PDA

View Full Version : Poly 88 case



Pages : [1] 2

falter
November 6th, 2017, 07:10 AM
I just picked up a Poly-88 case only.. wondering what you guys think my odds are of finding actual Poly boards? I seem to recall seeing some on ebay but can't remember how long ago.

An S100 bus is compatible with most any S100 card, right? I'm thinking if I can't find the actual Poly boards I can shove something else in there.. like this Applied Tech DG680 board I got.

glitch
November 6th, 2017, 09:45 AM
They do come up now and then, so if you keep an eye out you may well end up with a board set.


An S100 bus is compatible with most any S100 card, right?

Well, sort of. You can plug most S-100 boards into the Poly 88 chassis, but just because something is S-100 doesn't necessarily mean it'll work with other S-100 boards from a different manufacturer (or even the same manufacturer but from two different eras!). With a little chassis like the Poly 88, you should probably look for a single-board computer with at least the CPU, ROM, and console all on one board. Some very late SBCs will even have your full 64K of RAM.

Dwight Elvey
November 6th, 2017, 04:55 PM
Video boards are more common than CPU cards as they had early graphics
and were memory based.
The video board could also take a parallel keyboard, so no separate monitor needed.
They did make RAM boards but none of the Poly88s I've seen had them.
There are two holes on the back of the chassis that look to be for RS232. These would normally
have a serial buffer board and a cassette interface board. It is unlikely you'd find one of these
unless attached to a chassis.
Dwight

falter
November 6th, 2017, 08:29 PM
Question - *is* the Poly 88 historically significant? I had been led to believe it was but I'm not convinced it is.. doesn't seem to me it does anything unique vs. the Sol-20 and the Sol I thought was out earlier..

krebizfan
November 6th, 2017, 08:45 PM
The CPU card with RAM, ROM and cassette interface is probably mildly significant. I don't know if they were the first to integrate that much though.

falter
November 6th, 2017, 08:53 PM
Yeah. I'm just trying to figure out why they're so hotly contested. They're clearly very collectible.. I just don't understand why they stand shoulder to shoulder with a SOL.

Dwight Elvey
November 6th, 2017, 09:25 PM
It was also a single reset button machine.
It had a monitor in ROM rather than switches.
The cassette interface wasn't on the CPU, only the USART
was.
Dwight

glitch
November 7th, 2017, 06:36 AM
IIRC it was called the Micro Altair until they got told to cut it out :)

Dwight Elvey
November 7th, 2017, 08:18 PM
I think its claim to fame was the single button reset and no hardware front panel. The monitor in ROM was quite good. Since the video was memory mapped, it was fast to update. It showed a multi-line memory window, the registers and for the pointer registers, a single line of 8 bytes, as I recall. It highlighted the byte actually pointed to.
It had a tape read but there just wasn't enough room for the tape write. One had to enter it from the keyboard and then save itself to the cassette.
Dwight

DDS
November 8th, 2017, 03:50 AM
Years ago a friend gave me a Poly88 that was non functional. He also gave me a TEI enclosure with the Poly88 card set in it. One quirk on this Poly88 is that there's a cut out on each side that exposes card edge fingers on the motherboard. Supposedly this was a way to add S100 slots and power supply capacity to the system. You could just buy another box and connect the two motherboard sections somehow. These things were rare and I'd be willing to bet any original connectors or ribbon cables to connect two boxes would be even more so. And that's assuming Polymorphic didn't leave that as a DIY project for the buyer.

In this picture, you can see something with two card edge connectors attached to the near side of the motherboard. That could be a piece of original equipment that got separated from the unit I was given. Or it could be one person's way of making use of the expansion feature.

http://deramp.com/pictures/poly88_inside1.jpg

I was just cruising about for any info on a motherboard to motherboard connector. When I got to Herb Johnson's site, the name Dwight Elvey popped up.

http://www.retrotechnology.com/restore/poly_restore.html

Mr. Elvey, could you shed any light on the Poly88 expansion feature?

Dwight Elvey
November 8th, 2017, 05:56 AM
Hi
I'm afraid you know as much about the expansion as I do. Polymorphic went on to make the System88. I believe you are right that it was intended for expansion. You have to remember that the Poly88 came out right around the time that DRAM boards were becoming available so there was less need for the expansion of things like 8K static boards.
I made the DRAM modification to my CPU card and used a DRAM board. I modified the address decoder to be 4K or 8K ( I forget which ) to work around the monitor and video board when I had all 3 rows of 16K DRAMs. Since the RAM was all on one board, there was little need for expansion. As I recall it was expected that there was RAM starting a 2000H, not a 16K boundary.
The biggest issue with the Poly88 was cooling. Mine had the small fan kit installed when I got it but even with that there was not enough cooling so I always ran it with the cover off.
The person that might know something about the expansion is Marvin Johnston. I think he sometimes is on the classic comp news group and may be on this one as well. Marvin has quite a bit of the original stuff from the company when if folded.
One wonders if the one you have could have been used as is to plug a board into. I never looked at the pin layout. It would be useful to debug boards if that was the case.
I was noticing your unit had one of the I/O cards but it isn't clear if it is the serial card or the cassette card, connected to the CPU board. The picture isn't clear enough. I can see there is the card for the keyboard. These often had a resistor and zener diode on them for those keyboards that needed a negative voltage. The CPU card had the option for this as well but I suspect they found it more convenient to have the matching keyboard and connector than to deal with modifying the CPU card for each.
Dwight

DDS
November 8th, 2017, 04:08 PM
I was noticing your unit had one of the I/O cards but it isn't clear if it is the serial card or the cassette card, connected to the CPU board. The picture isn't clear enough. I can see there is the card for the keyboard. These often had a resistor and zener diode on them for those keyboards that needed a negative voltage. The CPU card had the option for this as well but I suspect they found it more convenient to have the matching keyboard and connector than to deal with modifying the CPU card for each.
Dwight

That's not my unit in the pictures. that's an example I just grabbed off the web in a hurry. Mine had the cards pulled out before I got it. The Poly88 case is in a stack of stuff I need to fix, and the TEI box the cards ended up in is in another stack. The good news is I'm not going to run out of projects any time soon.

But when it was given to me the original owner demonstrated how the monitor allowed you to edit memory and registers on the screen in hex then jump to the program. The display looked a lot like a typical DOS debug session. And this was when Altair & Imsai users were still flipping front panel switches. I later saw a similar layout when using the L$SNAP macro on a Univac 1100/80 and that wasn't real time. So the Poly88 user interface was way ahead of it's time.

I found this page that's a pretty good description of the software front panel.

http://www.polymorphic-computers.com/frontpanel.html

snuci
November 8th, 2017, 04:46 PM
This may help the Poly 88 expansion discussion.

I have the Poly 88 brochure that was mixed in with the documentation of my Poly System 8813. The specs say:

Expandability: Up to 4 chassis may be plugged together.

Here is also a page of the Polymorphic Systems price list that details the cost of:

1. An assembled and tested POLY 88 chassis @ $355
2. An assembled an tested "Expansion chassis for POLY 88" @ $375

Could be a normal chassis with dual S-100 card edge connectors connected together but that wouldn't allow for up to 4 chassis to be plugged together. As a side note, there are no pictures of the Poly 88 in any sort of expanded form or pictures of an expansion unit.

41864

snuci
November 9th, 2017, 03:38 AM
For those interested, I just uploaded a bunch of Polymorphic Systems marketing type files that have multiple brochures, price lists and newspaper articles on the POLY 88 and System 8813 computer systems.

http://vintagecomputer.ca/files-area/polymorphic_systems/

falter
February 24th, 2018, 08:40 AM
Question:

What are the bare essentials to have a running Poly 88? I know about the CPU card and Video interface.. did you need a RAM board or serial/cassette board of some sort? Also, is the monitor contained on just one ROM?

glitch
February 24th, 2018, 12:18 PM
IIRC there's some RAM on the CPU board, but not much. So, you want to have a CPU board, video interface, and some additional RAM. IIRC, cassette and serial are provided off of the CPU board (the interface on the back panel isn't just a cable, there's a circuit board that goes with it, and attaches by ribbon to the CPU), and the video card provides the parallel ASCII keyboard interface.

I'll probably end up putting a TDL ZPU and TDL SMB in the empty Poly-88 chassis I have.

smp
February 24th, 2018, 12:56 PM
I agree with all that Glitch said.

If one has a Poly-88 chassis only, I would suggest looking into the JAIR-8080 by Josh Bensdadon. It's available on eBay from time to time, but less expensive to contact Josh directly on the S-100 Group (https://groups.google.com/forum/#!forum/s100computers). This is an all-on-one-board system that is highly configurable to allow you to use other memory, I/O PROMs, disk interfaces, etc. Plus it has an SD Card interface onboard with CP/M too. An Excellent single board computer system, perfectly suited to a tiny chassis like the Poly-88.

smp

falter
February 24th, 2018, 01:06 PM
I do have a few interesting S100 cpu boards to play with.. one is an Australian DG680.. a Z80 board but the only problem is I'm missing the ROM and it needs a DG640 (I think that's the model) video card.

I also have that MicroDaSys card. And an unbuilt Byte MPU card.

I'm just a bit fuzzy on the whole active termination business and what's compatible with what for S100 machines.

glitch
February 24th, 2018, 01:26 PM
The Poly-88 is just a straight S-100 bus, no termination. Most older cards will work with it. You're not going to get a 10 MHz Z80 or 8088 or something to play nice on an unterminated bus, but it's fine for old 2-4 MHz systems. It's what was found in the Altair and IMSAI. You can add termination on a terminator card, or multifunction card, if you decide you need it.

It basically all mostly works together. Sometimes there are configuration changes or cut-and-jump changes you need to make on some boards to work in some systems.

With a chassis as small as the Poly-88, you want either a two-board system that was sort of meant to go together and give you a basic system, like the original CPU+Video board, or the TDL ZPU+SMB, or take it a step further and get a SBC. Josh's JAIR-8080 is one modern option, there are *lots* of vintage options. I personally like the Vector Graphic ZCB for a S-100 SBC -- it's a 2 or 4 MHz Z80, console serial port, parallel port, ROM, bootstrap circuit, and some RAM on a single card. It's simple and reliable, and can be made to look like an IMSAI or North Star Horizon serial console easily. The Vector Graphic ROM monitor is good.

PM me if you decide you'd like something put together that will definitely work -- you can either send me your boards and I'll get them working together, or I can see what I've got that'll fit the bill.

1944GPW
February 24th, 2018, 01:38 PM
I do have a few interesting S100 cpu boards to play with.. one is an Australian DG680.. a Z80 board but the only problem is I'm missing the ROM and it needs a DG640 (I think that's the model) video card.

My second computer was Applied Technology's S100 system using the DG680 and DG640 with the optional PCG board, I liked it a lot and ported TRS-80 software and the VDO text editor from BYTE magazine to it among other things.
If you are looking for the DGOS rom (I recall it stood for 'David Griffiths Operating System') or a DG640 you might try posting a want on the Microbee Users Forum https://www.microbee-mspp.org.au/forum/ The Microbee was the successor to that S100 machine, which I found out annoyingly they dropped like a hot rock just as soon as it was released.
The Microbee must have had a modified DGOS as I managed to backwards port some Microbee cassette tape software back to the DG680, the keyboard routines were different as the Microbee hardware I think used a light pen trigger to strobe, or something like that. The keyboard I had on my system was a blue GRI parallel ASCII unit absolutely identical to this one https://www.ebay.com/itm/New-Vintage-George-Risk-Industries-Professional-ASCII-Keyboard-Model-756/112812496243

You may find the DG680 will run with probably any 64x16 memory mapped video card starting up around D000H or thereabouts. A number of the games AT offered on cassette were ports of the Sol-20 ones, which I did not know until just a few years ago. I built a hand controller with left/right/fire buttons on it to play the 'shoot at overhead planes' game, can't recall its name. And a D/A converter/amp/speaker to play four-voice music from a 1970s BYTE article on waveform summation, a friend wrote the Z80 polyphonic code which I still have somewhere along with the D/A board.

falter
February 24th, 2018, 03:22 PM
Thanks guys.

I'd love to try making a DG640.. but have not been able to find any artwork. The microbee forum is really difficult to get through because it is tightly restricted due to legal issues. They have some sort of rain dance you have to do to grt access to the archive with the ROMs etc.

Dwight Elvey
February 24th, 2018, 03:53 PM
The video board has a connector that can go to a parallel keyboard. The video was one of the first to have some form of block graphics so are reasonably easy to come by, compared to the CPU card.
The CPU card had 512 bytes of SRAM. Not a lot but usually enough for the stack and monitor operations. The monitor ROM fit in one 2708. This gave quite a bit of information on the video output. The monitor was packed into the 2708 and didn't have enough memory for the cassette write. One of the first things a person would do is to enter the SMD code to dump a program to the cassette and saved it. It resided in that small space on the CPU card. There was also a cassette based asesmbler, 8K BASIC and a 12K BASIC. There were a number of cassette based games as well. It also has a single step but as I recall the code had to be in RAM. There were 2 more sockets on the CPU card for 2708s. These could be used for a version of PaloAlto Tiny BASIC. The binary code for this Tiny BASIC is available on the web. Normally, main system RAM needed to start at 2000H. There was a simple modification needed to handle DRAM boards but there was no issue with static RAM. The CPU board had a USART that could be used for cassette or serial to terminal/printer. These two functions required and additional board at the connector. The serial board was mostly just the level converters for TTL to RS232. The cassette board was a little more complicated and has at least one chip that would be hard to find. The keyboard also sometimes had a small board at the connector for the keyboard but this was usually only to take the ribbon cable to the 25 pin D. It some times had a zener or regulator to create the negative voltage for those keyboards that needed a negative voltage. The older video boards had circuitry to put a resistor and zener. Newer video boards needed the external board if a negative level was needed.
The Tiny BASIC was intended to be able to have instructions added to it. I added a PEEK, POKE and a SAVE instruction that I'd load from cassette. A programmed saved this way would auto load and leave one at the TB prompt to RUN or edit. I could have made it auto run the program but felt it was more useful this way. To run only needed a R. to start.
If one can find them, one could upgrade the Poly88 to a System88 but one needed to get the floppy disk card and a disk drive.
Sometimes empty chassis would be bought to expand the bus with a couple of 100 pin connectors or a single one if soldered to one of the chassis.
The one major problem with the Poly88 was lack of cooling. This is why chassis are often found without the orange cover.
Should you ever get enough to make the cassette work and have a serial card, I have a small bootstrap program that you can use images of the cassettes to bring up any of the cassettes that we've recovered.
Dwight

falter
February 24th, 2018, 04:25 PM
Thanks Dwight. I believe my Poly case still has the cassette board.. I found a picture that purports to be one.. it's a rather small board with a DB25 at one end and 7 or 8 ICs. Quite compact?

Dwight Elvey
February 25th, 2018, 04:16 PM
Hi
Then you need to find a CPU card. That is a more serious issue. There are just not that many, not in a System88 or Poly88 already. It was not a convenient board to do CP/M with so not as likely to be bought for that purpose. I've seen a CPU card on ebay once but it will be a tough one to come on.
Making a serial board is not too complicated.
Dwight

MicrocomputerSolutions
February 26th, 2018, 11:29 AM
Advanced Digital manufactured a couple of integrated single board S-100 Systems. The more well known processor board was the Super Single Six, which contained a 6mhz Z-80A processor. The Advanced Digital boards had processor, floppy controller, I/O ports, and system memory on them. There was even a design that had a hard disk controller on it.

