PDA

View Full Version : PC6300 no longer POSTs



NeXT
March 13th, 2012, 03:45 PM
I had my AT&T PC6300 working and booting before I put in storage last October. It was a fully punctioning system.
Before I put the system away I double checked that the battery was good. It was reading 2.8v and showed no signs of leaks. The drive was then parked and put away for storage on its side (drives closest to the floor).
I pulled it out the other day and tried to boot it and knew something was wrong. When you power on a 6300 the keyboard LEDs will go from solid to flickering, then they will go out and the system will show something on the screen (POST data). This time however the lamps remained solid and nothing else happened.
I opened the system and discovered that in five months the battery had gone dead and leaked. There wasn't any visible trace damage but the area around C10 to C0 (board is arranged into a grid) all showed light signs of corrosion. I pulled the system apart, cleaned the pins on socketed chips and edge connectors, washed the mainboard and left it to dry over a heating register and cleaned the video card. In the process I found the corrosion had gummed up the reset button and it was stuck on. Some contact cleaner and TLC fixed it.
After letting the boards dry (completely) I reassembled the system and tried again with my POST card installed.
The POST card constantly reported FF (wrong because that means the ssytem is booting) but had the right voltages and blinked the CLOCK light (and in turn the keyboard controller started flickering the LEDs again so that was working again too) but otherwise nothing else happened. It's now showing the same symptoms I saw on my Olivetti M24: Power is fine and visually it's okay even though the battery had leaked however it was otherwise doing nothing else (though in this case I actually had a keyboard and monitor).
Components that show corrosion on their pins are a 0N74LS08N, two M5M4256P memory chips soldered to the board, two socketed MB8264A-15 memory chips, and a diode. Other damaged parts include the power LED no longer works (but receives +5v) and the speaker got hit pretty hard (no clue if it's still any good and if I'm missing a error beep or something)
I can't confirm if the monitor is getting power. This particular monitor never gave any high voltage squeal when it was on though visually it was working.

Chuck(G)
March 13th, 2012, 03:51 PM
Odd--I've liked the 6300 setup because the battery is normally upside-down, so you'd think that any goo would just drop onto the bottom cover plate. Guess not. On the other hand, I've had the speaker drop off its sticky tape and break its leads--I just replaced it with a small piezo unit.

Does the flickering stop? If so, I'd hook up another speaker to see if it's trying to tell you something.

NeXT
March 13th, 2012, 05:11 PM
Flickering never stops. It should though as the POST tests progresses. The flickering however isn't because the BIOS is testing the keyboard controller I think. It's probably the clock signal. You can have the EPROMs removed and it will still do that so I assume it stops because the BIOS is chattering to the keyboard controller.
I attached another PC speaker and got not a chirp out of the thing.

Like I said, the only reason I got battery damage was because I stored the system on its side. If I had known this was going to happen it would of been stored the regular way.

Chuck(G)
March 13th, 2012, 09:15 PM
Let me guess--you stored it so that the battery was on the highest end? :facepalm: I'd probably start by taking the mobo out and inspecting traces under each IC where the electrolyte leaked, even if it means removing the IC.

Do you need a schematic?

Lotsa fun...

NeXT
March 14th, 2012, 04:12 PM
Stop reading my mind. I facepalmed too when I realized who unwise my decision was but like I said, I swear that battery was fine before I stored it. If it was the other way around it would of dropped right off the board. I got no schematics but chip removal should be okay after I borrow my friends reflow station.

NeXT
March 26th, 2012, 02:13 PM
Got the chips desoldered and removed. Same with the speaker. Corrosion was cleaned and there was no damaged traces. The speaker fell apart though from a lot of corrosion.
Soldered the components back down and the problem persists. :(
All I know for absolute certain that there is a clock signal. I can't say what else might be wrong.

Chuck(G)
March 26th, 2012, 02:34 PM
Well, there is a maintenance manual on the web, along with schematics. Time to get busy, I guess... :(

NeXT
March 26th, 2012, 04:43 PM
Damn.

NeXT
February 9th, 2014, 08:04 PM
NecroBUMP.
I've come back to this machine again.
To catch anyone up with my situation The machine was working when placed in storage. Several months later it was taken out of storage and found that the battery had leaked and dripped down the front 2" of the mainboard. The machine now no longer POSTs. The PCB is visually undamaged from the leak but the power LED has died and the speaker has been ruined. Both have now been replaced. Corrosion was found on two sockets in ram bank 1 and were replaced with new ones. Several chips were desoldered so I could inspect for damage under them and found nothing. I reinstalled the chips with sockets.
I have two POST cards, one tells me voltages, clock presence and a two digit POST code. The other card displays DMA, interrupts and two digit POST codes as well. +5 and +12 are good, a clock is present however only one card reports a FE code (or 00, or FF randomly when you turn the power on) and the other reports nothing. After a period the INT15 lamp will come on and if left for a long period you will get DMA 7 lit as well. The keyboard controller and POST cards respond to the reset button so that isn't stuck.

http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/6300t.png
Looking at part of the service manual I can tell that we are not reaching step 3 and I can't tell where in step 2 we might be hanging, if we're even reaching step 2 at all. I do now have a logic analyzer but in this case I am unsure what I should be looking for or at.

