Ok, does it do the same with math-co removed and jumpered no-math ? Odd you do not get any good readings off the Diag card but not all early boards posted to port 80 iirc....and sometimes you needed to check a few diff slots.
Also do not rely on an ATX psu on an old 286/386 board, and do not use a cr2032 coin on one = my rules because they just have issues working proper in many situations or attempt to recharge it (bad thing).
Glad you got the board to post even if it takes a few tries. Probably have some leaky tantalum caps causing issues...

I would start with replacing the cmos battery. I've had many many boards over the years either play dead or act really strangely with a dead,low, or missing bios battery. Once the system posts can you get into your OS?

Cmon guys! hpxca clearly stated he used not-a-varta and its even visible in the picture 😀

Problem starting can be due to bad Reset generating circuit, early on it was something super basic like capacitor to +5V.
Does it start from Glitched state after shorting Reset pins?

@Horun @chrismeyer6
I've thought about replacing all the tantalums on the board but I have not done it. I may yet give that a shot - it's fairly cheap to try and easy to do and there are not that many. None of them look visibly damaged or burnt and in my experience bad tantalums usually either short, explode or darken/burn.

As for the battery, it is a not-a-varta and it has the OR current switch on the bottom. That IC acts like a diode with a very low forward bias voltage and a high enough reverse breakdown voltage that there's no risk to the battery from the boards charge current but the output voltage is still over 3V (with a good battery). With the CR2032 installed in the not-a-varta the CMOS saves work as expected.

@rasz_pl interesting thought, I actually never made myself a reset button with a few leads attached but it might be worth doing and looking at the reset circuit.

Aside from the first boot attempt it behaves so much like it can't find anything in the ISA bus and the way the video card misbehaves has made me think about replacing the some of the 74LS244 line drivers. Actually one of them is an Hitachi 74LS244 and the others are national 74F244, I don't know if there is a reason they used the two different parts but i'm a little worried there is. Some are between the ISA slots and difficult to get to as well.

Also worth noting is that this board is actually rated at 33/40 MHz bus speeds and I am running it at 25MHz since at 25MHz CPU is all I have. I don't think that should be a problem though but if someone knows for sure otherwise please let me know!

I'm actually starting to suspect the 82C206 on this board. It seems to have everything in common with the unusual behavior:

-It handles the power good signal from the power supply, which could explain the strange behavior and difference between power supplies. When the power good is low, all address, data,data strobe and R/W pins are disconnected from the processor. This could explain the boot failures and no POST codes on the initial boot. Even though the power good is not low, all bets are off if the chip is faulty.
-It handles DMA and IRQ directly related to the ISA bus which could explain the strange boot behavior and video card issues on the 2nd boot attempt where it acts like can't find a video card or the video card puts out the purple screen.
-It handles the 32kHz square wave generated by the RTC oscillator and contains the actual RTC implementation, which could explain the odd RTC "ticks" in the CMOS setup after it finally boots.

Its going to be a bitch to get off of there without damaging those ISA slots though, and I don't have an vaccum de-soldering station 🙁

how about you try resetting it first? or putting scope on reset line in single shot mode.

>Aside from the first boot attempt it behaves so much like it can't find anything in the ISA bus

yes, any card can do this without seeing reset signal

Yes, i'll give that a try before I try to remove the 82C206. Rather then make one I just ordered a https://www.amazon.ca/gp/product/B09W3XFSR2 off amazon to make that a little simpler. I'll also take a good look at the pins and traces around the 82C206.

So I checked the ISA reset line (pin 2 of the ISA slot on the bottom row), it looks normal. It stays low the entire time after the system is powered on, going high only for a millisecond or so as the system powers down. This was the case on attempts when the system both did and did not POST including attempts when it gave the "no video card beep code" when there was one installed. When the reset button is pressed the reset line pulses high for about 100ms before dropping back low again and the system resets as it should. The problem is the same whether I reset it via the button or by power cycling the entire system.

You said it did work after power cycling, so does it start working after manual reset? or do you need keep resetting to get it to work? or it doesnt help at all?

Nice board really, hope it will recover.

@raz_pl It's the same cycle whether I use the reset switch or the power switch, meaning I need to either power cycle it 3-4 times or reset it 3-4 times (or any combination there of) to get it to POST.

After messing with it for more hours tonight I figured in for a penny in for a pound. An eBay seller had new old stock of that UMC82C206F for a fairly reasonable price so I bought one. I removed the old one (thank the can of perrier I had just drank and some kapton tape for the heat shield). I think the ISA slot came out of the deal intact. Will be several weeks until I get the replacement though.

Really hope this is it, I think we're beyond economical at this point anyway.

So I haven't given up on this thing, even though I probably should have. I replaced the 82C206 and it made no difference at all (sigh). I also replaced every tantalum cap on the board per an earlier suggestion - nothing.

I've stopped looking so much at the RTC circuit, it seems to be working fine and I think the glitch with the seconds in the CMOS setup is just something it does, I see it on some of my 486 boards as well that otherwise work properly.

I decided that since it behaves very slightly differently with different power supplies (the ATX with adapter vs the AT) I would try to replace ALL of the power filtering caps, including the ceramic disc capacitors. Many of the ceramic disc caps seemed excessively fragile, easily getting chipped or gouged. So I replaced them all and that actually helped - but it still does not boot all the time.

