I only tested it with this one. I will look around for a 5V one, but what would difference would a higher voltage CPU make when you look at the current boot issues?

HandOfFate wrote on 2020-02-28, 12:27:

I only tested it with this one. I will look around for a 5V one, but what would difference would a higher voltage CPU make when you look at the current boot issues?

Boot issues?
In your last posting you wrote, you have exactly the same behaviour with and without a CPU.
So trying another cpu is worth trying, and using a 5v cpu is a bit more straight forward (cpu gets the power more or less directly from the power supply)

Yeah, but on the other hand it's just a guess ...

Ah, you're right. But I did test the CPU in another board (with DX4 settings) and it worked well there.

That's why I called it a "boot issue" because I don't think the CPU is to blame, but I will try a 5V nevertheless. There are some 5V processors available here so I will try to get one of them for cheap.

A photo of that new POST card output (with all the LEDs visible) when the PC is turned on would help. And, since you have a scope, did you check the clock signals on the CPU and ISA slot? The CPU one should be there since ROM is being accessed but perhaps it's unstable. The ISA one could be missing and that will result in dead mobo.

I will make the photo tonight.

I didn't check for any signals besides the ones on the BIOS chip. I'll try to see if I can get anything from the CPU or ISA slots.

The photos of the POST card. And a video of the change of lights after turning the power on: https://imgur.com/nIvrK3p

I didn't have a chance to measure the signals yet.

The video is even better, you can see a nice reset delay on power on, and how it's applied on loss of PWR_GOOD signal but before the capacitors run dry as you turn it off. So at least that works properly. The next thing to check then would be the address lines on the ROM chip, which ones are toggling - perhaps there is a broken trace to either data or address line.

Do you mean checking the address lines from the datasheet (https://www.futurlec.com/Memory/27C010-150.shtml) for any pulses?

I had a random thought: since the clock LED isn't blinking, could it be that the crystal oscillator is dead?

There is one (I don't think there is another one though?) with markings that says it should be running at 14.31818 MHz, and from what I can find that frequency is used for the ISA (and PCI?) slots.

HandOfFate wrote on 2020-03-04, 14:21:

I had a random thought: since the clock LED isn't blinking, could it be that the crystal oscillator is dead?

It's pretty bold of you to assume your eyes would see the LED blink at 14MHz 😀

As for question above, yes, probe all the A* pins on the BIOS ROM with a scope. Use some sort of reset button hooked up to mobo pins, rather than powering the system on/off - way easier to catch only interesting transitions that way.

Deunan wrote on 2020-03-04, 17:52:

HandOfFate wrote on 2020-03-04, 14:21:

I had a random thought: since the clock LED isn't blinking, could it be that the crystal oscillator is dead?

It's pretty bold of you to assume your eyes would see the LED blink at 14MHz 😀

But isn't that what the clock LED on the POST card is for? Shouldn't you see some flickering, especially on video, or should it be a steady light like it is now?

Anything above 100Hz or so will not register as flicker to human eye. We can, in very specific circumstances (dark-adapted eye and a single short flash of light) detect shorter pulses than that, and I guess that's where this whole "144Hz" monitor hype comes from, but otherwise it'll be a steady glow. A camera can capture faster flicker but that goes only for specific setups (and usually high-speed ones) - a typical camera or a smartphone has automatic brightness and 50/60 Hz anti-flicker filter enabled by default.

You might be able to record flicker up to a few kHz with correct camera setting and/or if the frequency just happens to mostly match the scan speed. Above that, no way. As to what the LED is for, then, well nobody said the people who made these cards actually cared all that much. Preferably it should be a dual-color LED that changes the color depending on signal polarity. That way you'd expect to see a mix of both colors at once if the clock was present and close to 50% duty cycle. But hey, it's cheap. And you get what you pay for.

A POST-card is really only for the codes, the rest is just extra and sometimes useful (like the reset LED). For everything else there is the oscilloscope.