Edited: As a sanity check, I compared my EPROMs to Chuck(G)'s image files (http://www.vintage-computer.com/vcforum/showthread.php?28149-At-amp-t-pc6300-bios-v1-43) and they are okay so the BIOS is not corrupted either.

modem7
February 10th, 2014, 12:22 AM
I have two POST cards, one tells me voltages, clock presence and a two digit POST code. The other card displays DMA, interrupts and two digit POST codes as well. +5 and +12 are good, a clock is present however only one card reports a FE code (or 00, or FF randomly when you turn the power on) and the other reports nothing.
I cannot find anything that indicates that PC6300 machines output POST codes.

modem7
February 10th, 2014, 12:35 AM
I do now have a logic analyzer but in this case I am unsure what I should be looking for or at.
At reset, the CPU jumps to address FFFF0, which is an address near the end of the BIOS ROM range.
So what you could do is see if the Chip Select line on each of the two BIOS ROMs is being activated.

If not seen, the problem cause is a very low one.
If seen, then maybe try an odd/even pair of Supersoft/Landmark diagnostic ROMs.

NeXT
February 10th, 2014, 09:29 AM
If seen, then maybe try an odd/even pair of Supersoft/Landmark diagnostic ROMs.
Chip Enable lines are high. I see activity on the Output Enable lines or any of the address lines.
Will the Diagnostic ROMs work in the 6300? I thought those were for the IBM's and their similar clones? The 6300 doesn't even use an AT form factor board.


I cannot find anything that indicates that PC6300 machines output POST codes.
Ah, okay. I was going out on a whim because I saw the M24 had several codes, mind you none of them would be all that useful here.

Progress edit: I'm now fairly certain that we are getting stuck in some sort of loop in Step 2. Watching the RESET and READY lines on the 8086 I can see when the CPU responds to the reset switch and when it begins doing something with the EPROM. This can be confirmed by removing the EPROMs and observing the CPU remaining in a ready state after a reset. To further minimize my system configuration I have removed ram bank 1 and set the dip switches accordingly so now we only have the 512k that is soldered to the board.

modem7
February 10th, 2014, 09:53 PM
I was going out on a whim because I saw the M24 had several codes, mind you none of them would be all that useful here.
You are right. I had forgotten that the PC6300 and M24 share the same BIOS code.
The web page [here (http://www.bioscentral.com/postcodes/olivettibios.htm)] shows seven codes for the M24.


Will the Diagnostic ROMs work in the 6300? I thought those were for the IBM's and their similar clones?
Although described as a PC clone, speaking from a hardware perspective, the PC6300 is significantly different from the IBM PC. I don't have one (or an M24) myself. I know that a few VCF members have a PC6300/M24, so maybe they will comment.
It's possible that there is enough video hardware compatibility for the Supersoft/Landmark ROMs to at least display something on-screen. Suitable ROMs (odd/even) using the PC version of the diagnostics would need to be created.


Chip Enable lines are high. I see activity on the Output Enable lines or any of the address lines.
The PC6300 uses two 2764 type ROMs, i.e. active-low Chip Enable and Output Enable lines.
Both lines have to be low for the ROM to output data.
If the Chip Enable is always high, there will never be any output from the ROM. Are you sure you measured the right pin?

cr1901
February 10th, 2014, 10:06 PM
Although described as a PC clone, speaking from a hardware perspective, the PC6300 is significantly different from the IBM PC. I don't have one (or an M24) myself. I know that a few VCF members have a PC6300/M24, so maybe they will comment.
It's possible that there is enough video hardware compatibility for the Supersoft/Landmark ROMs to at least display something on-screen. Suitable ROMs (odd/even) using the PC version of the diagnostics would need to be created.

Nothing you can do about burning your own EPROMs, but as far as I can tell (without owning one), the M24 uses a CGA-like mode, and can run Flight Simulator unmodified (http://youtu.be/mUCh46_MzZU?t=16m19s)... so it should work.

NeXT
February 10th, 2014, 10:29 PM
The PC6300 uses two 2764 type ROMs, i.e. active-low Chip Enable and Output Enable lines.
Both lines have to be low for the ROM to output data.
If the Chip Enable is always high, there will never be any output from the ROM. Are you sure you measured the right pin?

In short, no.
All I could do was pull up the chip spec (http://www.futurlec.com/Memory/2764-200.shtml) and find a pin. If you say they have to be low then no, they never go low. They remain high. (this assumes we're talking positive logic right now, otherwise they're always low)

pearce_jj
February 10th, 2014, 10:44 PM
Chip enable might be driven from an identity comparator like 74LS520, 521, 688... follow it back and hopefully you'll find something socketed :)

modem7
February 11th, 2014, 01:57 AM
If you say they have to be low then no, they never go low. They remain high. (this assumes we're talking positive logic right now, otherwise they're always low)
In my many years of electronics experience, 'low' always corresponds to the lower of the voltage range for the logic used (TTL, CMOS, etc.), and 'high' always corresponds to the higher of the voltage range.

However, whether 'active' is high or low depends on whether a signal is 'active high' or 'active low'.
Active high: If the signal is named, for example, TRIGGER_1, then a trigger happens (activates) when the signal goes high.
Active low: If the signal is named, for example, TRIGGER_2, then a trigger happens (activates) when the signal goes low.


All I could do was pull up the chip spec and find a pin
Another 2764 spec is [here (http://www.alldatasheet.com/datasheet-pdf/pdf/126049/MITSUBISHI/M5L2764K.html)] and it shows that:
* Both the CHIP ENABLE and OUTPUT ENABLE pins are active low - to be driven low to accomplish their function.
* Page 2 explains how both of those pins need to be low for a 'read' operation.