They sold CPM-80 preconfigured for the system so there was no hassle trying to get the OS customized to work with the board.

Advanced Digital Boards show up on eBay occasionally for less than $300. The other choice is the JAIR8080.

falter
February 27th, 2018, 09:33 AM
I just scored two Poly 88 cpu cards and 1 video terminal card! For $195!

That brings my total investment to $395 including my case. Not bad since I've seen these things go for $800+.

Who knows if the cards work though..

glitch
February 27th, 2018, 09:48 AM
Doing pretty good there! Let me know if you're interested in parting with one of the CPU cards. If you have trouble getting them going, I'm happy to go through them for you.

Dwight Elvey
February 27th, 2018, 03:13 PM
You're on your way. The serial card can be made with as little as a tristate buffer and one of the maxim rs232 chips ( and possible inverter chip for glue logic ). There is a bit that comes from the CPU card that selects which card the USART listens to, either the serial card or the cassette card ( in case your wondering what the tristate buffer is for ).
Dwight

falter
February 27th, 2018, 05:50 PM
Will most any S100 serial board work? I see lots out there.. PT, Morrow, etc.

Glitch - sure -- perhaps we could do a trade? I believe you have that 480 board I'm looking for to properly complete my 500 'Superkit' setup. If interested.

glitch
February 27th, 2018, 06:36 PM
Will most any S100 serial board work? I see lots out there.. PT, Morrow, etc.

Dwight is referring to the little daughterboard for the Poly-88's CPU card serial port, which selects between cassette and the actual serial port. You can also use other S-100 serial boards in the Poly-88, though they make take some reconfiguring. A lot of S-100 stuff can be persuaded to work together, it just takes some effort sometimes. As long as you're not trying to do something totally mismatched, like trying to use a 16-bit only board in an 8-bit system.


Glitch - sure -- perhaps we could do a trade? I believe you have that 480 board I'm looking for to properly complete my 500 'Superkit' setup. If interested.

Yep, I've still got it. I'll send you a PM.

Dwight Elvey
February 28th, 2018, 02:36 PM
There isn't too much on the little serial board and you don't need to modify the monitor to use it. The monitor sets up the USART for you. If you add some s100 serial card, you'll need to write some code, put it on an eprom someplace ( if you use the space on the CPU card, you won't be able to use it for the Tiny Basic ).
The monitor is designed to easily patch in some other serial by what they call WormHoles. They just put the serial vector in RAM so you can patch it to someplace else, easily. I still recommend getting a small piece of vector board and make your own. There is almost nothing to it You can make one with original parts or do as I suggested and simplify it.
One other thing of mention is that Jef Raskin used a Poly 88 to create the manual for the early Apple II. This was because the Apple only had upper case letters. It is not clear though if he used a System88 or a Poly88 but his statement was a Poly88. It is just that someone would have had to setup a disk board for him with the Poly88.
Dwight

falter
March 8th, 2018, 04:21 PM
Okay my Poly cards arrived today.

So, if the (very bad) pictures in the scan of the manuals are stock configurations.. it looks like the CPU cards are all there... there is a single white NEC 2308084 ROM in ROM 3 and nothing in 2 and 1 on the first board. On the second board, there is a 2708 marked '035025B v. 75'.

On the VTI, it appears I am missing the MC6571 character generator. No problem, I have spares of those. I'm assuming the 'video out' header on the board, which is just two pins, is the standard composite signal and ground?

Are there particular slots these cards belong in?

glitch
March 8th, 2018, 04:43 PM
Yes, it's composite video out. No, it doesn't matter where the boards go -- the S-100 bus is just a straight-through bus, and the Poly-88 chassis isn't big enough for card placement to matter (on bigger unterminated buses, it's common to put the CPU and memory near each other -- I don't know if it really matters or if it's just something that helps out marginal systems).

falter
March 8th, 2018, 06:27 PM
Well that was quick.

Powered up the case... voltages seemed ok. So then I plugged in the VTI and CPU card.. BANG.. tantalum at C24 exploded. Nice sound and smell. Tested some chips off the board and so far no damage.

I noticed after that the cap right above the fan had pushed down into it.. preventing the fan from turning. Don't know if that precipitated things. Or if this is the same deal as the SOL with tantalums.

Dwight Elvey
March 8th, 2018, 07:26 PM
The fan shouldn't be an issue.
You do need to ensure you didn't open a diode on the bridge. There may not be enough filtering to use 60hz into the filter. If a diode blew, the filter will see 60hz instead of 120hz.
I'm not sure about the ROM on the CPU card. It should have a different version label. 035025B v. 75 doesn't sound like the boot polymorphic for the poly88. The numbers 035025 look like Polymorphic stock numbers so don't erase the EPROM without saving the data. I seem to recall something like v4. I seem to recall the eprom is in the first socket on the left, as you look at the CPU card.
When I get home I'll check that part number. It may be a replacement for the original ones.
Dwight

falter
March 8th, 2018, 07:58 PM
Ok. Guess I'd better figure out where the bridge is. :)

Nothing blew on the CPU card. But yeah.. I want to make sure I didn't do anything wrong before I risk having another projectile shot at me.

falter
March 8th, 2018, 08:36 PM
Okay, so checking the voltages again now.. now on pin 1 I have 11V, and on pin 2 I have -20V (?). Should be +8 and +16. So if it somehow went sideways between when I tested it last and plugged the cards in, I'm sure that -20V didn't jive with the tantalum cap.

So I guess we gotta check diodes at this point. I don't imagine this thing was built wrong years ago and just left like that.

I'm testing these voltages BTW on the 'sidecar connector'.. I lined up (with power off) an S100 card to make sure I knew (or thought I knew) where pin 1 was.

Dwight Elvey
March 8th, 2018, 08:50 PM
The +- rails for the +-18 could be about 20V. You'd better check all the rail voltages. It is better to check inside the cage. I'm not sure how the side connectors are arranged. I checked the number but I'm sure it is a Polymorphic ROM number. I have several image with numbers like 35110 and such. The only one with 35035 is for the double density 8 in. disk controller and that is Z80 code. All the others that I have look to be System88 images but I've not looked to carefully at them.
The original Poly88 ROM was a V4.0 ( you can fined binaries on the web ). The 35025 is quite likely the same or updated binary using the later numbering system that they used for stock of such stuff. I doubt they used the V4.0 numbering later on. If you get a chance, I'd like to get a dump from one of your ROMs. I can compare it with the V4.0 that I have on my Poly88.
Do you have a keyboard to work with it? It takes a parallel keyboard and connects to the video board.
Dwight

falter
March 8th, 2018, 09:01 PM
I decided to test the pins on the S100 sockets themselves rather than the pins of the side car. Yes, pin 2 appears to be +20V, and pin 52 appears to be -20V. Pin 1 and 51 are both +11V. So nothing to worry about there?

I'm wondering what I should replace the blown Tantalum with, or if I should risk plugging in another card to see what voltages were getting to the ICs.

I do have various parallel keyboards around. But first I want to not have it grenade on me. :)

falter
March 8th, 2018, 09:07 PM
Darn. I was looking through the scan of the manual I found online for the power supply checkout procedure -- but they didn't scan pages 3-16 for some weird reason. And that happens to be where that information is at. I guess I'll have to do another search. Bizarre.

Dwight Elvey
March 8th, 2018, 09:08 PM
I just checked the voltages pin 1 and 51 are both +8V.
Pin 2 is +18V and pin 52 is -18V are you sure you didn't mix 2 with 52?
If one of the diodes is blown open, the voltages will look normal but the ripple will be too much for most logic to work right.
I'm not sure it the +8V line is two diodes with a center tapped transformer or a 4 diode bridge.
Dwight

glitch
March 9th, 2018, 05:35 AM
Having a tantalum cap blow up isn't uncommon, some gear seems to be more prone to it than others (e.g. it's been my experience that North Star boards are really bad for it). As Dwight said, you want to check out the whole supply. I typically will test an unknown supply by loading it down with a dummy load (I use automotive tail light bulbs) and measuring ripple with an oscilloscope.

falter
March 9th, 2018, 10:09 AM
Ok. Will do.

So it's not a worry that the PSU is several volts over spec if ripple is ok? I'm guessing 11v just makes a 5v voltage regulator run a little hotter?

falter
March 9th, 2018, 01:26 PM
I'm kind of wondering about my DMM here.

I put it in diode mode and test with probes on correct ends. I get about a half volt. So far so good. Now i flip it around.. and I get 2V. My understanding of these things was that nothing should get through.. should it? Or does it change when they're installed? Every diode is consistent on this.. they all half a half volt one way and close to 2V the other.

Dwight Elvey
March 9th, 2018, 02:19 PM
Is this in circuit? There are multiple paths the current can take.
Is it a half bridge or a full bridge? Even if one of the diodes is open, both half and full bridge will show one forward diode as the transformer has such low resistance that the open diode won't show. If your going to measure with a meter, you need to disconnect one lead for a full bridge and the center tap lead for a half bridge. With an active transformer, you can use an oscilloscope, as Glitch says.
With the transformer active you can make a diode and small capacitor to use with an volt meter. If you put the circuit so that the diode is forward and the circuit in parallel with the tested diode, the voltage should never go over about .6V, with and active load. The diode is in series and the capacitor is in parallel with the meter. The capacitor should be about .1 to 1uf. With an ohm meter and a bridge, in place, you won't be able to tell much.
Dwight

falter
March 9th, 2018, 02:22 PM
Yes these are in circuit. I know the 'correct' way is to remove at least one end from circuit.. but I was hoping to avoid desoldering everything.

I'm not sure what half bridge or full bridge is or means. :)

Dwight Elvey
March 9th, 2018, 02:39 PM
A full bridge is 4 diodes and two leads to the transformer. A half bridge is two diodes and three leads to the transformer. Do a search for "full wave rectifier circuit diagram" You should see both types of circuit.
Dwight

falter
March 9th, 2018, 03:11 PM
Okay so after doing my best to check the diodes by removing one end from circuit.. I became fairly convinced the PSU and such were basically ok. I don't see any physical damage either.. no signs of rusting, etcetera internally. It's actually very clean in there. I also checked with a card plugged for shorts.. particularly pin 2 and 52 to make sure they weren't shorting together. Nope.

So I decided to gamble on my electronics non-skills and took the unit outside (paranoid, but I think prudent) plugged in the CPU card, got behind my steel door and plugged her in. No sparks, no pops, nada. After waiting 15 minutes with it running, I went outside and tested voltages on various ICs with my DMM. Everything is within spec. I've now removed a bunch of the ICs from the CPU card and checked them in my digital group machine and so far including CPU everything is good. So yeah. Either something is really wrong with the VTI board or it was just a tantalum cap waiting to fulfill its destiny as a Smurf grenade.

I would like to try again with the VTI but is there an alternative to tantalums? I don't like that whole 'bomb tech' feeling of working with them activated now.

PS: these CPU cards seem to be late revisions... one is a Rev E and the other is a Rev F.

glitch
March 9th, 2018, 03:47 PM
You may be able to replace the tantalum caps with aluminum electrolytics, depending on what they do in a given circuit; however, modern tantalum caps are a lot more durable than the old stuff, especially if you buy a good quality replacement. I'd recommend replacing with modern tantalum caps, perhaps upping the working voltage if the originals were a little too close to spec.

Dwight Elvey
March 9th, 2018, 05:15 PM
Usually on one or two blow. I'd just plug it in. You are even more likely to have bad RAMs on the video card.
Dwight

falter
March 9th, 2018, 09:15 PM
So 10uf 35v tantalums would work where 10uf 25v is called for, right? Same as with electrolytics, etc.. you can have a higher voltage than called for?

There seems to be mostly NOS caps in the capacitance i'm looking for. Is that putting me right back into the same risk of explosion? Should I be looking hard for new?

Also I looked up the white NEC chip.. apparently it contains the 4.0 monitor ROM and the EPROM with V 0.75 on it is for disk control, as Dwight mentioned.

daver2
March 10th, 2018, 03:34 AM
Yes, higher voltage device rating = less stress = better. But the physical size increases - so just beware of any physical constraints.

What I have noticed recently is that 'new' devices (capacitors in particular) generally have a higher voltage rating than their 'old' counterparts. Whether this is true or not...

My advice is - if you are going to keep the machine - replace the capacitors with new (unless your making one of your excellent 'period' cards up...).

Any NOS electrolytics will still need reforming.

My view only...

Dave

glitch
March 10th, 2018, 04:47 AM
Mouser, Jameco, Digi-Key, etc. should all have appropriate through hole replacements for you. Absolutely DO NOT BUY FROM EBAY unless you have trusted suppliers you already have a working relationship with! The sellers usually have no idea what they are dealing with, and many of the vintage/used parts crowd seem to not care about fitness of components.

The above is doubly true for tantalum caps, since they can in fact ruin sitting on a shelf.

daver2
March 10th, 2018, 04:48 AM
+1

Dave

falter
March 10th, 2018, 08:32 AM
I've only ever bought from ebay because I found Mouser etc's site difficult to navigate. But in this case maybe I'll make an exception.

I'm trying to figure out now if the tantalum 10uf caps on the CPU card are the same voltage as those for the VTI. If they are, should I take a chance and transplant one from my spare card for now to see if I can get the VTI up? The CPU card looks all happy.

One other oddity.. I had the 8080 out to test it (it's good).. while I was out I forgot and powered up the Poly.. the reset light for the first time lit up. When I reinstalled the 8080 it stayed unlit. Weird.

glitch
March 10th, 2018, 11:08 AM
eBay's a real trap when it comes to components. Half the time you get better prices through Mouser or Jameco anyway. There are a number of smaller shops that are geared toward hobbyists, too -- Anchor Electronics and Unicorn Electronics come to mind.

You can swap capacitors around if you like, you may also be able to operate the board without them, depending on where in the circuit they are. The board may be unstable without them.

I don't remember what the reset lamp connects to -- it may be as simple as a bad solder joint and unrelated to the functioning of the CPU card.

Dwight Elvey
March 10th, 2018, 11:10 AM
It is likely to work without the capacitor. Only the two next to each regulator are important.
Make sure to get the polarity correct. They bow even faster when backwards.
Dwight

falter
March 10th, 2018, 12:09 PM
It is likely to work without the capacitor. Only the two next to each regulator are important.
Make sure to get the polarity correct. They bow even faster when backwards.
Dwight

I'll check.. it's C24.. I was trying to follow the trace under the soldermask but it's tricky. Going to check the schematic.

NF6X
March 10th, 2018, 12:51 PM
I've only ever bought from ebay because I found Mouser etc's site difficult to navigate.

I also find Mouser's site to be difficult to navigate, and that's why I usually order components from Digi-Key. Their web site has the best parametric searching features I've found for electronics components anywhere. At my day job (I'm an electrical engineer), I've set Digi-Key to my home page in my browser!

