In recent days I have been looking for info on the keyboard controller, in this case it is the LANCE CORP chip. LT38C41, from what I found there are no compatible alternatives in case of chip failure, you have to use an identical chip, I don't know if another chip could be compatible, but I only have a NEC D80C42C chip available.
Having fixed the two corroded and interrupted tracks, he restored the electrical connection with the DIN connector.
I have checked other connections with the controller chip and they are ok. Ditto the fuse and some SMDs (resistors) in the area, even the 4 diodes are ok.
The problem seems to be due to a bad connection (oxidized?), This is because when turned on the three leds of the keyboard turn on and off (regularly), so at this point the next thing to do is remove the controller chip, and clean the PIN of the chip and socket.
From what I have learned underneath that controller chip, there may be another chip, and in case you have to see if there is oxidation there.
But looking at the back of the socket you don't see any other pins, so I don't think there's another chip, but I'll check.

A little while ago I removed the controller chip, as I imagined there is no other chip underneath, and I do not detect any oxidation.
Once the chip has been removed, a screen-printed message appears (visible in the photo), 8742 Keyboard controller, it also says J08-4017C, but I don't know what it means.
At this point, it remains only to clean the contacts well, and reassemble the chip, hoping that it will be enough to solve the problem.
But it's a job I'll do tomorrow.
Looking closely at the enlarged image, I notice that a track that reaches the chip is interrupted, being very small it was difficult to see, I think that by repairing it everything should be okay, and the PC can boot normally.
Obviously, I will also do this work tomorrow, and when the work is finished I try again to start the PC on the desk.

Update on 386 motherboard:
Today I solved, the problem as I thought was a bad contact.
What looked like a broken track was perfect, there was dirt on it and it looked broken.
Before reinserting the keyboard controller chip, I passed the PINs on a fine sandpaper, after the operation the PINs were like new.
When I turned on the PC, I heard several tac tacs and two beeps, on the screen I had access to the BIOS screen, by entering I could set various settings, but at the moment I was interested in getting to the OS start screen.
Exiting the BIOS screen, without memorizing the change (I will do it later), the PC points out the absence of a boot disk, at the moment I do not have the values ​​to enter (head cylinders etc ...).
From the main screen of the BIOS, in the bottom line, there is useful information to understand the manufacturer of the MB, I will check what it corresponds to.
Now I have to find the info for the 128MB CF (used as HD), and load DOS and WIN3.X, currently I don't have the coprocessor, but I think I order one, I have to choose whether the ULSI or the IIT, I choose these because they are both 40MHz, while i387 maxes out at 33MHz, so it should work asynchronously.
For the moment I want to configure it to make it work and usable, later I think to increase the RAM to 8MB and change the VGA ISA to a better one.
I also have various cards to plug in, including SB16 3COM-LAN and modem.

Today I made further progress, in the configuration of the PC, I added the ISA sound card (SB16 Value CT2770), but I had to move the controller to the ISA4 slot, because in the ISA2 slot there is a crystal that prevents correct insertion, that is the ISA card touches the quartz and remains raised.
I made a further progress with the 128MB CF, installing it in another PC (PCChips M810LR Athlon 800), I detected the values ​​of Cyln Head Sec, the capacity on the ATHLON800 is 128MB while on the 386DX by entering the same values ​​it switches to 123MB.
The numbers found are these: Cyln 980; Head 8; Sec 32.
Unfortunately on the 386DX I have an error (see photo), the operating system does not start, I have to understand what it depends on, that is, if it is a software problem (as I imagine) or the numbers entered in the BIOS are not good?

With the 386DX-40 I verified the start of the operating system, thanks to the CF taken from the 486DX2-66, in this there is already installed both DOS and Windows 3.1, plus other programs both Windows and DOS, among those for DOS there are various bench programs, I have used some but only one two three work, through these I realized that the PC probably works with the Turbo off, this is because it goes at about 20MHz instead of 40MHz.
So I have to format and make the 128MB CF bootable, but I would like to install DOS version 5.0 and Windows version 3.0, this is because they take up less space on the disk and because they were the ones used at that time on the 386, but to be honest the warranty seal indicates that MB of 04-1993 so W3.1 already existed.
The only doubt is if W.3.0 supports audio or multimedia, honestly I don't remember, but I think so, maybe not directly but with special programs (I guess), after I have to see for the drivers if specific ones are needed, not I think they are hard to find, in case I will try to find those for VGA Audio etc ...

