one more image, the IR scan of the board after turning on.
the CPU is getting hot, some other components as well. Not sure if helpful.

tomas

Hi,

so I am still stuck here but I have tried some different things. I have received a new CPU for the board since I was afraid that he old one was damaged (crack on the packaging). Replaced the CPU today but with the same "no beep, no boot, no code on a debug ISA card" symptoms - still with the BIOS I had there before. - the first unknown since the replacement CPU could be bad as well. Ok, quite disappointing but I hoped that maybe a different BIOS will do the different. Since I have got the TL866II plus I tried to reflash the BIOS using the image from here: https://theretroweb.com/motherboards/s/gemlig … 86umc#downloads. I have the GMB-386UMC v1.3, the BIOS says v1.02 so I guess it could work, or? Anyway, I do not have a UV EPROM eraser so I have ordered erased TMS27C512 - the original one was I think W27C512-45Z L27C512 - are these compatible by the way? - the second unknown. Anyway, again "no beep, no boot, no code on a debug ISA card" symptoms.

I have measured once more the voltage on the ISA rails and noticed that the negative values are -4V and -11V - is this normal?
Anyway, I have no idea where to look more to fix this, any suggestion?

thank you, Tomas

1.02 is the BIOS revision, not the board revision. I'm curious to know what was on the original bios if you could please dump it. There is apparently a rev 1.03 out there so maybe that's what you had ?

Deksor wrote on 2022-10-13, 16:07:

1.02 is the BIOS revision, not the board revision. I'm curious to know what was on the original bios if you could please dump it. There is apparently a rev 1.03 out there so maybe that's what you had ?

thx, yes, I though so but wasn't sure. The original BIOS was destroyed during extraction by a friend, few legs fell down, however, I have added it to another DIP socket and read it out. I have to say that I am not sure if this is the original one since it was laying around for some time. Inside the BIOS is however a string: 40-0102-428036-00101111-111192-UMC491, the BIOS on the website has as well DTK within the BIOS, this one no. Hence, I am not sure. If this is the original one, it looks like 1.02. Anyway, enclosed.

And a small update, I did today some cleaning of the board, vinegar, soap water and isopropyl alcohol. No difference, but, using the original BIOS that is now in another DIP socket and inserted within the DIP socket on the board I got some messages coming from the ISA debug card. All codes were referring to "normal" startup but just froze at one message. I am not sure now if the BIOS works or not or just partial since it is within the DIP socket to compensate for the broken legs but it gives me some hope. I am wondering now if the TMS27C512 I got as the BIOS replacement is ok for the board or not. Or it was just a coincidence. I do have some different 386 BIOS chips but I do not have the eraser and my attempts to erase it were so far not successfully.

So if no one has any idea, I will try to solve the BIOS chip issue and then test again. If this will be a no go, I will try reflow and then just use this as spare part board, sadly.

tomas

You could use an EEPROM (like the Winbond W27C512) instead so you can try different BIOS files easier and quicker without waiting for the erasure process to finish every time.

majestyk wrote on 2022-10-14, 19:40:

You could use an EEPROM (like the Winbond W27C512) instead so you can try different BIOS files easier and quicker without waiting for the erasure process to finish every time.

good idea, thank you, ordered already a few of them.
will report back if I will find something that is causing the malfunction.

tomas

Hi all,

so I have got the Winbond chips and flashed the BIOS there but I am getting more or less the same behaviour of the board. The board or better the debug card post from time to time something - hard to reproduce and once I have got even a clicking sound from the speaker. Hence, it seems that it is not completely dead but I can not figure out why it wont post, boot, etc... I am out of the options on what to try next but since the debug card is posting something from time to time there is something either random or changing over time.