I do know of one guy in the retrocomputing community who finds both the Mouser and Digi-Key sites to be pretty overwhelming, and he prefers Jameco. I can see how the huge selection at Mouser and Digi-Key (both of which primarily cater to engineering professionals) can be confusing to folks who don't live and breathe electronic components. Sometimes there are hundreds of viable options for a particular component.

Mouser's site is far from the worst, and I order from them occasionally. If you're price-conscious, you might want to search both Mouser and Digi-Key, because one or the other will sometimes have a significantly lower price on a particular component. Also, they don't both carry all of the same manufacturers, so sometimes you'll find what you need at one but not the other.

falter
March 10th, 2018, 01:16 PM
I think for me the difficulty is compounded by the fact that I'm still learning electronics. So if I go looking for a crystal oscillator.. Mouser pumps out 500 pages of them, many or most without pictures and I have to try and figure out if we're talking about the same thing. I find it very intimidating.

Generally I've had no problems buying from ebay, where at least there's always pictures and the search system is excellent. I've bought whole lots of 40+ year old ICs and so far with a couple of exceptions it all works.

Dwight Elvey
March 10th, 2018, 01:19 PM
If JameCo has it, I'll order from them because they are closer and I get them the day after I order them, if I order in the morning.
Be aware of shipping time and cost. It can be more than the value of the components by an order of magnitude.
Dwight

falter
March 10th, 2018, 01:30 PM
Okay so I checked and our cap does in fact serve a voltage regulator.. the +12V 78L12 at IC43. So now I'm wondering if said regulator maybe caused our trouble and blew up C24?

glitch
March 10th, 2018, 01:53 PM
Okay so I checked and our cap does in fact serve a voltage regulator.. the +12V 78L12 at IC43. So now I'm wondering if said regulator maybe caused our trouble and blew up C24?

That's pretty well impossible -- it's almost certainly just a bad tantalum cap. They grow tantalum crystals when exposed to moisture which produce shorts.

Yeah, Mouser does present a lot of options, I could see how it would be totally confusing to most people. I like their parametric search, generally I find it more useful than Digi-Key's, but there are some things it gets completely wrong. I almost always turn to Digi-Key for finding mechanical hardware to go with electronics (heatsinks, card guides, brackets, etc.) as their search handles them better. Digi-Key is usually cheaper on hardware, too -- I think it helps that they carry basically everything that Keystone makes :)

I buy parts from a mixture of sources and have taken to testing every single item from eBay if it's to go in kits. Sometimes it's the only place to find an item. I've had plenty of duds (I'll *NEVER* buy another AT28C64B EEPROM from eBay, even from some of the suppliers I use for other components...way too many failures). I've had a few duds from Jameco, almost none from Mouser and Digi-Key. Capacitors are just one of those things I don't buy surplus though, with the exception of soda can sized electrolytics. I've had some surplus fall in my lap, where I knew its storage conditions and purchase history (old stock from a previous employer) -- being essentially free, it was worth my time to test caps and characterize the lots that were available.

And of course some things are fine to buy from basically anywhere -- like resistors :)

NF6X
March 10th, 2018, 02:17 PM
While we're talking about distributors, McMaster-Carr is my go-to shop for a whole lot of mechanical stuff.

glitch
March 10th, 2018, 02:22 PM
While we're talking about distributors, McMaster-Carr is my go-to shop for a whole lot of mechanical stuff.

Indeed! All of my nuts, bolts, washers, and that type of hardware comes from McMaster-Carr! I also use them for many of my chemicals and chemical handling needs -- stuff like liquid soldering flux, wash bottles, 99.9% isopropyl alcohol, chemical-resistant storage bottles, and various lengths of cotton swabs :) Usually not the rock-bottom-price cheapest source, but it's always quality stuff, and they get it here the next day.

falter
March 10th, 2018, 02:50 PM
I think part of my problem too is I don't always know the name of something. For example, for my TVT I need a clip to secure the first couple inches of the AC line cord to a bolt on the transformer so the solder point isn't getting yanked on. But when I put clip or ac clip into mouser.. yikes.

falter
March 10th, 2018, 03:42 PM
Okay another inane question.

Another documentation fail - while they spell out the voltages for these 10uf caps in the manual for the VTI, they do not mention voltage for the CPU board tantalums.

That said, they look to be the same externally.. they have a brown band on top, black band in middle, and light blue on bottom and legs. There is a green stripe on one side which I assume indicates polarity. That compares more or less exactly to the known 10uf tantalums still on the VTI, although some look a tiny bit bigger.

falter
March 10th, 2018, 04:15 PM
Pointless question. I just tried to desolder one of the tantalums on the spare CPU board.. it's impossible. They're soldered on both sides and my iron just can't do it.

Chuck(G)
March 10th, 2018, 05:09 PM
The rule of thumb on capacitors is to determine what the working voltage of the point to which they're connected and add about 20% for a safety margin. For example, if it's a 5V line, you may be safe with a 6.3WVDC cap, but you're probably better off with 10WVDC. Anything higher is irrelevant--up to a point (i.e. A 1KWVDC cap is probably impossibly physically large).

For your AC line cord, you want either a "strain relief" or "cable clamp", depending on how you're going to attach the thing:

Strain relief:

http://heyco.com/blog/wp-content/uploads/2015/01/PH-65-1060B-1.jpg?a8f7f6

Cable clamp:

https://www.tubesandmore.com/sites/default/files/styles/uc_product/public/uc_products/p-hcableclamp-10.png

Back in the day of metal chassis, a knot in the power cord behind a plain old rubber grommet would do the trick:

Grommet:

https://media.rs-online.com/t_large/F1879453-01.jpg

A simple cable tie can also do the job.

NF6X
March 10th, 2018, 06:52 PM
Pointless question. I just tried to desolder one of the tantalums on the spare CPU board.. it's impossible. They're soldered on both sides and my iron just can't do it.

Well, I guess you need to upgrade your iron. I swear by Metcal irons, but they're stunningly expensive. A lot of folks seem to like Hakko irons. Sometimes when reworking a board, you might actually need to add some solder and/or liquid flux to help conduct heat into an uncooperative joint, before you start removing solder.

glitch
March 11th, 2018, 06:24 AM
I have a Hakko desoldering station, which I like, but I use Hexacon irons. Still made in the USA, still the same product line as the 1960s :) If your iron can't handle a double-sided board then it's definitely time for something better -- I used a 15 Watt Rat Shack iron for a long time, growing up, and did a lot of acceptable work with it. Make sure the tip is cleaned and tinned, and like NF6X said, you may need to add fresh solder if the original is badly oxidized.

Ksarul
March 11th, 2018, 09:32 AM
I use a pace soldering/desoldering station, but that is overkill for most hobby users. . .

falter
March 11th, 2018, 11:21 AM
What is the difference between a soldering station and just a soldering iron?

Side note: I managed to get my tantalum cap from the other board, but I think it may have taken out the through plate 'rivet' as I removed it from the CPU board.. ugh. Anyway, I put the cap into the VTI and plugged in again.. no explosions. So now i need to figure out how to fashion a connector for composite since my case doesn't have one anymore.

NF6X
March 11th, 2018, 11:53 AM
A good soldering station provides excellent temperature regulation with good transient response. It can supply enough power to maintain tip and joint temperature when soldering something that requires a lot of heat, like large surface pads, large connectors, etc. It has good transient response, so the temperature doesnít plummet when you put the tip on a large thermal mass, and it doesnít overshoot when you remove it from a large thermal mass. A really good station can solder a large connector efficiently, but it also has good enough temperature regulation to go right on to a tiny surface mount component without burning it up (assuming the tip is appropriate for both components). It will be easy to change tips when needed. There will be some way to change the tip temperature, either with controls on the base unit or by swapping tips, so you can use a higher temperature for lead free solder, an intermediate temperature for the good tin/lead stuff, or even a low temperature for soldering sensitive stuff with special solder alloys. The tips will have a plating that the solder will wet, because if the solder beads up on the tip then heat transfer will be poor. The tip plating needs to be tough enough to last for a while, too. Once the plating wears through, the tip is junk. Tips are consumable items. Some stations support accessories like desoldering guns, hot tweezers, etc. I donít like hot tweezers, personally, because they usually have blunt tips with poor alignment. I just use two pencil irons, one in each hand, when I rework 2 terminal surface mount discrete components.

A soldering iron is a pointy bit attached to a little heating element, with a mechanical thermostat. They have poor temperature regulation, and are usually underpowered for large joints and too hot for small ones. They suck for most tasks.

falter
March 11th, 2018, 11:55 AM
Okay so no life from the VTI. Darn. I think I have the cable connected correctly.. I've got the composite signal wire attached to the pin on the left, and ground to the pin on the right (if you're facing the board). But yeah.. no joy.

falter
March 11th, 2018, 12:28 PM
Checked the MCM6571 with power on.. I get -3.03V on pin 1, +5V on pin 2, and +11.83V on pin 3. I feel like pin 1 should probably be -5v?

EDIT: -3V is correct according to the manual. Hrmm.

falter
March 11th, 2018, 12:57 PM
Okay.. seems like I may have a problem at IC31 pin 8. According to the schematic we have a signal coming from the shift register at IC35 to pin 9. There seems to be some kind of clock signal there. But nothing coming out of pin 8, which is where whatever comes in on pin 9 should be going out. Bad 7407?

Chuck(G)
March 11th, 2018, 01:13 PM
Falter, remember that the 7407 is an open-collector device, so the output pins must have a pullup to work. Check the socket also, as well as the power. But 7407s, like all ICs, have been known to fail.



A soldering iron is a pointy bit attached to a little heating element, with a mechanical thermostat. They have poor temperature regulation, and are usually underpowered for large joints and too hot for small ones. They suck for most tasks.

The Weller TCP line uses what I'd call a "mechanical" thermostat--special tips with a slug of metal with a specific Curie temperature, contacted by a magnet connected to some contacts. Elegant and simple.

I'd still call it a soldering station--one the better time-tested ones.

falter
March 11th, 2018, 01:22 PM
Ok. There's no socket.. most of the TTL chips on this board have been direct soldered.

falter
March 11th, 2018, 01:29 PM
And I've got the concept right, right? What comes in on Pin 9 should be going out on pin 8?

NF6X
March 11th, 2018, 02:13 PM
The Weller TCP line uses what I'd call a "mechanical" thermostat--special tips with a slug of metal with a specific Curie temperature, contacted by a magnet connected to some contacts. Elegant and simple.

I used to think those were great, until I finally used a really good soldering station, on tasks where they make a big difference. It sounds like a new or used Weller TCP with a good selection of new tips would be a very nice upgrade for Falter, and probably quite suitable for most things from the through-hole DIP era.

Maybe I'm a soldering iron snob, but I've spent years doing SMD rework with 0402 and 0201 components, sharing boards with big brass coaxial connectors, with multi-layer ground planes and anywhere from 4-18 layers. I would not use a TCP for that kind of stuff. Its thermal control, peak power capability, and tip selection aren't up to those tasks. Naturally, I use the same tools for my vintage stuff, too, since I already have it.

P.S.: I still do not like working on 0201 components, even after practicing enough to be able to do it. Those little things are right at the limit of my comfort level.

Chuck(G)
March 11th, 2018, 03:14 PM
I somehow don't think that falter is up to SMT work, which is why I mentioned the Weller TCP--if you're into 70s vintage tools, it's about as good as it gets.

I'll do SMT work when required, by neither my eyes nor steadiness of hand are quite up to the small stuff. Soldering 50 mil pin-spaced TQFPs and 0402 is about my limit--and that's using either an OptiVisor or stereomicroscope. I have lousy depth perception. I can do those tasks with a Weller TCP or a hot-air rework station if the need arises.

I'm amazed at the guys who are steady enough to do cellphone repair.

Dwight Elvey
March 11th, 2018, 04:52 PM
I have two TCP setups. One is just a handle with a RS 24 volt transformer. The other is real Weller setup. I like to do most of my initial soldering with a small wedge #6 and use a #7 for heavy desoldering or that new junk solder. I've got a large 1/4 wide #8 when needed.
It would be better to have a fancy temperature controlled setup but with two irons that warmup and work in less than a minute, I've just not ever felt the need.

Chuck(G)
March 11th, 2018, 05:08 PM
Truth be told, I resort to the 2-tool setup when it involves large PCB planes with soldered-in through-hole components.

But my second iron isn't an iron--it's an old Weller D550 gun. At 325 watts, that bugger will pretty much melt anything.

I got rid of my American Beauty 175W iron a long time ago--kept burning myself with it (hint: don't reach for one without looking!).

I note that AB has a 60W TC soldering station. Don't know anyone who uses one, though.

glitch
March 11th, 2018, 05:21 PM
I've got a 100W Bell System tagged Hexacon for the big work -- soldering stuff to galvanized chassis boxes, etc. :) It'll get the job done.

OK, so I dug into my Poly 88 this evening to see if I could help a little. My CPU board is off doing its own thing, but the VTI seems to be operational except for a faulty SRAM. I went to look in the manual and figure out which socket maps to which part of the screen...only to discover that the online copies of the manual are for the older Polymorphic VTI and not the v1.2 board that I have! Do you have a newer manual, or does your board match up with the old manual?

PROTIP: when testing the VTI with a front panel machine like the IMSAI, it blanks video while you're examining memory. Also, set bit 7 of the byte high for ASCII mode -- low is graphics mode. Graphics mode is inverted, so writing 0x7F to a location clears it.

falter
March 11th, 2018, 06:02 PM
Yeah I ran into that too. However in the actual Poly 88 system manual they cover the assembly of the whole system including the later VTI card. I regret I am on my phone and it won't produce the link in my searches. But one of the sites with manuals has it. I think it is Volume 2.

falter
March 11th, 2018, 08:28 PM
Okay so no dice changing the 7407. Same deal. I can see pulse on pin 9 but not pin 8. So maybe I have a problem with the resistor (R19)? I've definitely got 5V on pin 14.

falter
March 11th, 2018, 09:01 PM
I followed pin 8 back to R19. R19 is low, voltage is less than 1V. The leg of R20 adjacent to that is 5V. For an experiment I tried jumpering pin 8 to 9 but that caused pin 9 to go low. Not sure what's up here.

Chuck(G)
March 11th, 2018, 09:06 PM
I assume that we're talking about IC30 in the VTI. If not, please correct me.

If that's the case, you have the situation that a steady "low" on any of pins 1, 3 and 5 will pull the output on pin 8 low. See how pins 2,4,6 and 8 are wired together? This is what's called a "wired-OR" (actually a "wired NAND", but the former term is a term of art, so we'll call it that). The output on pin 8 will be high only if the inputs on pins 1,3,5 and 9 are all high. If any is low, pin 8 will be low.
The point of IC30 is to combine several inputs to create your composite video signal.

falter
March 11th, 2018, 09:26 PM
No sorry it's IC31 actually.

The schematic is near the bottom of this manual: https://drive.google.com/file/d/1_uf3acVZMUIAPgXM7qQtWJ_tpd8_Ikhb/view?usp=sharing

For curiousity, I followed from pin 1 of IC31 there to see what was going on. There doesn't seem to be anything coming out of IC4 pin 8, which if I've read the schematic correctly is connected to IC31 pin 1. The clock pin (pin 11) on that same IC has nothing going on, and IC5 pin 3 has no signal of any kind either. Back past that I don't know what I'm looking for though. Or if I'm even finding anything significant probing the VBLK line.