Motherboard designers do different things. So on one motherboard type, the OUTPUT ENABLE might be tied permanently low, and the CHIP ENABLE line gated low by the ROM decode logic. I've seen the opposite done - CHIP ENABLE tied permanently low, and the OUTPUT ENABLE line gated low by the ROM decode logic. Or it could be that both lines get gated low.
The circuit diagram will show you how it is being done in the PC6300.

RuudB
February 11th, 2014, 06:54 AM
I cannot find anything that indicates that PC6300 machines output POST codes.
What about replacing the BIOS with one that does? I wrote my "own" BIOS by taking parts of other BIOSes (including the Anonymous one) and inserting my own POST codes. At this moment I use the address 80h and a self built card with two intelligent 7 segment display. But having the source code, I could use the address of a LPT port. Sources are free, assemble with NASM.

Trixter
February 11th, 2014, 10:46 AM
Although described as a PC clone, speaking from a hardware perspective, the PC6300 is significantly different from the IBM PC. I don't have one (or an M24) myself. I know that a few VCF members have a PC6300/M24, so maybe they will comment.
It's possible that there is enough video hardware compatibility for the Supersoft/Landmark ROMs to at least display something on-screen. Suitable ROMs (odd/even) using the PC version of the diagnostics would need to be created.

First off, OP should grab the 6300 Service Manual: https://docs.google.com/viewer?a=v&pid=sites&srcid=ZGVmYXVsdGRvbWFpbnxhdHQ2MzAwc2hyaW5lfGd4OjJj MTRiYmM3MTAwZjY1NjA
(to download, click the arrow to the right of the filename on https://sites.google.com/site/att6300shrine/Home/downloads )

Secondly, the 6300 is indeed very different and if anything in the supersoft/landmark ROMs relies on something specific to the 8088 (like the prefetch queue size), or something speed-related (xtal is 24MHz, CPU runs at 8MHz), then the diag roms won't work. I suppose it doesn't hurt to try, but my fear is that the differences in architecture will result in false positives.

modem7
February 11th, 2014, 09:31 PM
What about replacing the BIOS with one that does? I wrote my "own" BIOS by taking parts of other BIOSes (including the Anonymous one) and inserting my own POST codes. At this moment I use the address 80h and a self built card with two intelligent 7 segment display. But having the source code, I could use the address of a LPT port. Sources are free, assemble with NASM.
Per earlier posts, it was identified that the supplied BIOS in a PC6300/M24 does output some POST codes.


Secondly, the 6300 is indeed very different and if anything in the supersoft/landmark ROMs relies on something specific to the 8088 (like the prefetch queue size), or something speed-related (xtal is 24MHz, CPU runs at 8MHz), then the diag roms won't work. I suppose it doesn't hurt to try, but my fear is that the differences in architecture will result in false positives.

NeXT is in that situation of having what appears to be a 'dead' motherboard, and having done the basics (reseat chips, etc.) to try to get it going. In such cases involving an IBM 5150 motherboard, running the SuperSoft/Landmark diagnostics is a good next step because the first thing the diagnostic does (before any tests) is initialise video and display something on-screen. If a display is seen at all, it tells a lot about the state of the motherboard (i.e. it is basically running). And that was the basis for my suggestion to NeXT to try running the SuperSoft/Landmark diagnostics.

But as we know now, the ROM sockets are not getting the required signals to read data from fitted ROMS, and so fitting SuperSoft/Landmark diagnostic ROMs, the standard PC6300 BIOS ROMs, custom ROMs, whatever, is not going to achieve anything.

NeXT
February 12th, 2014, 01:13 PM
Trixter: I got the service manual already form Bitsavers and gave that a good read over. I saved their ST225 diagram because it's way easier to read than Seagate's ASCII art. The POST test note I had earlier came from the manual but that was as in-depth as it went. There wasn't even any schematics.

Modem7: Chip Enable remains high but Output Enable will toggle and I can go burn the Supersoft Roms in a few days. It's a bit harder for me because I have no spare 2764's so I have to wipe the chips, burn, verify and do it again when I want to put the old BIOS back on.

Edited: Actually, I stand corrected. Now that I'm more closely observing with the logic analyzer I'm noticing that things change between resets.
For the following photos, right to left in columns is cpu RESET, cpu READY, rom high CHIP ENABLE, rom high OUTPUT ENABLE, rom low CHIP ENABLE and rom low OUTPUT ENABLE

http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/IMG_5180.jpg
http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/CGS_0635.jpg
http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/CGS_0637.jpg

For reference, here's with the reset button held down.
http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/CGS_0636.jpg

modem7
February 12th, 2014, 10:07 PM
Edited: Actually, I stand corrected. Now that I'm more closely observing with the logic analyzer I'm noticing that things change between resets.
For the following photos, right to left in columns is cpu RESET, cpu READY, rom high CHIP ENABLE, rom high OUTPUT ENABLE, rom low CHIP ENABLE and rom low OUTPUT ENABLE
Okay. There are times when the high ROM has both its CHIP ENABLE and OUTPUT ENABLE pins low together, and there are times when the same thing happens for the low ROM.
If this were an IBM PC, I would be saying to myself, "Time to throw in a Supersoft/Landmark Diagnostic ROM and see what happens."