Well, this is surprising and/or embarrassing. Still no successful POST but I did make some progress.

This week I found someone selling a Am486 DX2 66MHz (a 5V CPU) close by for a soft price and I just installed it. While I was re-arranging the CPU pin jumpers I noticed a glint of light. It appears that the metal of one of the jumpers was pulled out from it presumably when I adjusted the jumpers for the 5x86 CPU. And what do you know, something is happening now that I removed it.

The POST card now shows that the process stops when it's doing step "07" which, according to the manual, means that the CMOS is being validated and if the battery is bad. (the manual says that the previous code, "06", is "reserved" so who knows how accurate it is). I think this points to the Dallas battery (the code stays the same when I pull the Dallas out, but that doesn't have to mean anything)

I did replace the Dallas battery but there is no guarantee that the replacements that I bought are any good. It's also possible that I made a faulty joint when soldering on a socket to put them in. I will measure those legs first, and then probably order a modern, guaranteed firsthand battery, or go for the Dremel route and mutilate one of the Dallases to operate on a coin cell battery.

New video

(P.S. I didn't put in the 5x86 back in again because I'm not worried that it was the cause since I know the CPU itself works. And this would explain why the POST card behaved the same before when I booted the system without the CPU.)

Edit: As far as I can tell, all the pins on the Dallas socket are making contact. I exposed the internal contacts of one of my Dallases and it measured at 3.3V. I was expecting max. 3V, but I guess it's okay

Edit II: The Vcc to the Dallas is a clean 5V. I measured it because the datasheet says that the chip can't be read/written to if the voltage is below 4.5V

Sorry for bumping, but any ideas?

I will attempt to measure the address lines this weekend, but I have to come up with an easy, one handed way to reset the board. Would just shorting two wires work, or does there need to be some resistance?

Mobo should have areset button header, it's two pins - the input and GND. Shorting those two will cause the chipset to reset the mobo. Any button will do, or even a switch as you can just flip it both ways. Shorting two wires together is an option as well but not as easy since you have to position them right and make sure they don't touch anything else.

HandOfFate wrote on 2020-03-07, 01:03:

I did replace the Dallas battery but there is no guarantee that the replacements that I bought are any good. .....

Yes, that's likely. Did you order it in China for like 2 to 4 US$ per chip?
I am asking, because ....
I tried it, got 3 chips in total, all 3 had an empty battery inside.
So when you think, the reason is the empty battery, then do the "mod job"

CoffeeOne wrote on 2020-03-12, 20:03:

Yes, that's likely. Did you order it in China for like 2 to 4 US$ per chip? I am asking, because .... I tried it, got 3 chips in […]
Show full quote

HandOfFate wrote on 2020-03-07, 01:03:

I did replace the Dallas battery but there is no guarantee that the replacements that I bought are any good. .....

Yes, that's likely. Did you order it in China for like 2 to 4 US$ per chip?
I am asking, because ....
I tried it, got 3 chips in total, all 3 had an empty battery inside.
So when you think, the reason is the empty battery, then do the "mod job"

They were that cheap, yes. But I opened one up already and it measured at 3.3V. Maybe it's the only one (of the five) that works well though, who knows 😜

*Bump* I finally found someone with a chip programmer and some 486 motherboards. He sent me some BIOSes of similar boards (486 with PCI and ISA slots), 2 Awards and 2 AMIs.

The Award ones stop at the same BIOS code ('07', that is zero-seven) on the POST debug card as my original BIOS, so no progress there.

Both AMI BIOS chips that I show different codes if RAM is missing, but when it's present it stops at code 'C7' and produces a single, long beep. I'm not 100% sure what that beep means because most pages only describe short beeps, and long beeps followed by short beeps. One guide says that a single beep (no mention of long or short) means "The memory refresh circuitry has failed"

Step 'C7' is supposed to be "ROM BIOS checksum passed. CMOS shutdown register test to be done next". I'm not sure what to make of that in terms of where things could be going wrong.