Dwight Elvey
March 12th, 2018, 06:27 AM
I've got a 100W Bell System tagged Hexacon for the big work -- soldering stuff to galvanized chassis boxes, etc. :) It'll get the job done.

OK, so I dug into my Poly 88 this evening to see if I could help a little. My CPU board is off doing its own thing, but the VTI seems to be operational except for a faulty SRAM. I went to look in the manual and figure out which socket maps to which part of the screen...only to discover that the online copies of the manual are for the older Polymorphic VTI and not the v1.2 board that I have! Do you have a newer manual, or does your board match up with the old manual?

PROTIP: when testing the VTI with a front panel machine like the IMSAI, it blanks video while you're examining memory. Also, set bit 7 of the byte high for ASCII mode -- low is graphics mode. Graphics mode is inverted, so writing 0x7F to a location clears it.

Just pull the RAMs from the sockets one at a time until you find the right bit. No schematic needed there. ( my board is the older one same with schematics )
Dwight

glitch
March 12th, 2018, 06:37 AM
Just pull the RAMs from the sockets one at a time until you find the right bit. No schematic needed there. ( my board is the older one same with schematics )
Dwight

Heh, yeah, I know how to trace them down the hard way -- but I thought I'd check for a workable schematic first!

It looks like the schematic that Falter linked is one of the "old version" ones. I think I've got my CPU + VTI operational, there were two resistors busted off of the bottom edge of the CPU board (probably from the seller shipping the boards in the chassis), now it jumps into EPROM address range and clears the screen (except for the quadrant with bad RAM). I need to find a parallel keyboard to use with it, or interface something else to the keyboard port. I've got a George Risk 771 keyboard but can't find documentation on it.

falter
March 12th, 2018, 06:57 AM
Is it the wrong schematic? The pic of the board looks like mine..

glitch
March 12th, 2018, 06:59 AM
Maybe you've got the older board -- mine has the regulator in the upper-right corner, and it's a TO-3. The layout for the board that the schematic references has a TO-220 in the upper-left. There are other differences, like the placement of option jumpers.

Dwight Elvey
March 12th, 2018, 07:20 AM
Don't go shorting across the 7407. It is an open collector, not a TTL drive.
1, 5 and 11 should be mostly high with low going pulses. Pin 9 should be mostly low with
high going pulses. Looking at the output of the 7407 won't make much sense as it is mostly at
a low level.
If you have a newer board, you'll need to trace the wires as the locations may be different. The circuit is
mostly the same.
Dwight

falter
March 12th, 2018, 08:25 AM
Pin 1 has nothing at all. 5 is pulsing. 11 is low with no pulse. Pin 9 is pulsing.

Chuck(G)
March 12th, 2018, 08:52 AM
When you say "nothing", does that mean low or high or "indeterinate"? At any rate, check to see what's coming into (pins 11 and 12) and leaving (pin 8) IC4--a 74LS74. That may provide a hint.

I was using the original VTI schematic here (http://www.hartetechnologies.com/manuals/PolyMorphic/Polymorphic%20Systems%20Video%20Terminal%20Interfa ce.pdf)--it appears that you have the Rev. A version.

falter
March 12th, 2018, 08:59 AM
"Nothing" means my logic probe doesn't detect anything at all.. no high or low signals, no pulsing.. just dead, as if I was probing a piece of paper.

Pins 11 and 12 just seem to be low.. no sign of a pulse there.

Pin 8 at IC4 is 'dead' also.. no signal of any kind on that one.

Chuck(G)
March 12th, 2018, 09:06 AM
Since the LS74 isn't a tristate device, you may have part of the answer. Check the power pins; if they're good, swap in a known-good LS74.

falter
March 12th, 2018, 09:25 AM
Will do.. power looks to be 5V so yeah, maybe it's bad. Really wish they had been socketed rather than direct soldered.

Dwight Elvey
March 12th, 2018, 09:56 AM
Pin 11 also depends on the 74LS74 so it is not unexpected that pin 11 of the 7407 isn't pulsing.
Dwight

falter
March 12th, 2018, 10:08 AM
Thanks guys. Is there anything I should probe further up the chain from the 74LS74? Because these aren't socketed I don't want to do any more desoldering than I have to of course.

Chuck(G)
March 12th, 2018, 10:13 AM
If the output of the LS74 is truly floating *and* it has both power and ground, my first course of action would be to replace it. As I mentioned, the LS74 isn't a three-state device--the outputs don't "float"--they're either high or low.

Dwight Elvey
March 12th, 2018, 10:13 AM
VBLK- is for the vertical negative pulse. It is at 60Hz. I believe DBLK- is for those parts of the field that do not normally have text. I suspect it is a relatively high frequency. VIDEO+ is the white data for the characters or block graphics. CSYNC- is for the horizantal negative pulse and should be at about 15KHz.
Dwight

glitch
March 12th, 2018, 10:50 AM
74LS74 flip flops do seem to have a somewhat above-average failure rate. And, as Chuck said, the output should never be indeterminate -- the only other thing that can cause the output to *appear* indeterminate is a short in the board, or in the next logic gate following it. In either case, you're removing a device.

If your iron is really so inadequate that you're having a hard time getting capacitors out, and you're pulling out the through-hole plating in the process, maybe you should get a better iron before continuing repairs. There's no sense in wrecking the board over a $25 tool.

If I'm stuck in a situation where I don't have proper desoldering stuff, I typically use flush cutters and cut the IC legs off as close to the IC body as possible, then pull each pin out from the top using needle nose pliers and a soldering iron with a blob of solder on the tip. You can then clear the holes using a solder sucker, wick, or even a pick that solder won't stick to (I've used a stainless steel dental pick in a pinch).

falter
March 12th, 2018, 12:01 PM
I think I found a method similar to yours that works for desoldering - I cut the pins down, then heat and remove quickly. Then I ready my Japanese made solder sucker on one side of the board and the iron on the other, apply heat and hit the trigger. Sucks everything right out of the hole, no damage or problems.

Chuck(G)
March 12th, 2018, 01:16 PM
Dave reviews the ZD985 desoldering tool here, in 2013 (https://www.youtube.com/watch?v=Ft50m8UU5WQ) and seems to like it very much. I've seen it offered on eBay. Definitely the budget tool, but maybe worth a shot.

falter
March 12th, 2018, 08:57 PM
Okay so I replaced the 74LS74 with a known good.. exact same problem. Pins 8 11 and 12 are all 'dead'.. no signal at all.

Chuck(G)
March 12th, 2018, 09:16 PM
Did you put the LS74 in a socket? We may be able to make some tests. Note that the Q and Q/ outputs of each section will always be the complement of each other.

And did you ever get an oscilloscope?

glitch
March 13th, 2018, 04:50 AM
I dug on my VTI more last night. DBLK is, I think, "Data Blank" -- it's used to blank the screen when the CPU is accessing the video RAM *or* the keyboard port (the latter bit is annoying and I'll probably try to fix that). When it is activated by the BS- (Board Select) signal, the only way for it to be reset is by the EOC- (End of Character) signal. See if pin 4 on IC 4 is stuck low. There are a bunch of things that talk to BS- but the signal originates from a DM8131 comparator near the address DIP switches.

I've got a copy of the Vector Graphic ROM monitor with a video driver for the VTI (this was a factory option), it polls the keyboard port instead of using interrupt-driven access. This causes the first character of every line to be lost, and a fair bit of "noise" in the video display. Lifting IC 4 pin 4 (or hitting the STOP switch) will bring the full display back, but results in "snow" on the areas of the screen that are supposed to be blank, presumably from simultaneous access to the video memory. I may end up fixing it, or not, since the VTI will go back to the Poly-88 eventually. For what it's worth, that shouldn't be an issue with the rev 0 video boards, it looks like the DBLK flip-flop was added in rev 0.1.

Oh, and the schematic in the manual you linked from Google Drive does seem to apply to the rev 1.2 boards. At least in the little bit of digging I've done.

Dwight Elvey
March 13th, 2018, 06:18 AM
Check that the ground lead of the 7474 is ground. It is possible that there is a bad feedthru or broken trace. It could be a failed input of one of the other chips pulling the output up. Can you measure the voltage on the two outputs of the 7474?

Dwight Elvey
March 13th, 2018, 07:09 AM
Okay so I replaced the 74LS74 with a known good.. exact same problem. Pins 8 11 and 12 are all 'dead'.. no signal at all.

I think we should make a list of places there should be clocks.
IC30 pin8
IC2 pin 15
IC15 pin 1
IC15 pin 13
IC16 pin 1
IC16 pin 13
IC13 pin 13
IC34 pin 1

Check these and list which ones have clocks and which don't.
Dwight

falter
March 13th, 2018, 01:42 PM
Will do when I finish work here. I can confirm that I was able to test the ground pin on IC4 (74LS74) and found it had continuity all the way down to the ground pin on the expansion edge connector. So I'm pretty confident that's AOK.

I'm of course assuming the graphics card will produce *something* onscreen if it's just left in the unit by itself? Even just a blank pattern or blocks? That's how I've been testing it.. without the CPU card installed.

glitch
March 13th, 2018, 02:26 PM
I'm of course assuming the graphics card will produce *something* onscreen if it's just left in the unit by itself? Even just a blank pattern or blocks? That's how I've been testing it.. without the CPU card installed.

Well there's your problem! -- unless your card has the crystal oscillator section built up (and there's no reason for it to, if it was in a Poly 88) then the board isn't getting a clock! No clock means nothing drives the shift registers, which means no video signal ever gets generated.

Be aware that the Poly 88 CPU uses a 1.8432 MHz system clock and puts it on the 2 MHz S-100 bus line, which means that a regular 2 MHz 8080 board will overclock the video card and cause horizontal sync to be over-frequency. If your monitor has a lot of range in the H HOLD pot you can compensate.

falter
March 13th, 2018, 02:45 PM
Doh! I should have noticed there was no crystal onboard but I assumed because I was seeing pulses at a few points that it was trying to do something.

I did plug it in with the CPU board and got nothing either. Glitch can you confirm that the composite video header has the signal pin on the left and the ground pin on the right? My unit didn't come with a composite jack and connector.. so I'm not sure if I have them reversed.

glitch
March 13th, 2018, 03:06 PM
That's how mine is hooked up. Make sure the CPU card's oscillator is actually running and putting out a 1.8432 MHz signal, and that it's getting to the VTI on S-100 pin 49.

falter
March 13th, 2018, 03:17 PM
Okay. Right now I am thumbing through the manual trying to figure out what some of these jumpers on the CPU card do. For example, there is a point marked 'VT1', connected at the bottom to pin 5. It has a wire soldered into it (immediately below R5) and that goes to a point marked A right beside IC15. No idea what it does but the fact that it's soldered into something marked VT1 has me wondering. The other CPU card I have here doesn't have that.

There's also a jumper under R9, I guesss connecting R9 to pin 9.. address? This manual is really hard to flip around in in PDF.

falter
March 13th, 2018, 03:42 PM
Okay so pin 49 is pulsing both on the bus and on the VTI card with the CPU plugged in. Interestingly, the pulse I was seeing at pin 9 of IC31 disappears when the CPU card is installed, becoming (at least to my probe) a solid 'low'.

However nothing has changed on IC4. We still have no signal of any kind whatsoever on pins 11 and 12.

Chuck(G)
March 13th, 2018, 04:49 PM
So, work backwards.

What's on pins 1 and 2 (they're tied together) on IC5? If nothing there, how about IC15? IC19? Just keep tracing back in the schematic until you see pulses of some sort.

falter
March 13th, 2018, 05:18 PM
Checked all the pins that Dwight suggested for clock signals:

IC30 pin8 - Clock signal present
IC2 pin 15 - No pulse.. just solid 'low' on logic probe
IC15 pin 1 - No pulse.. just solid 'low on logic probe
IC15 pin 13 - No pulse.. just solid 'low' on logic probe
IC16 pin 1 - Clock signal present
IC16 pin 13 - Clock Signal present
IC13 pin 13 - clock signal present
IC34 pin 1 - Nothing at all.. no pulse, no signal.

Dwight Elvey
March 13th, 2018, 06:24 PM
Now check IC1 pins 2 and 15.
The clock you see at places like IC16 pin 13 is the dot clock. That is from a free running multivibrator IC29.
The missing clock at IC2 pin 15 is the CPU clock, divided by IC1 !!
Dwight

Dwight Elvey
March 13th, 2018, 06:55 PM
Glitch
If I did my math right, you can use your 2MHz if you change JMP1. It should set the IC2 to divide by 14 instead of 13. IC1 divides by 9.
Dwight

Dwight Elvey
March 13th, 2018, 07:20 PM
Falter
Can you send me a copy of the code in your EPROM?
Thanks
Dwight

falter
March 13th, 2018, 07:35 PM
Thanks Dwight.

There doesn't seem to be any pulse on pins 2 and 5 of IC1. My probe just gives me a steady low.

Yes I can (attempt to) dump the code.. from the V .75 ROM you mean?

falter
March 13th, 2018, 07:47 PM
I must be going blind.. I'm looking right at the VTI board, pin 49.. and I don't see any trace leading from it at all. There is a pass through directly above it, but that connects to the pin on the opposite side. I can't find JMP5 and I don't see pin 2 of IC1 being connected to anything either. Some of the JMPs are not labelled, so that just adds to the difficulty. I do see what looks like a JMP point right between IC1 and 2.. Confused.

Dwight Elvey
March 13th, 2018, 07:55 PM
Thanks Dwight.

There doesn't seem to be any pulse on pins 2 and 5 of IC1. My probe just gives me a steady low.

Yes I can (attempt to) dump the code.. from the V .75 ROM you mean?

Yes, please. I'd like to have a look at it.
Dwight

falter
March 13th, 2018, 07:56 PM
Ahhhhh... ok.. I see the problem now. My board, a Rev F, does not have the trace from Pin 49 at all. It was removed somewhere between Rev 1.2 and F. I can see quite clearly in this photo (http://www.s100computers.com/Hardware%20Folder/PolyMorphic/Video%20Board/Video%20Board.htm) where the trace *was* but is not, at all on mine. So this schematic is not quite correct. And I have no idea now how clock signal gets to the board..

Dwight Elvey
March 13th, 2018, 08:02 PM
I must be going blind.. I'm looking right at the VTI board, pin 49.. and I don't see any trace leading from it at all. There is a pass through directly above it, but that connects to the pin on the opposite side. I can't find JMP5 and I don't see pin 2 of IC1 being connected to anything either. Some of the JMPs are not labelled, so that just adds to the difficulty. I do see what looks like a JMP point right between IC1 and 2.. Confused.

Are you sure you don't have a crystal version? I have one with a crystal on it that I got someplace in the past.
Make sure you have your numbering correct. Pin 2 of IC1 does need to be connected to a source.
Jmp5 is connected to IC2 pin 4 on the center. It changes IC2 from /13 to /14.
Look carefully at IC1's pin 2. it needs to be connected to some clock source.
Dwight

Dwight

Dwight Elvey
March 13th, 2018, 08:12 PM
A quick look on the ebay shows a video board with the keyboard cable. A lot of Poly stuff lately. The second one shows jmp1 and jmp5 in the pictures. The first one has the crystal as well.
How about a picture of yours?
Dwight

falter
March 13th, 2018, 08:23 PM
Here's a shot of my pin 49:

44269

And here's IC1 pin 2... unless there's a trace going underneath the IC.. yeah.. it's not connected to anything.