Does a bad capacitor could cause this?
Does it make sense to re-flow the board for cold joints - here I am afraid that the plastic parts will melt down?
Any idea what to try next?

thx, tomas

not software, not cap
You have cracked joints/vias or bad sockets. You can try percussive maintenance - smack it repeatedly in different spots looking for change.
Bad vias are really hard to locate, sometimes even act of probing will temporarily reconnect them. poking vias/pins with a wooden/plastic stick while repeatedly resetting might find something.

rasz_pl wrote on 2022-10-25, 06:22:

not software, not cap
You have cracked joints/vias or bad sockets. You can try percussive maintenance - smack it repeatedly in different spots looking for change.
Bad vias are really hard to locate, sometimes even act of probing will temporarily reconnect them. poking vias/pins with a wooden/plastic stick while repeatedly resetting might find something.

thx for your answer. I will try what I can do here. I have inspected the board under microscope several times but couldnt find any indication of a bad trace.
I will try to focus on the cracked joints but I will give this a few more attempts and then leave it as it is, already spend so much time on this.

thx, tomas

Have you tried changing bus speeds? JP1 & 2

I have a hunch on socket was removed and qfp was soldered on to get 40 mhz cpu on it.
Blue resistor is used to deselect the pga socket. I could be wrong.

Is cpu getting vcc correctly?

Nexxen wrote on 2022-10-25, 11:21:

Have you tried changing bus speeds? JP1 & 2 […]
Show full quote

Have you tried changing bus speeds? JP1 & 2

I have a hunch on socket was removed and qfp was soldered on to get 40 mhz cpu on it.
Blue resistor is used to deselect the pga socket. I could be wrong.

Is cpu getting vcc correctly?

hi,

yes, I have tried all configuration that are available, no difference.
The "blue" component you are referring was a capacitor (104) but it was definitely changed by someone since I saw there some traces of this before. The image on TheRetroWeb https://theretroweb.com/motherboards/s/gemlight-gmb-386umc has something similar there as well but I am not sure if the value should be 104 - this is the value I had in the socket on my board. What is odd here is that both pins where this capacitor is coming are on the ground, I do not understand this at all - no idea but even after leaving this out, the positions on the board where this part is coming are on the ground. There are no traces visible coming to this part and thus it is connected to the middle layer of the board (not sure how many layers this board features). Maybe the value is wrong, or it should be something else?

Also, comparing the images on The RetroWeb I see that I am missing BC11, next to the 14MHz OSC - never was there at least from what I can see the traces are clean.

Yes, the VCC ont he cpu is okey.

tomas

Added the BC11 there, just guessed that it could be the 104, no difference. I have no idea what to do next apart of heating the board with an SMD station. Desperate and ready to throw it out of the window 🤣

tsalat wrote on 2022-10-25, 10:33:

I have inspected the board under microscope several times but couldnt find any indication of a bad trace.
I will try to focus on the cracked joints but I will give this a few more attempts and then leave it as it is, already spend so much time on this.

there is stuff you simply will never find by just looking at it, thats why Im advocating for more hands on approach. Slightly bending whole pcb, fat fingering every component, all while repeatedly resetting the board and looking at the post card. Bad vias, track cracks under slots/chips, hairline cracks in resistors. The fact its intermittent clearly suggests mechanical failure mode. Theoretical worst case scenario would be internal defect in one of the chips, sometimes bonding wires rip/break from vibration/heat, very hard to find but also extremely rare.

rasz_pl wrote on 2022-10-25, 22:40:

tsalat wrote on 2022-10-25, 10:33:

I have inspected the board under microscope several times but couldnt find any indication of a bad trace.
I will try to focus on the cracked joints but I will give this a few more attempts and then leave it as it is, already spend so much time on this.

there is stuff you simply will never find by just looking at it, thats why Im advocating for more hands on approach. Slightly bending whole pcb, fat fingering every component, all while repeatedly resetting the board and looking at the post card. Bad vias, track cracks under slots/chips, hairline cracks in resistors. The fact its intermittent clearly suggests mechanical failure mode. Theoretical worst case scenario would be internal defect in one of the chips, sometimes bonding wires rip/break from vibration/heat, very hard to find but also extremely rare.

