charly2k18 wrote:

Deunan, the BIOS replace bin didn't help again 🙁. Here are the results of the test:

It wasn't supposed to help as in boot the system. Just to let me figure out things. The results are a bit unconclusive though.

What I gave you is a "BIOS" filled with HLT instruction so it should just stop the CPU, but there is some activity still on the /CE and some address lines. Though perhaps this is just a side effect of the incomplete decoding during memory refresh cycles which the mobo migt be doing on it's own as long as the clock is running. That's because the /OE line (pin 22) is now not toggling per your test so the BIOS is not being actually read.

Just to be sure, re-test and make sure there is no green LED activity on pin 22 with this image. And BTW that ISA test you did - was that on original BIOS or mine?

I have a few ideas but the noise on the address lines might be a problem. Assuming the CPU did stop and HLT code is read properly (that doesn't mean the data lines are OK, just that this particular 1-byte code is not glitching) we could check the address lines by doing up to 16 (since 2^16=65536 bytes) tests. For now though let's try a bit more optimistic approach again.
As before, program the file I've attached as BIOS image and run it. Probe the pins, but this time you don't have to do all of them if you don't have the patience. Only 1-10, 20-27.

Thanks Deunan!

I have re-tested pin 22 with the previous ROM image (BOOT ROM #2) and the result is the same, only red, no green led activity , and the ISA test I did, was using your bin file BOOT ROM #1.

Here are the results of the test on pins you mentioned using your bin file BOOT ROM #3. The tests were carried out in a dark room without light in order to better distinguish the LED colors 😀

That is confusing. The result is the same but image #3 had no HLT instructions in it...
Maybe something different then, try this one and now install the test card. See if shows any codes.
Whatever the result, probe the same pins again and also probe ISA slot B13 and B14.

Please upload your own BIOS image as well, since so far that was the only thing that showed any activity.

With BOOT ROM #4, the results are the same as BOOT ROM #3 😒. Results attached. And the results of testing B13 and B14 on ISA slot are both faint red color, no green activity and no boot codes with the test card, just "-- --". This is very strange!!

My assumption was the initial address is actually 0xffff0 but apparently that's not the case, and/or the bytes read are not correct. So now you'll try something else. Rather than prepare all the test images and upload them here I'll just tell you what to change so that you can streamline it a bit. You'll need a hex editor.

Start with your own BIOS image. Program it into the replacement chip, install, boot the system. Probe pin 22 and make sure you see green LED activity. If not, retry with original BIOS chip. This is to test if the replacement chip is not causing any issues - so that you won't waste time with all the work that comes next.

Now the actual test:

let x = BIOS size/2 = 32768
loop:
start with clean copy of your BIOS image
overwrite the first x bytes of the image with 0xf4 value
program, install, boot, check if you get green LED activity on pin 22 - if not, end the test, report back
if the activity is there, let x=x + (x/2)
jump loop

(so the x would go up like this: 32768, 49152, 57344, ...)

This is to test if we can even use this method to figure out where the BIOS execution actually starts. By the way, I've noticed you are consistently getting pin 7 low, this is address line A3 and I wonder what is causing that when A2 is high. Check (on powered-down mobo) the resistance between this pin and ground. See if you can trace what it is connected to.

Sorry Deunan, I'm trying to understand what do you mean I do with the hex editor. I'm not at home now so I can't test it but I would like to prepare all I can for test-it later. I attached an screenshot in this post in order you see what I have. Thanks again!

charly2k18 wrote:

Sorry Deunan, I'm trying to understand what do you mean

You know what, I can write a program that will generate test files for you. For now just test the chip with original BIOS image, I will send you a private message later today.

Deunan wrote:

charly2k18 wrote:

Sorry Deunan, I'm trying to understand what do you mean

You know what, I can write a program that will generate test files for you. For now just test the chip with original BIOS image, I will send you a private message later today.

Thanks for your time Deunan! Much appreciated!! 😀

Ok, I rechecked pins with original BIOS chip and the clone one W27C512 and the results of the tests are the same as before with previous ROM files. No green LED color on pin 22. Also, I tried to trace pin 7 and I found various continuity points but any resistor or capacitor between points, is it possible or I am missing something?! Attached are the results.

Just for your information, R100 and R14 resistors, have correct values. I have to disolder one leg of them and measured with the multimeter. The others of the image, checked on board, have correct values too.