and I can go burn the Supersoft Roms in a few days. It's a bit harder for me because I have no spare 2764's so I have to wipe the chips, burn, verify and do it again when I want to put the old BIOS back on.
Note that none of the ROM images of the Supersoft/Landmark Diagnostic ROMs at my web site are suitable for you. You need the PC version of the diags in an high/low (odd/even) form. The code in the PC version is 8 KB sized and designed to sit in the last 8 KB of the PC's 1 MB address space. I have created a couple of ROM images that should work for you, and placed them [here (http://www.minuszerodegrees.net/misc/SSDIAG_PC6300_UNTESTED.zip)].
If the ROMS present a display, then good, but because of diferences between the PC6300 and the IBM PC, you should not trust the results of the tests that then execute.

modem7
February 12th, 2014, 11:11 PM
Actually, what would be better to try first would be the ROM images that I have just created and placed [here (http://www.minuszerodegrees.net/misc/PC6300_OUTPUT_33.zip)].
The code only does one thing - outputs 33h to port 80h.
Expect to see "33" appear on your POST card (which I'm sure is monitoring port 80h).

(It would be good if someone with a good PC6300/M24 verifies that "33" appears.)

NeXT
February 15th, 2014, 06:12 PM
Tried your special "33 only" images and the card still sits at either FE or FF so I guess in that case we're back to determining why it's not reading?

gslick
February 15th, 2014, 06:54 PM
You're actually using a retro HP 1600A for real debugging work? That's kinda cool. I've been tempted to buy one of those for a while just because the display looks so cool. If you ever have a need to move up to some slightly more modern maybe I could make you a reasonable deal or trade you something for a 16500B setup.

NeXT
February 15th, 2014, 07:52 PM
I'll keep tabs on you then. I'd trade it away in a heartbeat if you were closer. It's nice and all but really I need something a bit newer.

NeXT
February 23rd, 2014, 02:14 PM
For the heck of it I tried the proper supersoft ROMs and those didn't work either. I'm also not finding any additional leak damage. This totally reminds me of the M24 I had many years ago. It didn't have a keyboard or monitor but it did have the battery leak and just like here, I couldn't even hear a POST beep. All we can conclude right now is something is stopping the machine from even reading the BIOS but there's still minimal and erratic amounts of activity so it's not completely brain dead.

NeXT
April 22nd, 2017, 11:07 AM
BuMP for continuation.

As per the thread on this series machine needing a parallel port POST code reader I purchased one of those and tried it out with no luck.
I have also now burned BIOS sets for all three major BIOS releases, all three of which do not seem to change anything.

Have also confirmed that all PSU rails are healthy AND the CRT is receiving power. A new speaker has been installed as well.

Photos of the machine and/or mainboard are available if anyone wants.

NeXT
May 23rd, 2017, 06:53 PM
I have been loaned a known POSTing Olivetti M24 by The Hackery (http://thehackery.ca/) for the purpose of testing the various pieces of hardware in my machine in an attempt to narrow down the probable fault. I have been asked however not to swap any IC's that are not socketed, so no way to test the DMA chip, but this means I have at least a known working BIOS and PAL set, a good CPU and a few other odds and ends.

1ST1
May 23rd, 2017, 09:19 PM
What does the parallel port POST card displays on it's LEDs. Did you follow the thread here regarding the diagnostic codes of M24/6300?

http://www.vcfed.org/forum/showthread.php?54771-AT-amp-T-6300-won-t-post
http://www.vcfed.org/forum/showthread.php?54605-How-to-read-parallel-port-POST-diagnostic-codes

By the way, you can safely remove the battery from M24 and it still runs. It will only loose real time clock and date. It's an XT, not an AT which stores relevant config data in nvram.

NeXT
May 28th, 2017, 04:51 PM
What does the parallel port POST card displays on it's LEDs. Did you follow the thread here regarding the diagnostic codes of M24/6300?00 or FF. AKA: nothing is happening.

So this afternoon I swapped parts between the M24 and my PC6300.

-I can cinfirm the 8086 is good
-I can confirm all motherboard PALs are good
-I can confirm the BIOS I burned is good
-I can confirm the keyboard controller is good
-I can confirm the WD8250 is good
-I can confirm the power supply is good

I also moved the main logic board from one machine to the other to ce if the video card or ISA board was causing problems. I was getting vertical stripes at this point but an audible POST beep and floppy access indicated to me that the machine was still running and likely there was a PAL revision conflict. The video board between both machines are slightly different with at least one PAL omitted. Anyways, I'll treat that as good.

It was when I got to the ram I started to run into problems. On the M24 bank 0 consists of eighteen 64k x 1 (two columns of 64k = 128K) chips and bank 1 is eighteen 256k x 1 (two columns of 256k = 512k) chips. On my PC6300 the chips in both banks are reversed and in both cases bank 0 is soldered in** whereas bank 1 is socketed.

The other thing I noticed was that on the M24 when the system came up form a cold start the keyboard LED's are solid, then go out and the POST screen is displayed one second later. The LED's would not light on a reset.
On my machine when the LED's should go out they start blinking. On reset they go solid again, then blink after one second. I already confirmed the controller is good so....



