Confusing Micro 8088 issues

[Steve Clarke]

I have a Micro 8088 machine that's experiencing some weird behaviour and I'm out of ideas of how to proceed. There are a couple issues, with the the most obvious symptom is that the serial mouse doesn't work reliably in DOS text mode, and not at all in windows or DOS games (however it has worked perfectly previously). I tried the obvious things of different driver, different mouse, different serial card none of which seem to restore functionality.

The plot thickens however as when running a system board check in CheckIt v3 my DMA controller fails (and that's the only component that does). I'm not entirely sure if this is connected to the mouse issue but I think I can independently corroborate this failure by switching my ESS audiodrive to use DMA 1 instead of 3 and the machine hangs on playing digital sounds.

My simple caveman brain concluded that there must be an issue with the Faraday chip as this contains the DMA controller. However, when I sourced a replacement and swapped it out the issues persist. Everything else in the system works perfectly and various tests and benchmarks bare this out.

I have my ISA-8 backplane stacked with a Sound Card, Network Card, Lo-Tek EMS card, Cirrus Logic VGA card, CF-Lite and Serial/Floppy card. On the off chance one of these is causing the issue I have removed everything except the Micro 8088, the VGA card and the CF-Lite but the issue persists.

The only other thoughts I have is that maybe U8 or U9 on the Micro 8088 might be faulty as they are the only other ICs on the address lines? But I'm well out of my depth now.

Can anyone offer any advice?

Cheers,

S.

[Sergey Kiselev]

The common way to troubleshoot such issues is by either elimination of controllers or by starting with the minimal configuration, verifying that it works, and then adding more cards/controllers.

Regarding the serial mouse issue:

- Make sure that nothing else conflicts / shares the interrupt with your serial port. Potential offenders are the network card and the ESS AudioDrive. Try changing the I/O port number and IRQ.

- Test the serial port using a loopback adapter

- Potentially the UART can be faulty. That happened to me once...

Regarding the DMA issue:

- CheckIt v3 will show a failure on DMA channel 0. This is because Micro 8088 does not use DMA for memory refresh. Other channels should pass. Is that what you're seeing?

- It is weird that switching ESS AudioDrive from DMA channel 1 to channel 3 triggers an issue. Unless you have another controller using DMA channel 1. The only difference in the signaling would be using DRQ1 an DACK1 vs. DRQ3 and DACK3. DACK signals are buffered on Micro 8088. But it is not very likely that the buffer IC is bad...

- Do you actually run the ESS configuration utility and switch the DMA channel, or just switch it in the games or whatever sound playing software you're using... Using a wrong DMA channel (different from what the hardware is configured for) would lead to system hangs...

It also could be a contacts/connectors issue.

- Try cleaning ISA card edge connectors with alcohol using a q-tip

- Do all the boards have gold plated card edge connectors? Tinned card edge connectors are not reliable, and can introduce all kinds of gremlins. Unfortunately people make and sell PCBs with them, because they are cheaper than ENIG (the second best option), and both are significantly cheaper that hard gold (the best option).
- Another thing which might cause reliability issues is the type of the logic used for the ISA buffers. The recommended logic family is 74F. It is faster and has larger output currents than other TTL families. It is also less likely to produce overshoots that fast CMOS logic (e.g. 74ACT) could produce without proper termination. 74ALS is also acceptable logic family. 74LS and 74AHCT potentially can work or cause reliability issues... Do not use non-TTL compatible CMOS logic, such as 74HC, 74AC, 74AHC...

Best regards,
Sergey

[Steve Clarke]

Hi Sergey, thanks for your reply.

I've been through the various cleanings, cards, resources and slots juggles with every combination resulting in the issues shown. The DMA issue is very odd indeed, and yes, I am setting the card up using the ESS configuration utility. Leaving all the settings the same except moving the DMA to 1 instead of 3 causes the issues. I've checked and double checked the various IRQs, Base Addresses and DMA settings of all the cards and there definitely is no conflicts. I can also confirm that all of ISA buffer logic are 74Fs. 

However, you do make one suggestion that sounds like it could be worth pursuing. All of my cards have gold connectors on the PCBs except for the Micro 8088 itself. It was the first one I built and it never even occurred to me to check... until now. I've ordered another PCB and I'll transfer all the ICs to that and see if it solves the issues.

Fingers crossed,

S.