44270 44271

And JMP5 is way over here now:

44272

I don't see a crystal anywhere on mine.. maybe you guys do. And we have some wire mods on the back. I link to my google drive so the pics are at a decent resolution.

https://drive.google.com/file/d/1yhAh-a40HgvOl8cC0vCugzAx-CuOPoMu/view?usp=sharing

https://drive.google.com/file/d/1siwfU3FWTzww0fwcczIBpCsuHThK7c1y/view?usp=sharing

falter
March 13th, 2018, 08:53 PM
Yeah I can't find a manual anywhere that explains this board. It's either Rev 0 or 1.2. I think I'm gonna close up this machine and put this all this aside for now. Maybe hope a Rev 1.2 or something closer to the schematic I have shows up one day. Kinda flying blind without a good schematic.

falter
March 13th, 2018, 09:15 PM
One thing I did note as I was putting stuff away... there is a spot on the board marked 'Y1'.. and on the other side are the remains of the legs of whatever was there... maybe that's where the crystal was? I don't know what Y means in the nomenclature of parts placement.

falter
March 13th, 2018, 09:36 PM
Answered my own question. There is definitely supposed to be a crystal there. Check out this photo of same revision board, in that area (crystal is top, just to the left of the second IC from the left):

44274

So that explains a lot. Now to find out what crystal it is.

Chuck(G)
March 13th, 2018, 10:09 PM
There is a board jumper to select either the crystal oscillator or a clock signal off the bus.

falter
March 13th, 2018, 10:25 PM
But my board has no connection to pin 49... what other pin could it get a clock signal from?

I'm wondering if I could patch the clock from pin 49 somewhere on this board for now..

Chuck(G)
March 13th, 2018, 10:44 PM
Look at the schematic, note "7" which says "Cut trace from 49 for on-board crystal". If your board has a 74S124 installed, then that's what someone's done.

falter
March 14th, 2018, 05:55 AM
Yeah. But that schematic is for rev 1.2 of the board. I am rev F. There is no trace to cut. Pin 49's trace has been eliminated in this revision. Pin 49 goes nowhere. I imagine I could try jumpering from that pin to where the onboard crystal feeds in..

glitch
March 14th, 2018, 05:55 AM
Look at the schematic, note "7" which says "Cut trace from 49 for on-board crystal". If your board has a 74S124 installed, then that's what someone's done.

Chuck, it's a different board, the schematic doesn't match. It doesn't look like there ever *was* a trace to pin 49, from the pictures.

I'd say you've found the issue with the crystal being gone. You can run a jumper down to pin 49, or put the proper crystal back in. Putting a crystal in would probably be the easiest way forward, since the schematic doesn't 100% match up to your board. You can probably use either a 2 MHz or 1.8432 MHz crystal, as Dwight pointed out there's apparently a jumper for selecting the prescale in the clock divider. Both are common frequencies (1.8432 MHz is common as the input to a baud rate generator as it divides nicely into standard x16 baudrates).

Dwight Elvey
March 14th, 2018, 05:55 AM
Can you measure the pins 3, 4, 5 and 6 of the two IC's, IC1 and IC2 to see which ones are connected to ground and which go to a pullup, with your ohm meter. With that I can tell you what frequency you need. I am suspecting it is a 2MHz, like the original had. The board also has a 50/60 Hz jumpers but that effects the vertical signal. Yours also looks to have the RAM for 64 column.
I suspect we can wire it to pin 49 but I need to see which pins are effected by the jumper between IC1 and IC2. I suspect the trace jumper is for the 2MHz crystal. Also, you can see a trace from the feed through below IC1 going to the IC next to the Y1 crystal. That would be the wire to pin 2 of IC2. We should be able to make it work with two cuts and on jumper to pin 49.
I do need to know what the current configuration is of the pins on IC1 and IC2, pins 3,4,5 and 6.
This way we can run it like the original POLY88 used, without the crystal.
Dwight

falter
March 14th, 2018, 08:40 AM
IC 1:

Pin 3 - 2.4k
Pin 4 - 2.4k
Pin 5 - 2.4k
Pin 6 - GND

IC2:

Pin 3 - 2.4k
Pin 4 - 2.4k
Pin 5 - GND
Pin 6 - GND

Just for kicks, I ran a jumper wire from pin 49 to the left side of Y1 and got this:

44291

I'm wondering if I should remove all the extraneous wiring on the back. Other Rev F and H boards I've looked at don't have any of that. Not sure what purpose the original owner had in doing that.

Chuck(G)
March 14th, 2018, 08:42 AM
And here I was going with what I was told was the schematic. So, is there a schematic for rev. F?

falter
March 14th, 2018, 08:46 AM
I thought it was the right schematic.. it was very close... I didn't realize there was such a huge difference between 1.2 and F when it came to pin 49. It looks like they had many different revisions over the years. I have looked and looked but have not found any schematics for F. May not exist.

Dwight Elvey
March 14th, 2018, 09:30 AM
Usually it is more random on the values but it looks relatively good. IC1 and IC2 are configured for the 1.8432 frequency, so that is good. The image you have does not seem to indicate any particular failure at this point. When you have the V4.0 monitor and keyboard connected, we'll know more. You should still cut the lead going from pin2 of IC1 to the IC next to the crystal. You'll be loading down the driver on the CPU card and possibly shortening its life.
In general I would not remove jumpers. Often these are factory installed so to solve routing issues. The boards is still very similar to the original. Can you note which pins have wires and which IC number and what 74XX part they are, in case the IC number doesn't match the schematic. The boards should be almost identical in function, except a little extra circuit for the additional columns.
Dwight

glitch
March 14th, 2018, 09:45 AM
Like Dwight said, cut the trace and rejumper to there -- you can't just hook pin 49 into one of the crystal lead inputs!

falter
March 14th, 2018, 09:50 AM
Ok. Just fyi that pic was taken with the cpu board installed..

So just for clarity.. I want to cut the trace leading from IC1 pin 2, and then jumper from pin 49 to there? Or to the IC pin it was leading to?

falter
March 14th, 2018, 09:58 AM
You know, I'm looking at it here.. I don't see *any* traces leading from IC1 pin 2 to anywhere. On the back side it's not connected to anything. On the front side, I can see a trace running beside it.. but that's it. Unless the trace runs backwards under the IC... I'm not seeing it connected to anything.

EDIT: NM.. I found it.. it does run under the IC and then out to IC45 pin 7 (74LS04)

Dwight Elvey
March 14th, 2018, 11:19 AM
Cut the trace and run the wire from 49 to IC1 pin 2. Are you sure it is 7? that is ground!!!!!
I suspect it is pin 8. Pleas check!!!


If the CPU doesn't have the right boot code, at most it should clear the display. As I recall, you have to do something to get into the monitor. I forget what.
I'll be off to lunch soon so won't be able to check on it until after I get back.
Dwight

falter
March 14th, 2018, 01:12 PM
Might be pin 1. Not at home but I'm looking at my picture of the back of the VTI.. I see the passthru under IC1.. and the trace goes to the bottom right most pin. I see the ground pin (I think) at the other end of that row. So maybe pin 1? I get confused when I flip these things over. :)

Dwight Elvey
March 14th, 2018, 02:00 PM
There is a feed through, under IC1. I'll bet that is connected to pin two as it goes in the desired direction, towards the crystal. Use your ohm meter! Don't guess.
Dwight

Dwight Elvey
March 14th, 2018, 03:23 PM
One other thing I recall, the System88 used a different memory location than the Poly88 V4.0 code did. If it came from a System88, the address decoder would be set wrong. I don't recall what it was for either.
a more readable schematic at: http://bitsavers.informatik.uni-stuttgart.de/pdf/polyMorphicSystems/Video_Terminal_Interface_Manual_1977.pdf
Dwight

I did some digging. I think the System88 was at D800 and the Poly88 was at F800. The jumpers may have to be changed to match.
I'll look some more when I get a chance.
Dwight

falter
March 14th, 2018, 04:20 PM
Okay so I'm home and did a continuity test. I am starting to go a bit dyslexic with flipping these boards upside down.

IC1 pin 2 goes directly to IC45 pin 7. For sure.

falter
March 14th, 2018, 04:21 PM
I did change the jumpers to what I thought was F800. However, there are different jumper settings for that address between Rev 0 and Rev 1.2 and they may well be different again for Rev F. No way to know.

Dwight Elvey
March 14th, 2018, 05:37 PM
Okay so I'm home and did a continuity test. I am starting to go a bit dyslexic with flipping these boards upside down.

IC1 pin 2 goes directly to IC45 pin 7. For sure.

That is either a design error or a solder bridge. Please measure from the top of the board where you can clearly see what you are measuring. If IC45 is a 14 pin part ( typical 7404 ) pin 7 would be grounded. You can check that by measuring to another ground. The fact that you got a raster leads me to believe that you were not driving ground from pin 49 of the edge connector.
Just to make sure we are doing it right. From the top, the notch is up. Pin one is on the left side of the IC. One is counting down the left side of the part. Picture please.
Dwight

Dwight Elvey
March 14th, 2018, 05:58 PM
I'm looking at the picture you posted of the bottom of the board. The trace in question is definitely pin 8 and not pin 7. There is no magic in the curcuit. I can see it clearly. That is the wire to cut.
Dwight

falter
March 14th, 2018, 06:45 PM
Man I'm really having a day today. Yeah, you're totally right on that one. I was just counting pin 1 from the wrong side. Sheesh.

Okay so with that trace cut, where is my pin 49 connection being made? To IC45 or IC1? Based on the 1.2 schematic, I'm guessing pin 2 of IC1.

Many thanks for the help and patiently sorting through my apparent ineptitude at counting.

falter
March 14th, 2018, 07:03 PM
I think the CPU is executing the ROM.

I noticed when I fired it up, the screen would very, very briefly flash a screen full of garbage, and then it would go to that garbage couple of lines in my picture. I notice too if I hit reset, I can see my 'lines' flicker, like a reset should. For an experiment, I removed the CPU and fired up... the screen full of garbage stays. So I'm thinking the CPU must be running something.

falter
March 14th, 2018, 07:18 PM
Dang.. I screwed something up.

I thought I'd switch my monitor over to the regular resistance level just to see what sort of picture I got... nada.. so I switched back.. now it won't stabilize to whatever i was looking at before. I have to crank the horizontal control *all* the way over counter clockwise to even see something similar. ugh!

Dwight Elvey
March 14th, 2018, 08:37 PM
You've put the jumper from IC1 pin2 to 49?
Dwight

falter
March 14th, 2018, 08:40 PM
Yes.

This is the picture I get now. Wavy.. can't stabilize it.
44309

falter
March 14th, 2018, 08:45 PM
And here's a shot with the CPU out. If you look very closely you can see actual letters amidst all the garbage.

Dwight Elvey
March 14th, 2018, 09:07 PM
What do you mean by the regular resistance level?
It is possible that you blew the output transistor. I'd suspect most any medium power NPN would work. It looks like you've lost contrast in the signal. About what you'd see if the transistor gave up.
Dwight

falter
March 14th, 2018, 09:17 PM
My monitor has two settings.... regular (75ohm?) and high resistance modes. Controlled by a switch.

Dwight Elvey
March 14th, 2018, 09:23 PM
I'd suspect the transistor may have given up. It may have been running right at the edge, after all these years. The
extra load of the 75 ohms in parallel with the 82 ohms may have been enough to make it give up.
I can see all the levels. I can see the horizontal pulse on the screen and the light areas from the data.
Dwight

glitch
March 15th, 2018, 05:32 AM
That's what one of my monitors looks like when you can't get H HOLD dialed in, when running on the wrong clock frequency (my board is set up for 1.8432 MHz, I was running with a 2 MHz system clock). Another monitor has enough H HOLD adjust to get a useful image. Perhaps your VTI was set up for 2 MHz crystal operation after all?

Dwight Elvey
March 15th, 2018, 05:45 AM
I looked at the wire modification of your board. It looks like a timing modification and I suspect it was a manufacture modification. You should leave it there. It looks like a timing issue related to reading the keyboard input.
Dwight

falter
March 15th, 2018, 10:49 AM
Ok. Regarding the crystal.. I'll see if I can find a 2mhz crystal. If not.. could I maybe find a way to make a 4mhz work and divide it down to 2?

Chuck(G)
March 15th, 2018, 11:00 AM
I've got a pile of 1.8432MHz XCOs in my hellbox that I have no use for... 14 pin DIP outline with the usual terminals at the corner pins.

falter
March 15th, 2018, 01:34 PM
Dang. I just missed two VTI boards on ebay. Went for $140 each. Same revision but one had a crystal and both were socketed.. which I am dearly missing right now.

falter
March 15th, 2018, 01:35 PM
That's what one of my monitors looks like when you can't get H HOLD dialed in, when running on the wrong clock frequency (my board is set up for 1.8432 MHz, I was running with a 2 MHz system clock). Another monitor has enough H HOLD adjust to get a useful image. Perhaps your VTI was set up for 2 MHz crystal operation after all?

Doesn't the system clock run at 2mhz? So if my board was set up for 2mhz it should be working right?

glitch
March 15th, 2018, 02:35 PM
Doesn't the system clock run at 2mhz? So if my board was set up for 2mhz it should be working right?

No, on a Poly 88 the system clock runs at 1.8432 MHz. It's unclear which frequency your board is configured for.

Dwight Elvey
March 15th, 2018, 03:35 PM
I've got a pile of 1.8432MHz XCOs in my hellbox that I have no use for... 14 pin DIP outline with the usual terminals at the corner pins.

That would work well for Falter as there are no jumpers to change.
He still has to find out what happened to his signal. My bet is on the transistor but it was only dropping less than 30 ma at about 4V. Most of those transistors can handle 300mw. It was running at worst around 120mw. I can't see anything else that would be causing an issue.

Dwight Elvey
March 15th, 2018, 03:38 PM
No, on a Poly 88 the system clock runs at 1.8432 MHz. It's unclear which frequency your board is configured for.


Remember, I had him give me the levels of the two 74161s. They were in the 1.8432MHz state and not the 2.0MHz. I wanted to check that before he jumpered to pin 49.

1.8432MHz is
IC 1: /9

Pin 3 - 2.4k
Pin 4 - 2.4k
Pin 5 - 2.4k
Pin 6 - GND

IC2: /13

Pin 3 - 2.4k
Pin 4 - 2.4k
Pin 5 - GND
Pin 6 - GND

or 15,754Hz

2MHz is
IC 1: /9

Pin 3 - 2.4k
Pin 4 - 2.4k
Pin 5 - 2.4k
Pin 6 - GND

IC2: /14

Pin 3 - GND
Pin 4 - 2.4k
Pin 5 - GND
Pin 6 - GND

or 15,753Hz

With the transistor out, he would not have enough voltage range to drive the video. It would look like what he is seeing.

