• Please review our updated Terms and Rules here

Certain screen columns repeating themselves (text)

M

mvno_subscriber

Experienced Member
Joined
Feb 14, 2017
Messages
69
Location
Oslo, Norway
Hi,

don't know exactly what to make of this, I was sitting in the evening typing on my word processor (Raytheon VT302), and then suddenly the screen glitched, different characters were shown and I quickly decided I should let it rest.

So I did let it cool overnight, thinking it might be a temperature issue, as I'd been using it for a couple of hours (never been that long before), but to no avail.

Take a look at the following screen shots:

2017-04-01 08.19.21.jpg2017-04-01 09.16.55.jpg

  • On the startup screen, it should say "LOAD PROGRAM DISK", however it seems like something is repeating itself(?), instead showing "AD PROGRAD PROGR" (note it's exactly the same amount of characters, just messed up)
  • The word processor loads fine from disk, but again it seems like text from certain "columns" is repeating itself over and over (the additional garbage in the lower portion is somewhat new, but still repeats itself)
  • If I try to insert a disk, it won't load, showing the text "FULL" which I assume is because it thinks something is wrong -- this happens regardless of whether I use the disk that was in when the glitch occurred or not.

For reference, here is how it should look when everything is OK:

2017-02-14 20.29.55.jpg


I've tried reseating all reseatable chips on CPU/Memory/Display controller, in addition to swapping all DRAM chips (the CPU board with the 8085 CPU has two rows of DRAM chips, I just swapped those, thinking if it was a memory issue, it would behave differently, but it didn't).


I was writing stories for my daughter for Easter, I had promised they'd be done until next weekend - hopefully not all is bust.. suggestions would be most welcome, as I have no idea where to start here (especially since the error didn't seem to only affect the display).
 
If you can scope the screen memory address lines from the video controller, it's likely that you'll find one that's "stuck". That should lead you to the offending IC pretty quickly.
 
Hm, nice tip! I have to figure out where that goes then... (and seriously up my skills :))

I tried mucking a bit more with it, and noticed that the keyboard registered double presses (there's a "bip" for every keypress, and it beeped twice in a row). So definitely some timing issue, and I'm thinking it has to lie on the CPU/Keyboard/Printer/RAM board. It hangs after I've pressed a few keys.

So something isn't timing correctly, and something else goes out of sync as a result. Or, that's my wild guess at least.. Can't find any oscillators, so could it be that the internal clock in the 8085 CPU does most of the timing in the system?

A bit late now, I'll go to bed and probe the CPU tomorrow. Here's a full size picture of the board:

2017-04-01%2023.26.20.jpg


As always, all comments are extremely welcome! :)
 
Thanks for taking the time, Chuck(G), I knew I probably had missed something obvious. Please pardon my poor skills!

Had a quick session before having to go to the office (groan), if anyone would like to comment on the measurements here it would be great as this is new territory for me. The scope I use is brand new and my first, so please excuse any silliness.


Output from the OSC pin of the 8224:

DS1Z_QuickPrint6.jpg


Output from the OSC pin compared with the associated crystal:

DS1Z_QuickPrint5.jpg


OSC and VCC (5v):

DS1Z_QuickPrint7.jpg


OSV and VDD (12v):

DS1Z_QuickPrint8.jpg


Rise time seems very slow compared to fall time, but is all within spec (max 20ns on both). To my untrained eyes though the pulse doesn't seem as "punchy" as I would expect it to? It also goes max up to 4V (I would assume it should be 5V?).


Data sheet is here: http://pdf.datasheetcatalog.com/datasheets/228/500533_DS.pdf
 
Last edited:
Ok, I realized I didn't measure the correct pin. So here's Φ2 on the 8224 which is TTL clock out:

DS1Z_QuickPrint2.jpgDS1Z_QuickPrint3.jpg

Looks ok, but is rise time a bit long here?


I also measured CLK OUT from the 8085 CPU. It seems to drift a bit, but I have no idea if that really matters or not:

DS1Z_QuickPrint9.jpg
 
Ok, so I decided to be a bit goofy and replace all electrolytic capacitors and observe the behaviour some more.

The double keypresses are gone, but symptoms now seem to be narrowing down:

  • Characters aren't correctly shown, as usual
  • Word processing software loads, I can type - but whenever the cursor row changes (not column), the system hangs

Since I can't see what I type (and something seems to happen in the lower portion of the screen when I type), I'm just guessing that changing the row will lead to an invalid value causing the software to hang?



Here's a full size picture of the display board. The top connectors go to the monitor interface:

2017-04-02%2018.12.44.jpg


Please pour your intellect over me :)
 
Last edited:
So I noticed the 6x1024 bit SRAM in the lower right corner of the video board (AM9101DPC) - at first I thought "aha!", but it doesn't seem like enough memory to hold a screen of characters. Not to mention the fact that the machine has an "all page" mode which zooms out, showing everything on the page. Anyone have any idea what this could be used for?
 
Given the location of the SRAM, I'd venture that it was for a "soft" font. I note that the adjacent character generator PROM appears to be a 2732 (EPROM), which is a 4Kx8 device, so not too unlikely.