**The exception to all the chips on bank 0 are the two closest to the front of the machine which I had previously desoldered to inspect for any trace damage underneath. When tested in bank 1 of the M24 the machine would not POST. Either that means both chips (or more in bank 0 of the 6300) are bad or it did not like me mixing Mitsubishi M5M4256P's with Hitachi HM50256P chips. As a sanity check regardless I tried a Fujitsu MB81256 and that worked, so I'm leaning towards something iffy about that Mitsubishi ram.

Also, christ almighty the dip switch settings passed around the internet are awful. Can someone else with 256K chips in bank 0 and 64k chips in bank 1 (so that is 640K total with no expansion boards) tell me what their dip switch settings are? It could also just totally be right now my settings are wrong.

Trixter
May 28th, 2017, 11:51 PM
00 or FF. AKA: nothing is happening.

My money is on motherboard damage. You've confirmed PSU, you've confirmed parts via swapping... but POST is not even trying to run because you're getting nothing at all out of the parallel port during POST. IMO, the only hope for this board is to start tracing connectivity along every trace anywhere near the original battery damage until you find what's gone.

David_M
May 29th, 2017, 01:05 AM
There a few things that must be functional to get a POST indication on a POST card.

Good power supply, CPU, ROM, reset, data and address bus, a handful of IO control lines and a clock.

Focus your testing on those things.

NeXT
May 29th, 2017, 11:08 AM
My issue is that I can trace the damage path along one side of the PCB about two inches across.Almost every IC in the vicinity I have already desoldered to check underneath and reinstalled with a socket. There are however other components I simply could not remove because desoldering DIPs on multilayer boards simply sucks but that was before I bought a hot air rework machine. You mentioned before that you've seen bad DMA controllers in the past. Probably unlikely here but damn I don't want to test it because that's a lot of holes to clear.

Trixter
May 29th, 2017, 05:18 PM
My issue is that I can trace the damage path along one side of the PCB about two inches across.Almost every IC in the vicinity I have already desoldered to check underneath and reinstalled with a socket. There are however other components I simply could not remove because desoldering DIPs on multilayer boards simply sucks but that was before I bought a hot air rework machine. You mentioned before that you've seen bad DMA controllers in the past. Probably unlikely here but damn I don't want to test it because that's a lot of holes to clear.

While DMA replacement is on my list for my one 6300s, I get a POST code up to the DMA failure. You're not even getting POST, so you don't need to worry about DMA for now.

NeXT
May 29th, 2017, 05:20 PM
Perhaps more bad ram?
The fact that the first two I tried wouldn't work in the other machine raises a flag for me.

Trixter
May 29th, 2017, 05:39 PM
Bad RAM would not prevent a POST code from being output. Here's the very start of the POST, according to the BIOS ROM:


seg000:DB8F i_powerup: ; CODE XREF: seg000:E05Bj
seg000:DB8F ; seg000:EA73j
seg000:DB8F FA cli
seg000:DB90 B0 40 mov al, 40h ; '@' ; Checkpoint #0
seg000:DB92 BA 78 03 mov dx, 378h
seg000:DB95 EE out dx, al ; Printer Data Latch:
seg000:DB95 ; send byte to printer

The very first thing it does is output 40H to the LPT port. No RAM needed for that. Your RAM might be bad, but it's not the cause of the bad POST. In fact, try removing ALL RAM from your system and seeing if you at least get 40H on the parallel port.

NeXT
November 27th, 2017, 09:01 PM
Yes, six months later only NOW am I able to start pulling ram with a new Weller-Ungar 4024 desoldering iron. It took me an hour and a half to get one bank out. I'll hopefully have the other out by tomorrow night.
It's amazing how well proper desoldering tools work compared to chinese knockoffs. :P
So far however I am not seeing any sign of corrosion or trace damage, so it may still very well be ram.

If it doesn't hurt, I can probably now also desolder and inspect under 6B, 7B, 5C-8C, 5D-9D, 1B, 1C and 1D. Those are the only other components that were in the path of the leakage.

Trixter
November 28th, 2017, 09:42 AM
I "solved" one of my bad 6300 boards by purchasing a working one. If you get through with this repair and it is too extensive to fix, send me a PM. (As documented elsewhere on this forum, I have a board that starts to POST but stops at the DMA test, leading me to believe the DMA chip needs replacing.)

NeXT
November 28th, 2017, 11:22 AM
Thank you for the offer Trixter. I will keep that in mind. I was hoping I'd come across a salvaged board if I could not conclude what the problem was as opposed to buying a whole other machine just to salvage the motherboard.

NeXT
November 28th, 2017, 08:12 PM
In what took me an hour and a half yesterday I was able to finish the ram off, then go and desolder both the DMA controller and Interrupt Controller and solder in sockets for them both, just in case we have to come back to them later.

http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/CGS_7770.jpg
http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/CGS_7770.jpg

I'm lazy tonight. I'll see if it gives a different POST code tomorrow.

Chuck(G)
November 28th, 2017, 08:16 PM
Stupid question again--your video cards are complete, right? No missing chips?

Trixter
November 28th, 2017, 08:32 PM
Also: Was that speaker replacement yours, or did it come that way?

NeXT
November 28th, 2017, 08:35 PM
Stupid question again--your video cards are complete, right? No missing chips?
Yes they are complete. Remember that this machine as-is was working prior to the leak.


Also: Was that speaker replacement yours, or did it come that way?

The speaker was the only visible victim of the battery leak and couldn't be salvaged. I put in a temporary replacement in the hopes I could get SOMETHING out of it.