Dwight

Dwight Elvey
March 15th, 2018, 03:41 PM
Doesn't the system clock run at 2mhz? So if my board was set up for 2mhz it should be working right?


No, the Poly88 CPU is 1.8432 unless you changed it. The crystal is 16.5888MHz on the CPU card.
Dwight

falter
March 15th, 2018, 04:14 PM
With the transistor out, he would not have enough voltage range to drive the video. It would look like what he is seeing.

Dwight

I'm thinking it's the transistor too. I fired up the machine for the first time today and briefly I had that stable-ish screen back. I shut down, swapped two RAM chips and fired up again.. and back to being unable to get the sync right, even after swapping them back.

The only other situation I've been in like this was I think on my digital group machine or something... turned out to be a bad 74123... I always run into those.

BTW: Am working on getting a dump of that ROM. Now my SWTPC machine is choosing to be jerky.

falter
March 16th, 2018, 04:56 PM
Dwight - here is the S record I generated from dumping the v .75 EPROM:

S1130000C33A04C3AF04C37808C3E606C33705C3C1
S1130010C604C31307C3BF07C3CB07C30205C31674
S113002005C3D907C37E07C34105C3EC05C3E40573
S1130030C34B05C36508C3BC0475F3310010211C10
S113004018221E0C3E86D32BAFD328D32A3E0BD3C3
S1130050203E18D3213E40D3203E0CD3213E80D3F2
S1130060203EE6D3213EC0D3202100203E3236007C
S113007023BCC26E0421640022B32D22952D229349
S11300802D3E0A32B22D3EFF32A42D32A52D32A6CA
S11300902D21550622210C21280522862DCD160559
S11300A03E0CCD240C3E1A32982D3E0132922D3155
S11300B00010CDBF0745786563C3AF04CD13074572
S11300C0786563C36400CD020521C92D7EE640C274
S11300D0DD047EF64077EB229A2DC3AF04EB229C1D
S11300E02D21F704CD3705E52A9A2DEBCDD103E177
S11300F023CD3705C3F4040D284572726F722000B6
S113010029003A882DB7CA1605AF32882DE5212972
S113011005CD3705F321F3E5211E0622180C213FF6
S11301202D22822D22842DE1C90D28436D64662081
S113013061626F727429007EB7C8CD240C23C33763
S113014005D5E5C5EBCD3705EB0600CD200CFE7FCC
S1130150CA8905FE18CAB505FE17CAC00577230467
S1130160FE0DCA7A05CD240CFE09C27505E52A0EDA
S11301700C2B363FE178B9C24B0558C178B72B7EBA
S1130180CC240C43E3FBE1D1C9CD8F05C34B0578E7
S1130190B7C8052B7EFE09CA9D05FE20D8E53E7F23
S11301A0CD240C2A0E0C7DE63FCAB3052B7EFE7FC0
S11301B0CA9E05E1C978B7CA4B05CD8F05C3B505FD
S11301C0CD8F0578B7CA4B052B7E23FE30DA4B055D
S11301D0FE3ADAC005E65FFE41DA4B05FE5BDAC0A3
S11301E005C34B05FE41D8FE5BFAF205FE61D8FE5D
S11301F07BD0EE20C93AC92DE680C264003C32911E
S11302002D2A932D3A902DB7C26400CD0205E92A18
S1130210952D32972DC30406211701C30406DB185C
S1130220B7CA6400FE19CAF505473A982DB8CA182A
S1130230063A992DB8CA0F062A822D702D3EFFBDAD
S1130240C245062E3F3A842DBDCA640022822DC3C6
S11302506400E1FB763A882DB7C27D06F3E52A8473
S11302602D3A822DBDCA52067EF52D3EFFBDC273C6
S1130270062E3F22842DF1FB2A862DE3C9E52A8A26
S11302802D3E2DBCCA92067E23228A2DB7CA7E0635
S1130290E1C92A8E2D7CB7C2A406B5C2A406328851
S11302A02DC352062B228E2D21002C228A2DD5EB14
S11302B02A8C2D23228C2D2BC506013A892D4F3EE5
S11302C001CD7808C1D1D25206C3C604DB29E60F9A
S11302D0473AB52DB8C2DC062AB32DE921A22D3543
S11302E0CCE606C36400F53EFE32F72D3E00D3286B
S11302F032A32D32A22D3E06D32B3E04D32BF1C9BB
S1130300F322BD2DEB22BF2DC5F5E122C32DE12241
S1130310C12DC9CD0007D121040019E52A0020E52B
S11303202A0220E521CB07E5210420E521B72D0E83
S1130330041A7713230DC2310721C92D7EF680EEEE
S1130340807721002011B72D0E041ABEC26707134F
S1130350230DC24A072ABF2DEB2AC32DE52AC12D3E
S1130360E52ABD2DC1F1C9214F5622BB2D21B62D41
S113037036043A922DCD7E07DAC604C35507F5CD6F
S1130380D907C1D8C51AE61FC60326006F1911A7DD
S11303902D0E087E1223130DC293072AAB2D7CB5B4
S11303A0C2A907110102F137C9F1E607EB2AAD2D05
S11303B0E54F2AA72D06013AA92DCD7808E1C9CD2C
S11303C00007D121040019E5C32807CD0007E12265
S11303D0B92DE122B72DC33907E5F5E6074F3AA059
S11303E02DB9CAF907210000AF32A02D110028064B
S11303F0013E04CD7808DA5F08AF32A02DCD650840

Hope this is useful to you.

falter
March 16th, 2018, 05:12 PM
Also, I have ordered replacement transistors as well as 2.0mhz crystals. I'm going to reverse the mod I did to bring pin 49 up and try a 2.0mhz crystal on the board itself to see if it helps run better.

I believe at this time the CPU board is working. I can clearly see initial screen garbage being cleared, followed by what should be a prompt. I've noticed if I change the memory configuration jumpers on the video board to anything other than F800, when I power up, I get a full screen of garbage -- no clear, no prompt. It goes back to prompt if I return the jumpers to F800. I'm hoping the transistor/crystal solution helps matters.. otherwise it's down to ICs and since this board unfortunately isn't socketed, that's going to be a pain.

Dwight Elvey
March 16th, 2018, 05:43 PM
Also, I have ordered replacement transistors as well as 2.0mhz crystals. I'm going to reverse the mod I did to bring pin 49 up and try a 2.0mhz crystal on the board itself to see if it helps run better.

I believe at this time the CPU board is working. I can clearly see initial screen garbage being cleared, followed by what should be a prompt. I've noticed if I change the memory configuration jumpers on the video board to anything other than F800, when I power up, I get a full screen of garbage -- no clear, no prompt. It goes back to prompt if I return the jumpers to F800. I'm hoping the transistor/crystal solution helps matters.. otherwise it's down to ICs and since this board unfortunately isn't socketed, that's going to be a pain.

Do remember, the oscillator divider is set for the 1.8432MHz, not 2MHz! You have to cut the trace under the board and jumper that jumper between IC1 and IC2 to set it to 2MHz. You also need to reconnect the trace you cut from IC1 pin 2.
Dwight

falter
March 16th, 2018, 06:08 PM
Thanks Dwight.. I'm getting a bit confused about the crystal/clock speeds. My understanding is right now the clock the video board is getting from pin 49 is 1.8432mhz.. and that isn't working even with the divider set up for it. Is it your belief that my sole problem potentially is the transistor, and not the crystal?

Dwight Elvey
March 16th, 2018, 09:07 PM
Thanks Dwight.. I'm getting a bit confused about the crystal/clock speeds. My understanding is right now the clock the video board is getting from pin 49 is 1.8432mhz.. and that isn't working even with the divider set up for it. Is it your belief that my sole problem potentially is the transistor, and not the crystal?

Oops wrong button. Text in next post.

Dwight Elvey
March 16th, 2018, 09:14 PM
Yes, I believe the problem with the horizontal is the poor amplitude. The 7407 is driving some amplitude but is is not strong enough for the display to sync on. The code you have is not of much use. It is only 400H in size. There are calls and jumps to a number of addresses from 400H to over 800H. This means that it is likely missing the other EPROMs that would be in the set.
Most of the addresses are in the 400H to 700H range. I only found one stretch of text and it was (Cmf abort). The () are part of the code. There is a jump table of 19 jumps at the beginning of the code. It looks like 3AH is the entry. it does a number of outputs to port 20H and 21H. These are not normal Poly88 ports. There is a loop that I've decoded using the base address of 0:

6C MVI A 32
6E MVI M 0
70 INX HL
71 CMP H
72 JNZ 46E

Note the jump address. The EPROM is intended to run at 400h and not 0. As I said before, you need to get a copy of the V4.0 code. This EPROM has nothing that is useful for you to get the Poly88 running. The fact that it is clearing the video may just be a runaway push instructions. The data you are seeing is likely errors in the RAMs and not missing text. The EPROM does contain 8080 code.
As for crystals, since the board was wired to run the crystal and jumpered for 1.8432MHz, it is likely the original crystal was 1.8432MHz. It should run fine with the 2MHz crystal but you will need to change the jumper such that IC2 pin 3 is grounded and not attached to the pull up net that you saw the resistor on.
The V4.0 code is on the web someplace, I'll look for it.
Dwight

Dwight Elvey
March 16th, 2018, 09:22 PM
I believe the zip file has the V4.0 image. Do check the note at the bottom of the page. There may be a bit error in the monitor's file.
http://www.retrotechnology.com/restore/poly_dwight.html

Dwight

Yep I was right, the zip file has the Monitor code. It is in Intel Hex so you need to convert it to .S format. It has the two errors as noted in the text file.
Once you get your serial board made and if you have a machine that can run real MSDOS, you should be able to load all the images and make tapes. You do need to have RAM starting at 2000H as I recall but you need to check the Poly88 manual to be sure. You need at least 16K to run most of the programs.
For running my download program, do note, I state you need to run real MSDOS. That is not the same as running a DOS window in windows. Windows thinks it owns the I/O ports for the serial. It will steal the occasional byte, breaking the transfer. On most of the old machines, you need to edit the auto exec.bat to not start windows.
Dwight

falter
March 16th, 2018, 10:11 PM
Hi Dwight..

For clarity.. the chip I dumped that code from is the purple one marked v .75 on the left of the white ROM in this picture:

44367

The sources I have been reading from say the white NEC ROM is the 4.0 ROM. The board pictured is my second CPU board.. the first one and the one I'm actually testing with only has the one white NEC ROM.

Dwight Elvey
March 17th, 2018, 05:58 AM
Ah, my error. I'm getting old and have significant brain fad. Once you get things working, you'll want to use some of the other stuff in the ZIP file.
Do you have a keyboard connected yet? If the monitor is working, you should be able to bring the monitor front panel mode data up by typing a control Z. That will give you a lot of characters on the screen to look at.
See the page 61 of:
http://www.hartetechnologies.com/manuals/PolyMorphic/Poly%2088%20Vol%202%20Oper%20%26%20Software.pdf
It describes things.
The display you have doesn't look right for the normal boot. You might try it with the .v75 EPROM removed. I don't recall if the monitor ROM looks for other ROMs to do boot. I'd need to look at the listing again. In any case, you don't have the hardware it is expecting to talk to and the monitor should work fine without it.
Normally the V4.0 will boot with a square in the upper left. This is waiting for you to type B or P to select the cassette recording type.
Your screen was not showing that. It is either an issue with the .V75 EPROM being there or it could be there is a problem with on the CPU card.
The most common problem with these cards is the RAM chips. If things are working right, you should be able to bring the front panel mode up with a control Z.
It will help you work on the video issues as well.
I'm looking at the daughter board of your CPU card. I think that is the patch for using DRAM cards. I recommend using SRAM cards for the start. Later, you can switch to DRAM cards to see if they work right. In any case, you don't need any additional RAM right now to get the CPU and video card working. It would only be a distraction.
Dwight

falter
March 17th, 2018, 07:18 AM
Thanks Dwight.

Definitely won't be the v .75 chip causing what I've got onscreen.. I haven't had it plugged in for any testing that I've done... the spare card pictured that that chip came with has not been tested at all. I using my other CPU card with just the white ROM by itself.

I have tried swapping RAMs around but no change in display at all. I'm really hoping the transistor fix will give me something more readable.. and that I don't have to go removing other chips to figure this out.

I'm wondering for keyboard if I could create an adapter and connect an Apple II keyboard for now? Don't want to take apart my SWTPCs..

Dwight Elvey
March 17th, 2018, 08:08 AM
It is not all that complicated to wire up an adapter to use a printer port of a PC as a keyboard.
You'd need a 7404 to make the nAck and the Busy signals for the PC side. Since you don't need the negative voltage, you can jumper that line to the int* of the 8212. You invert it once for the Busy and once more for the nAck.
All the rest are wired across or jumpered to ground at the respective connectors.
You can then use a simple program in BASIC ( or language of your choice ) to send bytes from you input to the LPT port or LPTn through the system.
If this sounds too complicated, I can make up a wire list for you. You have to make the program.
I believe the Apple keyboard would work but the printer port is easier. I don't know if the Apple keyboard needs a negative rail. If so, you'd need to make it the right level on the video board ( there are some locations for that to be done with a resistor and zener ).
Dwight

falter
March 17th, 2018, 08:44 AM
Yeah I'm looking at the keyboard pinout right now and it looks a little complicated for Apple II. I'm trying to find the pinout for my Nascom1 keyboard.. that might be easier.

I forgot I also have my homebrew CT1024 with detachable generic parallel keyboard. If I could figure out the pinout for that that might be easiest.

falter
March 17th, 2018, 09:05 AM
Strange one.. just for my own curiousity.. I swapped 7407s between CPU and VTI boards... same screen as before but now the reset button lights up?? Looks like the 7407 works with the reset circuit.. so I guess the one I had in there was partially faulty.

Dwight Elvey
March 17th, 2018, 09:29 AM
Here is the adapter since I'm wait for the rain to stop before working on the Ford.
I hope it got my spacing right.
Dwight




Poly PC


1____________________________2 D0

2____________________________3 D1

3____________________________4 D2

4____________________________5 D3

5____________________________6 D4

6____________________________7 D5

7____________________________8 D6

8 ____________________________9 D7

9____________________________1 nSTROBE


10 GND GND 12 Paper End
11 GND
12 GND GND 20
GND 21
GND 22*
GND 23
GND 24*
GND 25

___________11 BUSY
1 2 | 3 4
13 INT- _____|>o____|____|>o____10 nACK


__________13 ON-LINE
|
14 +5V________4.7K___|_________15 nError


7404 pin 7 to GND
7404 pin 14 to +5V
On video board remove R14, CR2, C28, C29
Wire jumper IC41 pin 23 to KB pin 13
Ensure JMP8 is to +5V ( not +8V )
You most likely don't need all the GND wires but you should wire
those with * as minimum.

falter
March 17th, 2018, 03:55 PM
Made a bit of progress. Drawing on long experience, I removed the 74123 and replaced it. Now the screen is MUCH more readable and easier to dial in on my monitor (dang those 74123s!). I'm wondering if this is the square I'm looking for:

44408

If I play with the vertical hold, it does expose those other squares:

44409

I think the board is forming characters properly now with that 123 changed. Here is a screen shot of what happens if I set the memory jumpers on the video card wrong:

44410

I can't center that properly unfortunately. Won't let me.