So the video RAM must be shared with the CPU, which seems to be the case. Note the "RAx" pads for pullup resistor arrays near the lower left side. I'm guessing 8 bits data, + 16 bits address on those. Maybe that's a place to start looking.

Interesting in that there's 128KB of memory on the CPU board.

I note that Bill Degman has or had one of these things. I believe that he's still a member here.
 
Last edited:
Thanks! Maybe I'll dare to ping him :)

Do you think the SRAM buffers the contents in the EPROM, or could it be that the system actually supports more than one font? Given the memory size, it doesn't seem big enough to contain the entire character set of the machine (I counted on the keyboard + other characters I know exist, and that's at least 100, plus underline)?

I've catalogued all of the ICs on the board and it looks like there's a 4x4 SRAM chip there as well.Plus a ton of TTL logic, flipflops and whatnot. Getting late here, I'm tired and should go to bed... Here's a link to the component list - the position noted in the spreadsheet correlates to the number above the ICs as shown in the picture of the video board: https://docs.google.com/spreadsheets/d/14zpftSsPZLwm763qIjAuNkl5hm4iLLTripwCkDqtt6A

I'll check the thread first time tomorrow :innocent1: crossing my fingers I don't end up trying to debug a bunch of logic gates :eek:
 
Well, the good news is that there are only a few counters on the board. So that's a place to start. Clearly, you'll have a counter that advances for each scan line in a row and another counter that advances for each character--usually, these are combined into a single long counter where the scan line is advanced at the end of a row.

I suspect that the 9101s might hold an alternate character set--but that's just a guess.

If I've got time this evening, I'll dig around in bitsavers to see if I can find a terminal schematic that shows the general operation. After about 1976, most terminals used LSI controllers, so you don't have a bucketload of SSI TTL doing things.
 
So I probed the 2 counters (IC 105 and 106, SN74LS191N, 4bit counters). I had to solder wires to the pins of the ICs and hang them outside of the case, since the spacing between the boards are too tight to fit any probes (harsh, but is there any other smart method?)

They seem to count in intervals, followed by some long blank (not seen in images - didn't measure that one I'm afraid). Also, the speed of the clock cycles and how much they count up to differ whether they're in the "LOAD PROGRAM DISK" static screen or in the word processing program. They're in the lower right board and seem to be connected to the SRAM somehow. Anywho..

I'm not knowledgeable to know whether it is ok or not, but it seems like the last count in each interval is cut short. You'll also see more clearly on the zoomed-in version that the last clock pulse also is cut short.



CLK is top row, followed by the bits, from lowest to highest (I misplaced CLK in first image).

105 seems to count 0-5 in "insert disk" mode:

105_insert_disk.png



.. and 0-6 when in-program (notice increased frequency):

105_in_program.png



106 in "insert disk" mode:

106_insert_disk.png



.. and in-program. (notice increased frequency):

106_in_program.png



Here's a zoomed in view of the end-of-cycle for the 106:

106_in_program_end_cycle_zoom.png



Could it be that it these counters actually count the columns on-screen? They seem to count independently (all images have 5µs per division, with the exception of the zoomed in image), which would suggest to me that these are used for different purposes?

Ok, so probably I'm just way off base here. I should perhaps poke at the 4x4 SRAM and, if that yields nothing, try to monitor the RA1/2/3 pads and see if I can make any sense of them. However I'm pretty tired now so probably overlooking something. Please shout out if you see anything obvious I might have missed..
 
Last edited:
I would seriously have been lost without you, Chuck(G) :) :)

Looked at the screenshots with rested eyes, and realized the 105 does indeed seem to work in tandem with the 106, the pulses are longer and seem to correlate with every 0-15 cycle on the 106.

Further evidence:

  • When in "insert disk" mode, the characters are wider, so they are obviously different display modes. Less characters per line, which makes sense given that the counters don't go as high
  • I tried shortening some of the pins, and voíla! Horizontal positioning of characters changes


Maybe I should wire them up together and look at the combined output. It should also be worth mentioning that the repeating strings seem to be 8 characters long.

However, if these count the horizontal plane - what about the vertical plane? Is this typically done the same way, with counters?

EDIT: I just looked at my parts list again and the counters were highlighted in green. That you, Chuck(G)? :)
 
I've found some counters that might be a bit wonky - but then it worsened. I suspect that me soldering wires to the ICs to test them in the cramped space wasn't a very good idea, as I might have overheated some of the ICs while soldering. Now the vsync is a bit of a goner as well, so I've ordered spare parts and will just replace all counters on the video board and make them socketed in the process. I wasn't very comfortable doing this "destructive" testing in the first place, and now even less so.

I wonder how they tested these originally when repairing, they must have had some breakout cable or an extension of the card slot. Not sure how I would acquire this now, or what it's even called?

This has been a pretty horrible day for my electronics projects - stuff stop working and I'm getting ESD on everything else in the house. I even got a shock by touching a sausage when preparing dinner -- no kidding! Better wait and hope for better weather tomorrow I guess. Thankfully I'm working on a grounded mat when doing electronics repair work, but who knows if I might have fried something as well without noticing..
 
You must log in or register to reply here.
Back
Top