PhilipA
November 29th, 2017, 08:03 AM
Possibly stupid question but is the reset button (I'm guessing that's what the pushbutton is at the bottom left) not got crud inside and making a semi-contact or anything?


--Phil

NeXT
November 29th, 2017, 11:12 AM
That is actually a very good question and that was one of the first things I found when cleaning the mess up.
Originally it was stuck and would not register any presses because of internal corrosion. I had to flood it with contact cleaner before it began working reliably. I can confirm however it is not stuck in reset as a logic analyzer hanging off the CPU shows that everything drops dead when you press and hold the button and comes back to the same random garbage on the bus when released.. I do not however recall if I verified if the RESET pin was being polled.

NeXT
November 29th, 2017, 08:41 PM
Reassembled the machine and plugged in the POST code checker for a bench test.

From cold start (power switch)


01 - FF
02 - 00

From warm start (reset button pressed)


01 - 00

I verified that the reset line and circuit were behaving and I can see it operating fine all over the place but still no change from the machine. I can see something trying to move around the data bus and other various lines go high and low seemingly at random. What is your secret, computer?

http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/IMG_9081.jpg

Trixter
November 29th, 2017, 09:38 PM
Geezus, that's frustrating. It's almost an assassination mission at this point.

Kazblox
November 30th, 2017, 02:49 AM
My only advice at this point would be to burn a ROM that contains nothing but a few opcodes to write to the internal speaker in a loop. If it fails even there, extreme problems might be around.

1ST1
November 30th, 2017, 10:19 AM
Didn't you used the Olivett M24 Theory of Operations service manual? http://bitsavers.informatik.uni-stuttgart.de/pdf/olivetti/m21_m24/Olivetti_M21_M24_Theory_of_Operation_Nov84.pdf

Chuck(G)
November 30th, 2017, 10:23 AM
...there are also schematics of the thing around on the M24 site.

NeXT
November 30th, 2017, 04:24 PM
Wow, when did that manual get uploaded?
Yeah, that's all the trimmings in there. The schematics start on page 241.
I'm going to skim over this but if anyone sees anything I'm open to suggestions right now.

1ST1
November 30th, 2017, 11:58 PM
This is online since long time ago.

Trixter
December 1st, 2017, 12:10 PM
Yes, but to be fair, not a lot of people realize the 6300 is an OEM'd M24, and furthermore, a lot of Olivetti information online is in Italiian. I've owned a 6300 for decades and I only found this manual a few years ago.

NeXT
December 2nd, 2017, 09:18 AM
Alright, as an effort to eliminate a few suspect components I am going to desolder and socket the nine IC's that were also in the path of the leak and do the same on the loaned M24 so I can swap chips back and fourth until I see something fail. New sockets are on order but given the time of year and my short-sighted decision not to use digi-key we might be ready to test again in a month or two, depending on sea conditions in the Pacific Ocean. ;)

1ST1
December 2nd, 2017, 01:41 PM
Yes, but to be fair, not a lot of people realize the 6300 is an OEM'd M24, .

--> http://www.old-computers.com/museum/computer.asp?c=531
--> http://www.old-computers.com/museum/computer.asp?c=535

So I want to tell you all some secret information...

Xerox PC 6060 is also an Olivetti M24 ---> http://www.computerhistory.org/collections/catalog/102662896

Logabax Persona 1600, strange name, is also an OEM version of Olivetti M24: http://www.old-computers.com/museum/computer.asp?c=154

AT&T 6300 WGS is mostly identical with Olivetti M240
---> https://www.ebay.com/itm/AT-T-PC-6300-WGS-Model-XP-1050-FOR-PARTS-/322293335271
---> https://olivrea.de/olivetti-m240/

There is a AT&T 6386 WGS ---> http://0xea.blogspot.de/2013/11/the-at-pc-6386-wgs.html
-----> This is one of the Olivetti M380 models, but I don't know which one.

... just found another page containng some software for Olivetti based AT&T PCs... http://www5.ncr.com/support/support_drivers_patches.asp?Class=pc_library_63xx

NeXT
October 13th, 2018, 02:04 PM
Alright, as an effort to eliminate a few suspect components I am going to desolder and socket the nine IC's that were also in the path of the leak and do the same on the loaned M24 so I can swap chips back and fourth until I see something fail. New sockets are on order but given the time of year and my short-sighted decision not to use digi-key we might be ready to test again in a month or two, depending on sea conditions in the Pacific Ocean. ;)

Okay that ended up taking a year.
Unfortunately it's still a dud. I give up. I've done chip swaps and stripped the affected area down to the PCB to check for breaks and corrosion damage and I cannot locate the fault. The Hackery also wants their donor M24 machine back. Who was it that at one point was offering another board? I would like to take up that offer.

2icebitn
October 13th, 2018, 02:32 PM
Sorry I haven't scanned the whole thread. Have you allowed for a bad solder joint somewhere? I have to assume you've pulled every chip, cleaned sockets and pins. What about the p/s? Could it be delivering flakey power? Have you tried booting with no drives (or cards) connected? I don't know a lot about the 6300, have one here that displays garbage when powered on. Got to get busy myself.

NeXT
October 13th, 2018, 02:52 PM
(I have no rational way to word this post)

Trixter
October 14th, 2018, 12:34 AM
Who was it that at one point was offering another board? I would like to take up that offer.

