286 clone - corrupted video output and strange diagnostic result

[3pcedev]

Hi all,

I've got a restoration project on my hands, and I could use some advice on how to proceed....

Basically I have a 286 clone (Laser (Vtech) 286/3) which sat for many years. It worked when stored away, as I was the last to use it (circa 1999). It suffered the usual leaky battery, and also some mice damage, but overall it is far from the worst board I have restored.

The battery leakage confined itself to a very small area. All traces on the top/bottom layers are intact, and verified continuous. There was a DIL switch nearby which had some contacts which open circuited, however that has been fully repaired. As always, there may be damage to buried layers that I can't see though. The electrolyte did not make it's way into any IC's.

Mice managed to 'make their mark' on a couple of ASIC's. There is some corrosion around the legs, but again I have verified continuity to from the legs to the traces on the board. All is fine as far as I can test. From my experience the mice damage is usually worse than battery electrolyte, so I am leaning towards one of these ASIC's has an internal fault....

When I first booted the PC the display was garbled (lots of random characters everywhere) but there were some legible characters which eluded to 'PARITY ERROR' and something to do with a RAM fault.

The first thing I did was removed all non-essential expansion cards, and switched in a known good video card. This made no difference. I tried MDA, CGA and VGA, also tried different slots on the motherboard, all had the same result.

I then figured I would put in the Supersoft diagnostic ROM's to see if they could narrow down the RAM fault (CGA card installed). The result here was the video was still heavily corrupted, but you could make out the 'basics' of the diagnostic screen. The weird part is it seemed to go through and pass all the tests... I listened to all the beep codes that it output, and all I got was:

- 'Cannot Initialize Monitor', which is fairly common even in a working clone system.
- 'CMOS RAM', which indicates my BIOS settings don't match the system config (not surprising since I am running just a video card, no FDD/HDD controller etc).

There were no beep codes for bad RAM or anything else.

I am guessing something is messed up in the data or address bus; but if so why does the SuperSoft pass all the tests? Why doesn't the system just hang as the bus is used for all sorts of other things??

I finally swapped the RAM from bank 2 into bank 1 (in the blind hope something was bad in the first 64k), but it made no difference.

If anyone has any clue as to what to test next, I would be most grateful!

(Photo attached of SuperSoft diag screen, showing the video corruption).

 

[modem7]

I am guessing something is messed up in the data or address bus; but if so why does the SuperSoft pass all the tests? Why doesn't the system just hang as the bus is used for all sorts of other things??

Might this be something that affects only the ISA slots. For example, for uni-directional lines signals going to the ISA slots (e.g. address), maybe there are buffer chips between the ISA slots and other motherboard circuitry.

Have you tried a complete power down, i.e. battery removed for a period adequate for the motherboard ?

 

[acgs]

You don't mention if the monitor used to test the machine is from the same system or a known good monitor. Are you sure the monitor isn't causing the corruption? (Not that I think it would, but you seem to have isolated the graphics card itself from the equation.)

What about the video cable? Is it part of the monitor? If not, is it a known good cable?

 

[3pcedev]

Might this be something that affects only the ISA slots. For example, for uni-directional lines signals going to the ISA slots (e.g. address), maybe there are buffer chips between the ISA slots and other motherboard circuitry.

Have you tried a complete power down, i.e. battery removed for a period adequate for the motherboard ?

It would make sense it it only impacts the ISA slots - that would explain why the Supersoft diagnostic manages to run. The board uses a number of ASIC's, I don't know if they are used as buffers or as bus arbitrators though. Unfortunately I don't have any technical data on this motherboard. There are some basic 'glue logic' chips on the board which I can test using a logic analyzer though. Might be worth a try...

Also yes - it's been powered down for over a week with no change.

You don't mention if the monitor used to test the machine is from the same system or a known good monitor. Are you sure the monitor isn't causing the corruption? (Not that I think it would, but you seem to have isolated the graphics card itself from the equation.)