Thank you for your answer. Well understood 😀
I have tried to generate some pressure on the board at different places and tries to restart and as well turn off/on the PSU. No luck, when I thought I found a spot, it was not repeatable. The codes I am getting on the debug card are more random, I do not see any pattern when they occur and when not. I mean, I start the board, nothing, restart a few timea and something. Other times I turn on the board and some code on the debug card, restarting or starting again will not do anything, etc...

I will try to desolder the bios socket and look there but although I think the board is not completely dead, I have no idea what to after. Well 😀

Thx, Tomas

check all address/data lines between bios / isa and CPU (some will go thru buffer chips 244/245), check clock, then im out of ideas apart from desoldering and testing/replacing all 74xx series chips, that would leave you with bad chipset

rasz_pl wrote on 2022-10-30, 06:09:

check all address/data lines between bios / isa and CPU (some will go thru buffer chips 244/245), check clock, then im out of ideas apart from desoldering and testing/replacing all 74xx series chips, that would leave you with bad chipset

Hi, I have checked today the bios traces and almost all from the CPU still need to trace a few. At this moment, all traces are having continuity. The bios sockets reads voltage from 0,8V to 5V on all pins except the GND. I will check the CLK2 pin 15 on the CPU when I will be at work (I found only the datasheet for am386 sx that states CLK2 on pin 15). All OSC I have already checked and they are working.

If the pin 15 is the wrong one to check the CLK of the CPU, let me know.

Funny thing, the debug card get me sometimes an error code even without BIOS, is this normal?

Thx, tomas

some kind of resistor network should be pulling all data and 0-23 address bits high, while 24-31 low. ISA IRQs high, ISA DRQs low. D/C R/W M/IO high
is your bios/isa bus behind 245 buffer? even without bios that working pullups should never allow for any random behavior.

rasz_pl wrote on 2022-10-31, 20:00:

some kind of resistor network should be pulling all data and 0-23 address bits high, while 24-31 low. ISA IRQs high, ISA DRQs low. D/C R/W M/IO high
is your bios/isa bus behind 245 buffer? even without bios that working pullups should never allow for any random behavior.

thank you again for your reply.
I have measured today the CLK with the oscilloscope and I am bit confused. I have tried the position on the PGA socket to measure the CLK and as well the 15-pin on the QFP. Both were showing something like 2MHz and the signal was not looking as I would expect. I have uploaded a picture of the board here: https://www.sylex.repository.3d-sphere.com/in … d4pSdMLtfFHM4jX, and highlighted the points as well.

Underneath the BIOS is an SN74LS244N BUFFER - I guess you were referring to this one.
I guess the CPU should have the 40MHz clock, right? or not?

tsalat wrote on 2022-11-02, 18:53:

thank you again for your reply. I have measured today the CLK with the oscilloscope and I am bit confused. I have tried the posi […]
Show full quote

rasz_pl wrote on 2022-10-31, 20:00:

some kind of resistor network should be pulling all data and 0-23 address bits high, while 24-31 low. ISA IRQs high, ISA DRQs low. D/C R/W M/IO high
is your bios/isa bus behind 245 buffer? even without bios that working pullups should never allow for any random behavior.

thank you again for your reply.
I have measured today the CLK with the oscilloscope and I am bit confused. I have tried the position on the PGA socket to measure the CLK and as well the 15-pin on the QFP. Both were showing something like 2MHz and the signal was not looking as I would expect. I have uploaded a picture of the board here: https://www.sylex.repository.3d-sphere.com/in … d4pSdMLtfFHM4jX, and highlighted the points as well.

Underneath the BIOS is an SN74LS244N BUFFER - I guess you were referring to this one.
I guess the CPU should have the 40MHz clock, right? or not?

The pin where you're measuring 2MHz seems to be A6 signal, not CLK signal.
I think the correct pin to measure should be from the left bottom corner of the 386 footprint on the photo, the third column, 6th row, counting from the bottom left corner.
QFP pin 15 should be correct CLK signal. please verify, on photo, you measure left side of qfp package, 15th pin from top.