Dwight Elvey
March 17th, 2018, 04:16 PM
There are two pots on the video board. One adjust the width of the text, while the other adjust the centering. Did you try to adjust those?
One effects IC34. Just curious, which did you replace?
It is looking close. For gable, it looks like good garble.
Dwight

falter
March 17th, 2018, 04:22 PM
I replaced the 74123 at IC34. I think that's the only one there is. I will note I didn't have any 74LS123s on hand.. and am using a regular 74123 instead.. don't know if that makes a difference? And yes, I tried adjusting the pots all over the place. Can't quite get it.

Wish someone had a video of a working Poly-88 out there.. then I'd have something to compare to.

I'm thinking it's sort of working at this point. It'll take me a while to wire up a keyboard. Wish there was a way to simulate the keypress required to take it further.

Dwight Elvey
March 17th, 2018, 05:10 PM
You could program an eprom. It is memory mapped so you can load counts if you want.
Dwight

Dwight Elvey
March 17th, 2018, 08:35 PM
I was counting the number of lines. You are not getting enough lines. It looks like your only getting about 9 or 10 lines. You should be getting 16 lines. IC15 counts the lines and IC12 decodes the count to determine when to generate a reset of IC15.
IC15 has two parts. The first part clocked at pin 1 counts the scan lines of the character. I can see the right number of scan line for the characters so that part is working right. The part that doesn't look right is the part clock at pin 13. I wish you had an oscilloscope. It looks like either IC15 isn't resetting properly or IC12 is not decoding right. It is possible that IC4 is holding the enable on IC12 active all the time. We can see that the screen isn't blanked all the time, suggesting IC4 only clocks when it sees state D of IC15 going to 0, starting the video blanking. I'd say that IC15 is the likely problem in that it doesn't properly reset.
This really need to be looked at with an oscilloscope. I looks like IC15 isn't fully resetting. IC34 failing to make a reset pulse could also be an issue.It would leave a partial count in IC15. Instead of restarting at the top of the screen at 0 it would have a partial count that was used to create the vertical pulse.
Now, this all assumes that the video is syncing on the vertical pulse. It might be that there is no vertical pulse.
If you play with the vertical hold, can you get it to stop were is chops the lines at different places? This might implicate the 7407 you got from the CPU card.
Dwight

falter
March 17th, 2018, 09:23 PM
It is very hard to hold the vertical in place. The blocks that appear onscreen eventually end up out of sight at bottom. Or if I somehow get that one large block to top left.. it will hold for a while but then shift upward one line at a time until it disappears. Sometimes on power up that block will randomly appear over to the right or left. Not a lot of control.

Dwight Elvey
March 18th, 2018, 05:29 AM
In the video signal, there are 3 levels of signal. The white is the most positive. On this board, that is generated by the VIDEO+ signal. We can see that on the screen. There are two black signals on this board VBLK- and DBLK-. DBLK- is to blank the screen between characters, when the addresses and such are changing. VBLK- is to darken the screen when approaching the vertical retrace. It is not the actual vertical signal. That comes from the third level. It is sometimes called blacker than black. Both the horizontal and the vertical sync are of this level. On this board, they come from the CSYNC-. I assume that means Combined SYNC. They are combined at IC3 pin1. Input pin 2, for the horizontal, and pin 3, for the vertical pulses. The horizontal is straight forward. it comes from the oscillator divider, IC1 and IC2, and runs at about 15KHz. The vertical is a little more complicated.
You seem to be missing the vertical sync. With your logic probe, can you look for pulses at IC3 pins 1, 2, and 3.
Dwight

PS also check IC3 pin 6.

falter
March 18th, 2018, 08:36 AM
Seems to be pulses on 1 2 and 3.. 6 also but 6 sounds kind of strange. Here's a video of me testing all four pins:

https://youtu.be/PLtHZrz7iyo

falter
March 18th, 2018, 09:38 AM
Just fired it up again and now, again, I can't control the horizontal.. cannot get a stable picture. The randomness of when it does it is kind of bothersome.

falter
March 18th, 2018, 10:35 AM
I just remembered that I have this OSI 542 keyboard kicking around. I think it'll work.. just need to find the instructions with pinouts.

44418

falter
March 18th, 2018, 10:39 AM
Found the schematic for the keyboard here (http://www.osiweb.org/manuals/542B.pdf). Seems pretty straightforward.. my only bit of confusion is which pin is Strobe, and do I need to worry about all the others like NI, A2 A1 (I think that's what they are.. hard to read the scan quality is bad), R/W, etc..

Chuck(G)
March 18th, 2018, 10:42 AM
Still resisting getting an oscilloscope? :)

falter
March 18th, 2018, 11:07 AM
Not resisting, just taking time making a decision. This 2 channel usb doesn't seem to work at all. I have a bad habit of impulsive buys on stuff like this (like my first logic probe) before realizing ah.. this wasn't the right one. I think I've settled on a Hantek 4 channel USB.. 100mhz unless anyone thinks that's a bad idea.

Chuck(G)
March 18th, 2018, 11:12 AM
Why not the 4-channel Rigol? You can d/l it to your PC, but it's got its own screen--which is very handy.

falter
March 18th, 2018, 11:19 AM
I was looking at those but this is where I get confused. Some are $300 like this one: https://rover.ebay.com/rover/0/0/0?mpre=https%3A%2F%2Fwww.ebay.com%2Fulk%2Fitm%2F33 1941742120 ... but then others are $500-1000 plus. Whats the difference?

Chuck(G)
March 18th, 2018, 11:43 AM
The one you point to, the Rigol DS1054Z, is a 50 MHz, 4-channel scope, which is fine for your vintage applications.

Other models have a larger frequency range or a different number of channels. The one you point to is offered by the authorized Rigol dealer, so you'll probably get good support.

If you need a 100MHz scope on the same budget, you can get one for about $300, but you get only 2 channels. Memory depth also matters, though not as much as it used to.

The DS4024E, in contrast, is a 4 channel 200 MHz scope and runs about $2K. There isn't a linear relationship between max frequency and price--it's more like an exponential one.

Dwight Elvey
March 18th, 2018, 01:33 PM
Seems to be pulses on 1 2 and 3.. 6 also but 6 sounds kind of strange. Here's a video of me testing all four pins:

https://youtu.be/PLtHZrz7iyo

As much as I know about logic probes, it sounds about right. The pin 6 is 60Hz. I assume the probe is dividing it down. Pin 1 is a mix of the 60Hz and 15KHz, as is pin 3. Pin 2 is a partial count of 15KHz. This says the problem is likely either the 7407, transistor or both, unless the frequencies were way off. Can I see a video of it when it is close to locking? It would help to see if there is any signal drarker then dark.

This keyboard isn't going to do much good without something else. There is no indecoder. The SWT uses software to convert things from row and column into ASCII. The Poly88 expects ASCII as input. You could write some code for an Arduino ( not sure which size, I'd have to count the needed pins ).
Dwight

PS
I did find some other text in the .V75. I found (Error _) where the _ was a 00H. I suspect the string was intended to have a number added.
Dwight

falter
March 18th, 2018, 03:02 PM
Ok here's a clip of me starting up from off, and then trying to adjust. Based on the (hard to see) black borders.. it seems like that big block is supposed to be top left.

https://youtu.be/CMvgzopBnzc

falter
March 18th, 2018, 03:26 PM
Here's a photo showing the 'borders' or bands between things. When I took this shot, I fired up the machine and it was all wavy, monitor wouldn't dial it in. And then somehow, magically it pulled itself together. Vertical still kept things creeping upwards but anyway.. you can see the 'cursor' in what I think is the top left position, I'm guessing those other boxes above are supposed to be hidden from view at bottom.

44427

Dwight Elvey
March 18th, 2018, 03:50 PM
Yes, the big block is suppose to be on the upper left. The other things at the bottom of the screen image are something else wrong. I can't tell if it is from the code of errors in the video RAMs. It looks like the CPU is constantly writing to the video. It might be more useful to look at the uninitialized garbage, since I can do counts. The horizontal and vertical pulses look time about right but lack the contrast I expect to see. At one point there is loss of horizontal sync, again showing only a little bit of darkness. I still get the feeling that there is something wrong with output stage. The darker than dark normally has more contrast.
Can you do a video with the processor removed from the CPU card. That should show the garble but I can count characters across an down. It should help to analyze if there is some counting errors ( something that is much easier with an oscilloscope ).
How are is your parts order coming?
Dwight

falter
March 18th, 2018, 04:03 PM
Here it is. Note after a few seconds of leaving it be I start messing with the vertical and then later horizontal control.

https://youtu.be/nS315fxdhpQ

Dwight Elvey
March 18th, 2018, 04:08 PM
I was just looking at the still shot on post 207 and noticed that only 1/2 the sweeps seem to be displaying? I'd noticed that the video, that the vertical had higher contrast than the horizontal. The video was enough out of focus that it was hard to see the missing lines. This would tend to think that only every other horizontal sync was happening. I need to think about that a little. Video is scanned twice as a normal operation. That means for each dot, there should be two scans on the video for each dot.
I'd still like to see a screen of garble synced as best you can.
Dwight

falter
March 18th, 2018, 04:22 PM
This is as close as I can get:

44428

It's like it won't all fit on one screen. If I dial the width all the way counter clockwise it pulls everything onto the screen but wraps over itself. If I go the other way it spreads out so wide you can't see it all without using position trimmer to scroll sideways. There is no way to bring it all onscreen without it wrapping over itself.. as if you were folding an overhead transparency.

falter
March 18th, 2018, 04:28 PM
And again.. this is using an old Samsung VM4509 with the horizontal control cranked all the way in one direction. If I try this on my Commodore monitor it's just garbage.

Dwight Elvey
March 18th, 2018, 04:42 PM
I think we are expecting too much. There is a jumper above IC16. It is actually JMP4 in the schematics. First cut it from 64 and move it to 32 ( the poly88 expects to be 32 characters, not 64 ). This may help with the horizontal sync as well. The vertical still looks wrong. I'm giving it some more thought.
Don't worry about the wrap. I think that is related to the vertical problem.
Dwight

falter
March 18th, 2018, 04:45 PM
Will do. I had assumed it was supposed to be 64 characters because I had read somewhere in one of the scanned manuals that it was the default.

Dwight Elvey
March 18th, 2018, 05:53 PM
Ok, I've given the problem a lot of thought.
Except for the horizontal sync pulse, only the vertical counting uses the further divided clock. We are seeing both a vertical failure and a horizontal sync failure. IC15 counts the lines displayed of each character with the clock at pin 1. Each character uses a total of 16 lines. On the 16th line, it increments the other half of IC15 through pin 13. There are suppose to be 16 characters high lines that are 16 scan lines high. Then there should be some dead time to do the vertical retrace. This should be 262.5 scans. 16x16=256. That means there should be 6.5 horizontals between the last character and the and the top of the screen.
It looks like we are getting only about half as many horizontal counts as we should be. That would mean that one of the two counters, IC1 or IC2, is counting to a full 16.
I'm thinking IC1 since it is a divide by 9 but it could be IC2 or IC3 pin10 not pulsing or not pulsing soon enough. IC2 is a divide by 13, also close to 1/2 of 16. IC1 is the fastest one and the most likely to fail.
Dwight

falter
March 18th, 2018, 07:29 PM
And that would be one IC I don't have a spare for. Are there any substitutions that would work? I have 74163s...

falter
March 18th, 2018, 07:59 PM
Actually I'm wrong.. I have tons.. a whole bunch of working Apple IIs. I removed the 161 at IC1 from the VTI board, installed a socket, then installed one of the chips from the Apples. Made no difference.. so I guess on to IC2?

You mentioned IC1 is a divide by 9 and IC2 is a divide by 13.. on my board IC1 and IC2 are the same.. both 74161s.

Dwight Elvey
March 18th, 2018, 08:18 PM
Without modifications they are both 4 bit binary counters( you know 16 counts from 0000 to 1111 ). If you look at the diagram, you'll see that the terminal count feeds back into an inverting function to the load* pin. This causes the counter to load the input value on the data inputs instead of counting from zero.
Remember those pins I asked you to measure the resistance to ground. IC1 has the pins 3,4 and 5 tied to a pullup and pin 6 tied to ground. That means on the first count after 1111 binary, it will load the counter with 0111 instead of 0000, as it would do without the load, shortening the count to the next terminal count of 1111. Being:

0111
1000
1001
1010
1011
1100
1101
1110
1111 being 9 counts to the terminal count, outputting on pin 15

If the counters are not the problem, it may be the clock generator on the CPU card. It takes the crystal frequency and divides it by 9. Before hacking into the CPU card, you might wait for the 2.0MHz crystal. We'd need to change the load address for IC2, for the 2.0MHz oscillator. Right now it is 0011 for the 1.8432MHz. The 2.0MHz needs it to be 0010 instead. That is what jumper between IC1 and IC2 is there for. It removes pin3 from the pullup and brings it to ground. If you look at the schematic, you can see that IC1's terminal count goes through IC5 ( a nand function, wired as an inverter ) to IC2 clock input. That also feeds back to IC1's load* pin. We know that some type of clock is getting to the rest of the circuit so it is unlikely the a nand gate can divide by about 2. IC2's feeback goes through IC3, a nor gate, wired with both input together making it an inverter. It may be bad but I believe its function of tripping IC34 is working.

falter
March 18th, 2018, 08:50 PM
Okay. I ended up removing and changing out IC1, IC2 and IC3. I've finally gotten good at removing ICs without damage.. this Japanese solder sucker is amazing.

Nonetheless, it didn't help. In fact, the video has gotten worse and stayed that way despite swapping the old ICs back. It does that.. sometimes if I let the machine sit for half an hour unplugged it comes back.

falter
March 18th, 2018, 09:31 PM
Yeah there's definitely something transient going on here.

For 15 minutes I could not get the horizontal sync to pull the picture into something legible, no matter what ICs I swapped around or tried with the width and position jumpers. Then I unplugged and let it sit for half an hour. Replugged.. right away I can see stuff properly again. Seems to really like being unplugged for a while.

Dwight Elvey
March 19th, 2018, 05:11 AM
It is possible that it is the processors oscillator. I think at this point before going further, you should wait for the 2MHz crystal. You can then isolate the video card from possible problems of the CPU card. This is a difficult board to analyze remotely without an oscilloscope. We'd be able to check things without pulling parts.
I thought it might be a good idea to check your monitor frequency. I've been assuming your someplace were things are 60Hz and not 50Hz. Is that true?

Can we see another picture of the uninitialized data with the jumper set at 32 characters instead of 64?
Dwight

falter
March 19th, 2018, 07:06 AM
Yes everything is set to 60hz as far as I can tell.

I will changeover to 32 char today. I was initially reluctant because the 'jumper' between the pads that currently set it to 64 char is actually a trace.. so this card was set up from factory to be 64 char. Thought I'd play with it a bit more before I cut that.