[Steve Clarke]

Hi Sergey,

I have rebuilt the micro 8088 with a new PCB with gold contacts. All the components are new with the exceptions of the ICs. This has resolved the issues with the DMA and I can use channel 1 for the ESS Audio drive without issues and CheckIt now reports that only DMA channel 0 is failing. However I'm still having issues with the serial port mouse. I have tried two different mice, two different serial cards in many different slot, IRQ and COM: port configurations and still get no joy. To be clear, I have removed all cards except for the Micro 8088, CF-lite, the VGA card and whichever serial card I'm trying with. Both CTMouse and MS Mouse 8.2 find whichever mouse is attached at the configured com: port and I get some intermittent functionality in DOS (edit is my go to test) but nothing in games (Secret of Monkey Island) or Windows 3.0.

If I run the serial port tests in CheckIt I sometimes get a register failure with either serial card but the baud rate check always fails. One of the cards is your serial and floppy controller (the floppy works perfectly regardless) which I have tested with a Max Linear ST16C550IJ44-F and a TI TL16C550CFNR , the other is a new old stock ISA 40pin 16C550 ASYNC adapter card. Unfortunately I don't have any other ISA equipped motherboards to test the cards but it feels unlikely (but not impossible) for both cards to fail the same way.

I'm confident that the Faraday chip, v20, 8087 and ROM/RAM chips are all good and numerous checks and benchmarks support this. Once again I'm back to considering the ISA bus logic ICs, does this sound even plausible that they could fail in such a way? I have a TL866-II programmer which I believe does have some limited checking of 74 series logic... do you think there is any mileage in pursuing that?

Or is there anything else you can think of?

Thanks,

S.

[Steve Clarke]

Just an update and I think I might be on to a solution. I tested the 74 logic and everything checked out fine. I had a bit of a revelation that the only remaining constants I now had in this issue were the VGA card and the CF-Lite card. I removed the CF-Lite card and booted from a floppy and the mouse seems to work perfectly.  I then re-inserted the CF-Lite card minus the compact flash and booted from the floppy again and the mouse works perfectly...

I have some other CF cards I can try but it wont be until next week now.

[Steve Clarke]

Well that was a dead end. The CF was not the issue. I can reproduce the issue on demand whether I'm booting the machine from a floppy with no XTCF-Lite installed or booting from seemingly any compact flash. . However all this testing has highlighted a much quicker and repeatable testing mechanism. If I start dosshell.exe from whatever drive I boot from the mouse will not work. However, if I change the display mode to any of the text modes, the mouse suddenly springs into life (The mode switch is causing some sort of reset I assume?). If I then change back to the original text mode the mouse works. The mouse doesn't work in any of the graphics modes, but if I switch back to a text mode it starts working again. I get the same result using cutemouse 2.1b4 or MS mouse 8.2.

I get the same result with all the other ISA cards installed or just the VGA, Micro 8088 and Floppy+Serial card. I have tested with 3 different VGA cards (Cirrus Logic GD5401, a GD5402 and a Trident TVGA9000i), two Micro 8088 boards (with the chips moved between them) and two different serial boards. The only other things I haven't changed are the back plane and the PSU. In desperation I have removed the 8087 but other than the machine running cooler it has had no impact.

Just to provide the complete picture I have an NEC v20 with 2 x 512kb memory chips installed. I'm running BIOS 1.0.0. All the testing is run under MS DOS 6.22.

Any advice would be greatly appreciated.

[Steve Clarke]

Solved it. Finally. 

It turns out the floppy and serial card had a mechanical issue with a faulty PLCC44 socket. I was checking every solder joint in the entire machine again (electronics is fun...) when I noticed one of the corners of the UART chip looked ever so slightly raised. I attempted to re-seat the chip but that one corner always seemed to "pop up". However, I noticed that if I press the chip it stays in place and seems to make good contact. I taped the chip down and tried it in the machine and the mouse worked perfectly. Checkit also completes all the serial port checks. I removed the tape and reran the checkit test and the register fault and bit rate fault returned along with the odd text mode only function of the mouse. 

Upon closer inspection the faulty socket looked fine with no visible defects, but both of my UART chips wouldn't sit flush even when the socket is removed from the card. I luckily have a spare socket from an RC2014 UART project, so I swapped it out. The chip sits flush in the new socket and everything seems to be working perfectly.

That was incredibly frustrating, but at least I have almost enough parts to build a whole new machine... that should keep me busy.

[Sergey Kiselev]

I am glad you've figured it out! And I apppreciate all the updates.

Have fun with your Micro 8088!

- Sergey