What about the video cable? Is it part of the monitor? If not, is it a known good cable?

The monitor and video cable are fine. I use them on a number of other systems routinely.

The characters displayed look perfect, albeit randomly placed all over the screen. To me this suggests that the video card has a fault (which I have eliminated by trying multiple different functional cards), or the data being fed to the video card is corrupt.

I'll have a play with the logic analyzer... who knows, maybe I will get lucky.

 

[acgs]

... Unfortunately I don't have any technical data on this motherboard. ...

Perhaps if you shared a picture of the motherboard, someone here can identify it. Then, with a little more luck, someone might be able to help out with a service manual, schematic, or something.

 

[3pcedev]

Perhaps if you shared a picture of the motherboard, someone here can identify it. Then, with a little more luck, someone might be able to help out with a service manual, schematic, or something.

Unfortunately I've been down this road before. It's a Vtech clone 286; who used their own in-house motherboards (not an off the shelf one used in many clones). I don't even know if they released a technical reference for it.

I have found a problem though! I was working through the glue logic on the board, and I came across a 74F244 which is not following its truth table.

I've attached the output from the logic analyser. Basically when 2OE goes LOW, the 2A* input should be output to the 2Y* output (i.e. when 2OE is low, if 2A1 was high, then 2Y1 should be high as well). If 2OE is high, the outputs should be high impedance. You can see that 2OE is low for most of the capture, so the inputs should equal the outputs at those points.

2A1 and 2Y1 do not track, likewise with 2A2 and 2Y2. 2A3 and 2Y3, and 2A4 and 2Y4 do seem to be working though.

My gut tells me this IC is actually working OK, but another IC on the bus is not going into high impedance when it should, and it's overpowering the output pins on the 244. I will keep testing other IC's on the board, and see if I can narrow it down further. If I find something that is mirroring it's input to its output, regardless if it should be high impedance, then that is likely the culprit.....

 

[3pcedev]

Well I found the problem, but I think I am out of luck....

I probed around some more on the board, and found multiple problems.

The first is that a bi-directional buffer (74F245) which directly interfaces with the ISA bus exhibits the same basic symptoms as the 74F244 above. The outputs are not always equal to the inputs when OE is low. I also noticed the OE never really went high on the logic analyser - it just stayed low 99% of the time, and then very rarely blipped high. By blipped; the blip was so small that it would not be registered as valid on a 12 MHz system.

After a bit more probing around, I actually found the OE on the 74F245 was shorted to ground (board completely switched off, tested with multimeter). This is obviously as issue, because the outputs are now permenately tied to the ISA bus.... Maybe if they were using the 74F245 as a pure buffer this would work, but as far as I can tell the OE line does actually go somewhere (i.e. a trace leads away from it, not just tied to ground).

A bit more basic probing around with the multimeter and I found that a small PROM on the board had 3 of its data lines also shorted to ground. This also makes zero sense. Since the PROM was socketed, I removed it and powered up the board. I then got a very different 'garbled' pattern on the screen, but the same overall behavior. Further tests on the glue logic on the board showed nothing remarkable.

I then probed around some of the chipset ASIC's. The board uses a TACT82000 chipset (info here). Unfortunately I could find no datasheets or technical references for the chipset, so I had to fly blind. What I found was several data lines (I think) on the 82301 which were also shorted to ground. I couldn't follow them very far due to the mess of traces and VIA's on the board, but it's obvious that they are not supposed to be grounded at all times.

So it looks like the 82301, and possibly the 82303 is likely dead. Both have some minor corrosion due to mice, so I expect something internal has failed. Unfortunately I can't seem to find any of these chips in quantities less that 100 (and at that quantity, the price is ridiculous). If anyone has any leads on them, please let me know!

I guess the best option moving forward (unless I can find replacement IC's) will be to find another AT clone motherboard. Unfortunately my collection has a huge amount of 386 onward spares, but no 286...