Now it behaves like this:

1st boot - POST diagnostic card shows codes, but the codes appear very slowly almost like the CPU is grossly underclocked. After about 30 seconds, the codes stop changing and the computer does nothing more until rebooted or reset. Rarely I get the afore mentioned purple screen from the video card and a boot loop instead.

2nd boot - Almost always completes the POST and boots normally.

One trick I have found is that if you power the system ON while holding it in reset for about 30 seconds, then release reset it always POSTs and boots normally.

Still open to any other suggestions of course. There are no aluminum electrolytics on this board. Between the tantalums and the ceramic disc capacitors I have replaced nearly every cap on this board. All that remains capacitor-wise are the small ceramic axial caps, but I don't think any of those are used for any sort of power filtering.

There's no voltage regulation on this board either. The +5v from the PSU is fed directly to the IC's - CPU included, and to the ISA bus.

I've got a 33MHz 386 on order that i'm going to try, as this board is designed for a 33 or a 40MHz CPU and I have a 25 in there now (with a 50MHz oscillator), but I really don't think the slight underclock would cause these issues.

reset signal is not reaching one of the components, start with three big UMC chips and keyboard controller

rasz_pl wrote on 2024-05-23, 14:11:

reset signal is not reaching one of the components, start with three big UMC chips and keyboard controller

I'll take a look at this as soon as I get home. My scope probe is too fat to get onto these pins, so i'll probably have to solder a wire onto the pad to see it.

Reset seems to be pin 146 of the 482 chip (system controller) and pin 18 of the 82C206. I found the pinouts for all 3 on the retoweb

https://theretroweb.com/chip/documentation/um … 8b321761597.pdf
https://theretroweb.com/chip/documentation/um … 02349458608.pdf
https://theretroweb.com/chip/documentation/um … 69312425138.pdf

I can't seem to locate the reset line pin the 82C481 memory controller though.

resmpc pin 133?

could it be some bad IC that has a internal connection that when the chip gets warm will connect and when it gets cold it will fail ?
on that line I would warm up individual chips with a soldering iron or just something hot and try the board util I find the faulty chip...

I also had a board that used to stop at 06 and 07 post code and I found out it was the jumpers setting that my CPU didn't like ... it seemed well set but the CPU didnt like
but that was on a 486 motherboard and your board is 386 maybe it is simpler

good luck

Nunoalex wrote on 2024-05-23, 15:27:

could it be some bad IC that has a internal connection that when the chip gets warm will connect and when it gets cold it will f […]
Show full quote

could it be some bad IC that has a internal connection that when the chip gets warm will connect and when it gets cold it will fail ?
on that line I would warm up individual chips with a soldering iron or just something hot and try the board util I find the faulty chip...

I also had a board that used to stop at 06 and 07 post code and I found out it was the jumpers setting that my CPU didn't like ... it seemed well set but the CPU didn't like
but that was on a 486 motherboard and your board is 386 maybe it is simpler

good luck

Thanks for your suggestions Nunolex. I am hoping it's not the 82C481 or 82C482, there's new old-stock of these available on eBay but they are very expensive, it would be cheaper to get another board then to replace them. I've already swapped the 82C206 with no success and aside from some buffers/line drivers/inverters there's not a lot left on there anyway. I'm going to dig into the reset signal. It almost certainly goes through one or more of those inverters and maybe one of the line drivers as well.

On this board there's no CPU jumpers, just some to set the amount of cache, indicate if there's a 387 co-processor and one relating video (mono/cga/other). CPU clock speed is simply set by replacing the clock oscillator near the CPU. Just for kicks I've put a 2nd 14.318 MHz oscillator in there a couple times and, once it boots (same issues occur) it runs quite happily, but rather slowly at around 7Mhz.

Out of curiosity - did you check if the problem is related to temperature? For example lets say it takes 5 resets to get this mobo to work. Does this number change if you press the reset button 5 times in a row right after power up, or if you wait a minute or two between each reset?
Did you try replacing the BIOS ROM with something else? If you have EPROM programmer maybe consider buying another EPROM chip and make a copy (just make sure the read is stable). You can even get a modern 70ns (or faster) OTP EPROM, these are so cheap you can just "waste" one with the 1:1 copy.
Remove and reinsert the CPU. I've never had issues with these sockets but it's worth trying. Try removing the cache chips, see if it changes anything (if it completly stops the mobo from working you'll know the cache is always active and therefore any issues with these chips can affect the boot process).

It's fixed! @rasz_pl you were absolutely right, it was the reset line.

I was tracing the reset line last night and I started drawing the circuit to try to reverse engineer it. I started at the reset switch header and worked back from there, but the line was very difficult to trace because it ran all the way from the lower right corner of the board where the header is located and up to the upper left corner near the battery. And that last word should ring all the alarm bells - yes near the battery. I never would have seen it if I had not been tracing the line with my multi-meter, but sure enough there it was. I made a very short video of it and you can probably see why it was so hard to find

https://www.sejr.ca/erik/pages/386/board.webm

Or for a closer look:

https://www.sejr.ca/erik/pages/386/microscope.webm

In hindsight I should have inspected the battery are much more thoroughly, it seems that the battery did leak at some point and the person I got it from cleaned it up but it was probably already too late for that resistor. The other ones above it are fine though.

One 470 ohm resistor later and the machine boots every time.