UPDATE: I have uploaded a video to Youtube testing pins with original BIOS chip. The pins tested are 1-10 and 20-27. There's a little pause between tested pins and a long pause between 10 and 20. Hope it helps!

https://youtu.be/gahgQxxJfdk

Photos confuse me.
I see no xtal osc in the socket, but instead some pins as black dots.

By the way, you mentioned that sometimes the LED flashing pattern is different.
I had mobos with defective reset circuit (cap too low capacity -> too short reset time, CPU starting before all hardware was ready, resulting in random dead looping instead of clean system initialization).
Short the reset key post for two, three seconds and test to see whether this still results in different flashing patterns. Repeat this a few times.

We did try reset button, didn't help, and the signal does registerer properly on the test card. The missing parts and black dots are because of tracing pin 7 I assume.

So, ROM pin 7 goes to pin 12 of that '245 nearby, and to the Forex chip. And from YT movie there was activity on it with the original BIOS chip. But the ROM pin 22 is also not active this time around - so I wonder how is that possible. I assume you always verify the data written to W27C512? You can double-check this by dumping the chip now and comparing the resulting file with the original BIOS image. There should be a 100% match.

So it might actually be that the CPU is not starting in ROM at all but somewhere in RAM (due to address lines not working correctly) and it randomly hangs or not depending on what sequence of bits it finds and decodes as instructions. However the problem with this is I'm not sure I can help you any further if I can't rely on the ROM to test. Normally in situations like this you connect a logic analyzer to the mobo to figure out what the first few bus cycles after reset actually look like.

So the problem would be either the connection between CPU and chipset, or perhaps between chipset and ROM socket. The second one is easier to test so you can try - each address line (take a look at 27C512 pinout) should be connected to the big Forex chip. Doesn't matter which pin exactly, just that there is connection. There should be since pin 7 (A3) is connected directly so other ones would be as well. Not sure about data pins, you can try and find out too.

You only care about good connections, so use the beeper function. Or you prefer to measure resistance - it should be around 1 ohm or less for signals. Anything in kiloohms or higher you can ignore.

Deunan wrote:

The missing parts and black dots are because of tracing pin 7 I assume.

Yes, you are correct, missing parts are because of the trancing pin, I forgot to put the chips again on the mobo before taking the photo.

Talking about the dump images of BIOS/W27C512, I double checked that are 100% match and it is, original and copy. If I had another "clone" BIOS dump for this motherboard, we could check it too but I didn't found any other on Internet 😒.

Regarding continuity/voltage test, attached are the results. In voltage mode, apparently all is OK i pressume. Anyway, this motherboard must have several problems around all the circuit. Visually are all correct, no bursted chip, resistors, caps..ah, and let's not forget the strange noises coming from the speaker.. they are still there! very weird!

FYI: I also measured the connections between SN74lS245N and the Forex chips (all of two), the Morse chip and the CPU, just in case. Results attached below.

All of these look OK. There are no weird values and everything seems to be nicely grouped. The '245 is a bus transciver and it separates the Forex chip signals from (probably) ISA bus slots. You can check this by testing pins 2-9 with ISA slot address signals. If not ISA then perhaps the SIMM memory slots. One of these two.

The important thing is all pins on the ROM socket are directly connected to the Forex chip - so it should be able to properly read the BIOS code. But it's not doing that so that leaves us with 2 possibilities: the chip is faulty, or the CPU signals are not getting to it correctly. The second one might be a mobo problem or a faulty CPU.

So, since there is nothing else to do anyway, why don't you try investigating the connection between CPU and Forex chip? Download this PDF: http://datasheets.chipdb.org/Intel/x86/386/da … ts/24126703.PDF
On page 3 you'll find pinout for the PQFP 386DX. You'll need a good, sharp probe tip - you can use a needle for that. Just as before, check each signal and it doesn't matter much which pin on the Forex chip it ends up at, just that there is a good connection. Some pins on 386 might not be connected to other chips (like the FLT#) so it's not always a bad thing.

EDIT: Oh, and you can probe some of the CPU pins as well while at it. Or you can even start with that. The CLK2 pin (24) should give you both LED colors for example. ADS# (pin 27) as well. If not, don't bother with futher checks, report back.

I've had a case whereby pressure from the probe made a joint look connected, but in fact it was the probe pressure itself that was making two points in contact. It was not at all obvious and I missed even after a few rounds.

Guys, sadly I'm going to have to temporarily park this project. It is consuming me a lot of time which unfortunately I do not have at present 🙁 But in any case, thank you very much for your assistance, for your help, for sharing your knowledge and above all for your patience 😉. Maybe in the future I will return to this project, who knows! Many many thanks again guys!