Me, I think. I've got 6300 parts I can, uh, part with. Send me a PM.

2icebitn
October 15th, 2018, 08:34 PM
When you get yours fixed, we can work on mine:

48771

Trixter
October 15th, 2018, 09:41 PM
Is that before POST, during POST, or after POST?

2icebitn
October 15th, 2018, 10:12 PM
Is that before POST, during POST, or after POST?

That's all she wrote. Plug her in, turn her on, wammo, instant Antideluvian chess board.

Trixter
October 16th, 2018, 10:06 AM
Saw you're selling the mono monitor -- giving up on this system?

2icebitn
October 16th, 2018, 10:28 AM
Nein! I only got it a few months ago. The monitor is superfluous to my needs. And it's not even gone yet. I could opt to sell a Tandy VM-1 instead, as I have 2. I have no qualms about setting the AT & T monitor atop a Tandy 2000. Others would, I wouldn't. Or maybe I'll sell all 3 and have a 3 way switch feeding my NEC Multisync 1. But none of them work so votz the point?

In short, I have to move a number of items soon.

Trixter
October 16th, 2018, 06:09 PM
That's all she wrote. Plug her in, turn her on, wammo, instant Antideluvian chess board.

Bonus points for Antediluvian. One more question: With a 6300 keyboard plugged in and turning the power on, the CAPS LOCK and NUM LOCK lights blink in rapid succession show that the CPU is going through POST. Do your keyboard lights blink when you turn the system on? If so, and you wait 15 seconds, does the system unit beep?

2icebitn
October 16th, 2018, 06:13 PM
You forget that my kb is in pieces. I suppose it's still pluggable ... I would have to make sure. All in good time.
Clearly the Antideluvians are back. And they're making themselves known by sending coded messages through vintage computer screens. This ain't some random fault, I know what's going on.

2icebitn
October 18th, 2018, 08:51 AM
Ok I plugged my keyboard-carcass in and turned it on. The feature lights came on solid for half a second, then just blinked at a furious rate. A different image appeared on the screen, those fluxing anti-deluvians, and I believe it even morphed. Take a look -

488504885148852

2icebitn
October 18th, 2018, 09:04 AM
And no beep/s.

PeterNC
October 18th, 2018, 09:47 AM
AT&T 6300s are fairly common on eBay. You could buy a replacement.

2icebitn
October 18th, 2018, 10:28 AM
Or I could devote the funds to a better spam filter.

Trixter
October 18th, 2018, 06:12 PM
The solid lights, then blinking lights, means the CPU is running through POST, which generally means the board is still worth salvaging. If you got zilch or no flashing lights, that would be "bad".

If you hook up a "laptop diagnostic code" diag board to the parallel port (extended conversation here (http://www.vcfed.org/forum/showthread.php?54605-How-to-read-parallel-port-POST-diagnostic-codes)), the 6300/M24 POST outputs various checkpoints to the port. I have a system that outputs 43h and then nothing else, and looking at the BIOS source code (contained in the "System Programmer's Guide AT&T Personal Computer 6300 (ftp://ftp.oldskool.org/pub/drivers/ATT/6300/Documentation/Original%20Scans/System%20Programmer's%20Guide%20AT&T%20Personal%20Computer%206300.pdf)" PDF), I can see that's right before the 8237 DMA chip test, so I know I probably have to replace the DMA chip on that board someday.

2icebitn
October 18th, 2018, 07:20 PM
Many thanks Jimbo.

NeXT
November 7th, 2018, 06:53 PM
With thanks form Trixter I have received a board that is "better". It's still not a fully working board but it at least initiates POST.

The tester by the way indicates it hangs up at 45h, so either it starts testing the base ram and runs into bad ram (?) or it's actually stuck on a Programmable Interrupt Controller fault. I will have to swap some IC's from the dead board to see if I can rule out the PIC but for ram......the tip on my Weller 5088AS is shot and I do not know where to find replacements anymore. Would piggybacking be of any use?

2icebitn
November 7th, 2018, 07:19 PM
Dollar stores often sell soldering iron. you may have to find the old kind of dollar store. My weller lost its plating. Poc.

NeXT
November 7th, 2018, 07:39 PM
Yeah, it's not that kind of Weller. This station was around $560 at one point.

Trixter
November 8th, 2018, 08:31 AM
With thanks form Trixter I have received a board that is "better". It's still not a fully working board but it at least initiates POST.

Sorry about that -- it was untested, but glad it gets further than your other board.


The tester by the way indicates it hangs up at 45h

According to the BIOS source listing (page 8-90 of the System Programmer's Guide), it outputs 45h when it has completed everything it considers fatal, then puts the original vector table back in place at the first 1024 bytes of RAM, then determines system settings from DIP switches, then enables video, then displays all of the tests that passed onscreen.

Have you verified the DIP switches are set correctly on the board?

NeXT
November 8th, 2018, 11:25 AM
Oh, by no means am I complaining it doesn't fully POST yet. I'm already further ahead than I've been in over half a decade. ;)


Have you verified the DIP switches are set correctly on the board?
Likely no. I think I defaulted them to the base config for the Hackery's M24 board but I believe its ram arrangement was different. I will have to double check those tonight.