Dwight Elvey
March 20th, 2018, 05:45 AM
Hi Falter
Can you check the value of C34. It is suppose to be a 22pf.
I'm trying to think of anything that could cause the problem.
Dwight

Chuck(G)
March 20th, 2018, 07:13 AM
I'd be suspicious of any components around the 74S124 VCO. Try tapping on the various passives. In particular, trimmer resistors are notorious for going intermittent with age. I wouldn't expect a low-value ceramic capacitor to change much with warmup. Resistors, yes.

falter
March 20th, 2018, 09:31 AM
I'll test these later today. Thanks guys.

A batch of 2mhz crystals will be here next week so I can reconfigure the VTI for that and see what I get there.. I think Dwight had the right idea.. separate the video and CPU card so we can figure out where the problem actually lies. It bothers me that I still can't get my regular Commodore monitor to sync horizontal at all while my Sanyo requires the horizontal control dial cranked all the way in one direction just to barely eke out an image. I think until we get the the VTI standing on its own, we might be chasing our tails.

glitch
March 20th, 2018, 10:04 AM
Again, "H HOLD all the way to one end" was the symptom I observed with my VTI running at the wrong frequency (2 MHz system clock instead of the 1.8432 MHz system clock it was designed for). If you've got another S-100 system up and going, you *can* test with that -- IIRC you've got a SOL-20 that works, right?

Chuck(G)
March 20th, 2018, 10:25 AM
See, a 'scope on the horizontal sync input could identify *exactly* what the frequency was.

Just saying...

Dwight Elvey
March 20th, 2018, 03:14 PM
I'm going back to someplace we looked early on. Do you get a tone at IC4 pin 8. I would expect one similar to the 60Hz we heard before in the video. I recall you said that even after replacing it, it didn't show a high or a low. Since we replaced both IC4 and IC31, that only leaves IC12 as a possible problem source.
If later you should decide to replace it, please inspect carefully for possible solder bridges or even a possible trace bridge. Only pins 4 and 5, of IC12, are suppose to connect together.
Dwight

glitch
March 20th, 2018, 03:19 PM
See, a 'scope on the horizontal sync input could identify *exactly* what the frequency was.

Just saying...

Sure would, rather than guess and check.

falter
March 20th, 2018, 03:25 PM
I'm working on it Chuck. It's been a long time but I think I'm confident enough to buy one.

Glitch: My Sol-20 does work.. I am a bit nervous about plugging any of this stuff into it though.. it's a little cranky sometimes.

Just thought I'd post a pic of the backside of my CPU card.. looks like it's been modified.. anyone have any idea what's going on here?

44453

Dwight Elvey
March 20th, 2018, 03:41 PM
There were a number of modifications added to the CPU card. Look at the manuals on:
www.hartetechnologies.com/manuals/PolyMorphic/
page.
The one on the far right looks like the 60cycle clock interrupt.
I'm not sure about the rest.
Dwight

Chuck(G)
March 20th, 2018, 03:52 PM
I'm working on it Chuck. It's been a long time but I think I'm confident enough to buy one.

Just buy the 4-channel Rigol 50 MHz one. It will meet your needs until you get into some really aggressive projects. Then you can trade up. Scopes in working order generally hold their value better than PCs.

It will hugely simplify your troubleshooting. You'll not only be able to verify the presence of signals, but also observe their shape, frequency and level, all at the same time.

It's not as if the oscilloscope is anything new--it was invented in the 19th century, after all.

glitch
March 20th, 2018, 04:01 PM
Yeah, a 20 MHz "Student 'Scope" or "TV Tech 'Scope" will do almost everything you need for old computers. Pretty much any non-toy modern offering will do.

My CPU has similar modifications -- possibly identical, I don't have it in front of me. I'm not 100% sure mine is functional (it clears the screen and puts up a cursor, but I can't get into front panel mode).

Well, putting it in the SOL-20 if the SOL-20 is also unknown is probably not a good expenditure of one's time. I thought you had it up and stable, that's why I'd suggested it.

Dwight Elvey
March 20th, 2018, 05:04 PM
I'd be suspicious of any components around the 74S124 VCO. Try tapping on the various passives. In particular, trimmer resistors are notorious for going intermittent with age. I wouldn't expect a low-value ceramic capacitor to change much with warmup. Resistors, yes.

I'm not so much as expecting the capacitor to change as it to be the wrong capacitor.
As for resistors, these old carbon comp. are known to not only become temperature sensitive but also to change value over time. Usually this is when run close to their wattage rating. These usually increase in value but I've seen a few in tube circuits actually get lower in resistance. So much more useful information would come from an oscilloscope.
Dwight

Dwight Elvey
March 20th, 2018, 06:28 PM
When trouble shooting any piece of equipment, it is always necessary to understand how it is suppose to work so that one can better understand how it can go wrong.
I thought it might be useful to describe what is happening with the clock input at different points in the schematic. First, when I am talking about a counter, I will use the clock pin as the reference. In other words, I will refer to IC1 as IC1-2 counter. This will help to stop confusion with counters that are two in the same IC. For gates, I'll use the output pin, even when talking about some input. Please follow the schematic as I go through the circuit. In this post I'll describe the Horizontal sync and the vertical sync. It is a little complicated but I hope I have broken it down into it various steps so that it makes sense.
First, the 1.85432MHz clock comes in to IC1-2. It is divided in frequency buy IC1-2 by 9, to 204.8KHz.
It then goes to IC2-2 were it is divided by 13 to 15.754 KHz as the horizontal frequency. It also creates the horizontal sync signal through IC3-13. In the 60Hz divide jumpers, it only holds the horizontal sync for 1 cycle at 204.8KHz because it starts the count at 0011. The two high bits are only 0 for one count.
The terminal count output of IC2-2 goes three ways.
First is to the left, to IC3-10, to reset IC2-2 and to trigger IC34-9 oneshot. It resets various parts the happen each line. We'll get back to it but lets follow the clock timed functions the will go to the vertical operations.
As we go back to IC2-2, we come to IC3-4, were it blocks the vertical syncs during the horizontal resets.
Last it goes to IC15-1 clock input. This counter is used to count the lines used by the character. Although, a text character is only 9 line high, the additional lines are use for space until this counter reaches 15. When it does, two things happen. IC14-8 and IC14-6 reset IC15-1 through its async reset pin. The counter only stays in the count of 15 for only a short time as defined by the delays in its reset, and the two parts of IC14, but IC15-1 Qd output has been high for 7 horizontal pulses from IC2-2.
This pulse also goes to IC12 pin1 and IC15-13 counter. This counter counts the number of characters. It has outputs that drive address lines to the RAMs to select the character line to send to the character ROM IC37. When counter IC15-13 counts from 15 to 0, it trips IC4-11. Since we are done with the text, we need to blank the screen, VBLK-, while we are getting read to make the vertical sync. IC15-13 uses a neg-edge clock. This means that clock pin just went to zero so that pin 1 of IC12 is zero. IC15-13 also just reset to 0000 so that its outputs going to the decoder IC12 pins 2 and 3 are also 0. This decodes to a low on the output of the decoder IC12 of pin 15. This goes to IC3-4 where it is combined to be the CSYNC- vertical sync. IC12 pin 15 will stay low until IC15-1 Qd output goes high, 8 horizontal pulses from divider IC2-2. At this time IC12 output pin15 goes back high, ending the vertical sync pulse. The vertical operation isn't done yet. IC12 pin14 is now low ( not used in 60Hz ). Counter IC15-1 continues to count horizontal pulses, another 7 horizontal clocks. When it again goes from 1111 to 0000, IC15-13 counts once to 0001. Remember, IC12 pin1 is zero when the counter IC15-13 counts so the decoder now has pin13 low. Another 8 horizontal counts and IC15-1 Qd goes high making decoder IC12 pin 1 also high. Remember IC15-13 still has 0001. Value on the decoder IC12 is pin1=1, pin2=1 and pin 3=0. That decodes IC12 pin 12 to low. We'll get to what that does in a bit. That extra delay after generating the vertical sync was there to allow some blank space at the top of the screen.
Back to what happens at when IC12 pin 12 goes low. Follow the trace to IC34-1. Its outputs do several things. It resets IC15-13 that counted the character lines an the vertical retrace time. It reset the counter IC15-1 so that it would start at the top beginning of the character. It reset IC4 that was used to start the vertical sync signal.
OK, we have completed one vertical cycle. This is the only part of the display that is specifically controlled by the crystal frequency. We will get to how the characters are created across the screen in the next how it works post.
Dwight

PS I just made an edit. IC15-1 Qd only stays high for 7 Horzontal clocks plus a tiny amount of time that it resets to zero because of IC14-8 and IC14-6.

Chuck(G)
March 20th, 2018, 08:46 PM
Basically, you have it right, but when troubleshooting, I take the telescope from the other end.

For example, on the VTI board, I know from experience what a composite video signal should look like (if not, I look it up).

Put a scope on it and see if the waveform and frequency matches.

If not, ask in what way it differs from what's expected--and what could produce that. Work backward through the logic to see if signals at each stage match what's expected. Eventually, you'll settle on the failing or misconfigured component.
--------------------------------
Programming can be done in a similar way--and in many cases, is to be preferred. Define the end product--if it's report or screen or API, define that first, then work backwards defining components as you go to see how to accomplish the result.

Top-down vs. bottom-up. In the 70s, there was a guy, I think was named Robert Pride ( or something-Pride). He traveled the world, giving seminars on what was called "The Pride System" that followed this approach. He made a lot of money (for the time).

falter
March 20th, 2018, 08:57 PM
I tried getting a read on C34 with my DMM but it just endlessly cycles.

Here is a pic of it:

44470

I think that's a 3 on the left side.

Dwight Elvey
March 20th, 2018, 09:21 PM
It looks to be a 390 that would be a 39pf. Close enough to work.
Now back to the pin 8?
Dwight

falter
March 21st, 2018, 08:12 AM
Very low but distinct pulse on pin 8 of IC4.

Dwight Elvey
March 21st, 2018, 11:47 AM
One more place to probe. IC12 pin 10 should never go low. The feed back from IC12 pin 12 to IC34 should cause the reset of the counters to not let it count that far. It will be a slow pulse at about 60Hz if it does make a pulse.
Dwight

falter
March 21st, 2018, 04:20 PM
I'm not picking up a pulse (too low for probe?) but if I set the probe to TTL mode it shows 'high' and kind of makes a noise similar to previous tests.

Dwight Elvey
March 21st, 2018, 06:54 PM
Can you do a video of the sound.
Dwight

Dwight Elvey
March 22nd, 2018, 09:14 AM
Years ago a friend gave me a Poly88 that was non functional. He also gave me a TEI enclosure with the Poly88 card set in it. One quirk on this Poly88 is that there's a cut out on each side that exposes card edge fingers on the motherboard. Supposedly this was a way to add S100 slots and power supply capacity to the system. You could just buy another box and connect the two motherboard sections somehow. These things were rare and I'd be willing to bet any original connectors or ribbon cables to connect two boxes would be even more so. And that's assuming Polymorphic didn't leave that as a DIY project for the buyer.

In this picture, you can see something with two card edge connectors attached to the near side of the motherboard. That could be a piece of original equipment that got separated from the unit I was given. Or it could be one person's way of making use of the expansion feature.

http://deramp.com/pictures/poly88_inside1.jpg

I was just cruising about for any info on a motherboard to motherboard connector. When I got to Herb Johnson's site, the name Dwight Elvey popped up.

http://www.retrotechnology.com/restore/poly_restore.html

Mr. Elvey, could you shed any light on the Poly88 expansion feature?

Someplace, I saw a price list for the poly88s. They had a separate chassis with an edge connector added to it, making the buss and power extension. How well this worked, I can not say.
Dwight

PS found it:
http://vintagecomputer.ca/download/polymorphic_systems/Polymorphic-Retail-Price-List-1.pdf
Extension Chassis for a POLY88 $375

Dwight

Dwight Elvey
March 22nd, 2018, 09:58 AM
Can you do a video of the sound.
Dwight
Any sound hear is likely a problem with your friend, IC34. If it isn't making a short enough pulse, IC15 can continue to count until it eventually overflow to zero. If you've socketed IC34 you can get one of the original 74123 in there and see if that pulls things back to normal. The time constant of a 33pf and a 4.7K is 0.155us. I seem to recall, without looking it up that the 74123 is suppose to trip as a oneshot someplace, about 1 time constant. A horizontal scan is about 60 microseconds. To make IC15-13 count once takes 15 scans, or about 1 millisecond. It would need to count about 2 milliseconds to get a low out of IC12 pin 10
So, that doesn't leave much. Either IC34 isn't anyway near fast enough or IC12 isn't working correctly. A 74LS138 is easy to find. One more socketed part is probably a good idea.

falter
March 22nd, 2018, 12:05 PM
Here's video of it:

https://youtu.be/LhmywHgiFRk

I usually have my probe set to CMOS and pulse to hear pulses.. not sure if that's correct but that is how I have found pulses before. However as you'll see.. I only detect anything in TTL mode. Switched to CMOS there is nothing.

Dwight Elvey
March 22nd, 2018, 06:25 PM
The circuit is TTL, so you should be reading in the TTL mode. The CMOS assume a larger signal swing in the positive direction. Without pullups, TTL usually doesn't get more than about 3.5V to 3.75V. It won't go to 5 without pullups.
Having a signal there is not a good thing. Again, I'd suspect the IC34 or the IC12.
Dwight

falter
March 22nd, 2018, 07:47 PM
Okay. I don't think it'd be IC34.. I've swapped several different 74123s in there with no change. However, I do note that the original was a 74LS123.. could that make a difference?

If not, I guess we gotta go in and swap IC12.

falter
March 22nd, 2018, 09:05 PM
Ok.. pulled and swapped IC12. Same problem despite trying 3 different chips.


To make matters worse.. I was using the keyboard port to run my logic probe and accidentally managed to short the 5v and ground together. Now I don't have 5v at the keyboard port. Not sure what I blew up to cause that. It was literally a fraction of a second between when I heard the 'bzzt' and cut the power. Don't see anything obviously burned.

Dwight Elvey
March 23rd, 2018, 06:16 AM
I believe it is all on the same regulator. You may have blown a trace or damaged the socket for the keyboard. Those TO3 regulators can put out 3 amps.
If you have 5V elsewhere, it is an open 5V or ground at the socket.
Back to the original problem.
It is possible that IC15-13 counter is not resetting when it gets the pulse from IC34. There is something going wrong in this general area.
Maybe there is a shorted or broken trace someplace.
I still recall you saying that IC4-8 looked to not have a proper level. I'd love to see a scope shot of some of these signals.
Can you borrow a scope from someone?
Dwight

falter
March 23rd, 2018, 05:11 PM
I must have toasted the ground pin for that socket... if I connect to a ground elsewhere on the board with my VMM I get 9V. Which surprises me, because I see this board still has the factory trace in place and was led to believe it was set for 5V by default.

I have a USB scope here but I cannot get the thing to be recognized properly. And unfortunately I live in a small town where this is not one of the popular hobbies (as opposed to mountain biking and climbing) so nobody to borrow from. But I have tentatively placed and order for a Rigol.. so we should be remedying this situation in a couple of weeks, however long it takes for Canada Customs to get their pound (or twenty) of flesh.

I'm still kinda thinking it might turn out to be glitch who's right in the end.. that somehow we're getting the wrong clock frequency to the VTI and that's causing everything to go haywire. Nothing we have done has made the video one iota more adjustable.