NeXT
November 8th, 2018, 06:15 PM
Verified and properly set the dip switches to reflect the hardware and ram config (640k onboard, with 512K on bank 0 and 128K in bank 1, no NPU, one 360K floppy, 8250 UART, 80x25 Monochrome display and no ISA cards installed)

No change. 45h, blinking keyboard LED's and nothing on the screen. :(

http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/IMG_1830.jpg

Went over the board and verified all other jumpers were good and swapped in the known good 1.43 BIOS EPROMs and PAL chips. Replaced the battery damaged power LED and replaced the reset button as only one contact side was working. Also verified continuity on the dip switches. Still nothing.

NeXT
November 9th, 2018, 06:56 PM
*Squishes bug with thumb*

Got it.

I thought about how Trixter said the machine progressed once 45h was passed and if it was not reaching display initialization and was not reaching the programming of the keyboard controller and it could not be hanging on the dip switches the only thing that it could hang on is the vector relocation in ram, indicating the first bank of ram could have a problem.

Bank 0 is 18 of the usual 1-bit 256k DIP chips with 16 pins. Bank 1 is 18 64k chips. I've seen configurations where these were flipped and others were it was all in sockets but on this machine it's all soldered in so testing ram is a pain in the ass. I've never really tried piggybacking as the mileage seems to vary wildly but I had nothing to lose here so I fit in 18 chips and tried again. It worked.

http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/CGS_8202.jpg

Through the very careful process of elimination I pulled chips one at a time until it failed to POST again. Eventually I found one chip would fail every time the piggyback was removed.

http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/CGS_8203.jpg

Okay then. I guess I will have to see if I can get one last shot out of the tip on my weller and we should be good to go. Thanks Trixter!

2icebitn
November 9th, 2018, 07:15 PM
Just leave it piggybacked. Solder it in place. Eventually, in maybe 40 years, you'll have a veritable leaning tower of pisa made of ram chips (my spelling corrector dropped the a at the end of pisa and substituted an s. I should have left it).

NeXT
November 9th, 2018, 07:31 PM
I wouldn't trust it for anything. It might pass the basic POST ram check but if the old chip is sticking or losing bits it won't be long until that becomes a problem as it's still in-circuit.

Trixter
November 9th, 2018, 09:04 PM
Brilliant! I've never had piggybacking work for me, but glad it helped you! And your monitor is working too, nice (2/3rds of my 6300 mono monitors are dead and I lack the skills to fix them)

What's odd is that there is a DMA test earlier in POST that is supposed to test the first 64K and not proceed if an error is found. I guess it exists mostly to test the DMA functionality and not the RAM itself...

NeXT
November 10th, 2018, 08:06 PM
After some fudging with an incorrectly jumpered hard disk controller and some REALLY position sensitive MFM cables it's all together and happy again. That wraps up the thread folks!

http://i11.photobucket.com/albums/a166/ballsandy/Computer%20related/CGS_8210.jpg

1ST1
November 11th, 2018, 01:37 AM
Nice!!!

Valerio
November 11th, 2018, 08:42 AM
Beautiful!

2icebitn
November 11th, 2018, 08:52 AM
You're not truncating this thread before mine is fixed mister! :)

NeXT
November 11th, 2018, 11:47 AM
Alright, the wheel is yours. :P

2icebitn
November 11th, 2018, 11:59 AM
I ain't got the time :(. Maybe in a few weeks.

I need to get one friggin xt working, so I can do chip substitutions. Currently none work. Icarumba!

MrArgent
November 11th, 2018, 05:05 PM
A beautiful specimen! You did great!

2icebitn
November 11th, 2018, 06:22 PM
Specimen - an Italian astronaut? LOL LOL LOL LOL

Valerio
November 21st, 2018, 05:46 PM
When you get yours fixed, we can work on mine:

48771

I'm happy to report that your monitor arrived safe and sound and it works just fine - so the chessboard problem you saw is due to something else...

49506

(not sure why it shows it rotated...)

2icebitn
November 21st, 2018, 05:57 PM
I knew the problem wasn't with the monitor. Maybe a lot of people thought the opposite, and there was only 1 bid :). Glad you're happy. And if you should ever happen to come across a schematic of the DEB, can't remember what Olivetti called it, or photos even, please keep me in mind.

If you're using an Android or IOS device, it happens ...

Trixter
November 22nd, 2018, 10:52 AM
The DEB was the Display Enhancement Board, an add-on that granted 640x400x16 colors and other effects. It's not the actual display board.

I have a spare display board; PM me in December.

2icebitn
November 22nd, 2018, 04:41 PM
You have a spare stock display board? Don't need that. I need lots of colors from my 6300. What's life without color ... except when you're running in monochrome.

I have scanned some portion of the technical manuals from AT&T and Olivetti. But not exhaustively. Is there nothing but a mention of the deb? Would be nice if they published the schematic at least. I mean every other computer of the day that was capable of 8 or 16 colors in high resolution either offered it stock or by means of a chip upgrade. Why did these guys have to be such fuddy duddys?

Trixter
November 23rd, 2018, 12:25 PM
You have a spare stock display board? Don't need that. I need lots of colors from my 6300. What's life without color ... except when you're running in monochrome.

All stock M24 display boards display color in text and graphics modes. The DEB just gives more colors in 640x400, and extra capabilities.

DEB info here: ftp://ftp.oldskool.org/pub/drivers/ATT/6300/Display%20Enhancement%20Board%20information.pdf

2icebitn
November 23rd, 2018, 07:35 PM
thnx, but no schematic or artwork :(