The faulty search and repair work on the 386DX motherboard is practically done.
I was unable to get the SanDisk 128MB CF card to work, I thought it was a solvable problem, but it seems it is not recognized by the BIOS, because booting from floppy (A: \) it does not find the C: \ (CF) drive, maybe it is possible that it works on other PCs or with PCs with different BIOS (not AMI), so I'll try to order one with a capacity of around 256MB but not SanDisk.
Currently I have on a Canon 8MB CF, where I have installed a DOS version 4.01 (or 4.00?), I am trying to put a Windows version 1.03 in it, unfortunately the free space is not much, with DOS I have occupied about 1MB, currently I have about 4MB free, because I have installed a couple of bench and word processing programs.
However, the 8MB CF will work fine in PCs older than the 386, at most I would say an 8-10 MHz 286.
For a 386-40 you need something larger, at least 100MB, to install DOS 5.0 and W3.X.
So I think to switch shortly to a MB 486, I will try to solve the problems it has and fix it, I will choose one of the various MB486s I have, I think I start with a VLB with 486DX-2-66 and L2 cache not working (see photo).

Today I rebooted the 486 VLB motherboard with faulty L2 cache, the PC only works by disabling it, but the performance is obviously a bit poorer.
I remember that when it worked I had increased the quantity from 128KB to 256KB, adding the chips where they were missing and resetting the jumpers.
Today I tried to remove the chips and clean the oxide from the PINs, unfortunately despite the cleaning (which doesn't hurt) the problem is still present, that is, it has not been solved.
From the ISA post card I read in displays 96 and 95, so something happens at this point (96), I have to investigate exactly to understand what it is that creates the problem, I would exclude that they are the defective cache chips, because the ones I had added work, but it should be try these if they work in another MB, or you would need to test them somehow.

Between yesterday and today I started work on another 486 motherboard, however it is an ISA card (not VLB), it is a practically new card, hardly ever used, originally had a Cyrix DX2-80 CPU and it worked a couple of days, later it showed a blocking problem, and went to the assistance a couple of times, unfortunately they did not solve the problem, from a check done years later the CPU was found to be faulty.
I remember trying other CPUs many years ago, and the card would boot up displaying the BIOS screen, but it couldn't complete boot, that is, it didn't get to the OS loading.
Now I'm trying to get it back into operation, so why would it be nice to fix it, the small size allows installation in a mini-case, for power supply I would use a compact ATX (I don't remember what they are called) with ATX-AT adapter, the installable RAMs are both 30PIN and 72PIN, and if you want you can use both.
As a test CPU I am using a 486SX-25, alternatively I have another one plus a 486DX-33, these CPUs should be working, I have tried all three but the result when starting is the same, the post card always marks - - - on the displays, the keyboard LEDs flash and the floppy seems to start up, but I noticed a couple of things that maybe prevent it from starting, the lack of the 3.6V battery and an interrupted track in the back, this is perhaps not so much important because it goes from the pin of the keyboard chip (which is not present) to the pins of the keylock.
As a VGA I use an ISA CL-GD5420, I thought it had 256KB of memory, as it has two chips marked HY524800J-70, actually it looks like the memory is 1MB, while the I / O controller is still ISA with a winbond W837871F chip , for the ram I tried one from 4MB to 72PIN, but I will change it right away, and I'm using four to 30PIN of 1MB each, on two chips says NEC424256-70, on the third NEC D41256L -85.

Tonight I added the battery to see if it starts, unfortunately nothing has changed, so I don't know I'll try to fix the track in the back, but I don't think anything changes.
The symptoms of this motherboard are the same as many other bad cards that I have, unfortunately the post card is not able (in these cases) to understand what the problem may be.
Returning to the MB Opti 486, I ran other tests, to understand if the cache problem is solvable.
Unfortunately I didn't write down the BIOS setup when it worked fine, so I thought it might be some BIOS entry that needs to be changed to fix the boot hang problem.
But it is not easy there are several attempts to make, because some items have different values, I'll try to understand which ones are to be set, otherwise I would waste time for nothing.

Today I recovered another MB, it is a GMB-386SAL, basically a 33 MHz 386SX.
I had discarded it about twenty years ago, it had a problem starting, it made several beeps, in recent days I disassembled it to clean the oxide and remove the battery (now exhausted).
In the afternoon I tried it, and I discovered that the problem was due to dirt or oxide, it started without problems, and the beeps emitted are due to the low battery (nothing strange then).
For the tests I used the ram present, two pieces of 1MB each at 30 PIN, so in total there are 2MB, as a VGA I put a Hualon ISA with 256KB.
Next I put an ISA controller, to which I will add a CF-IDE adapter and a CF instead of the mechanical HD.
I put some photos to see how it is, unfortunately there is not much possibility to expand it, at the limit I could bring the RAM to 4MB, add a sound card and bring the VGA to 512KB (adding a couple of chips), I remember that there were some chip for the upgrade of the cpu (mounted on top), in practice it went from the 386sx to a 486slc, but perhaps it is easier to win the lottery than to find one (today).

In recent days I have increased the RAM to the 386-SX PC, I had only 2MB (two pieces of 1MB each 30Pin), I have added another 2MB with RAM of the same type (3 samsung chips), in total I now have 4MB, which is the maximum installable using 1MB modules, if I had a couple of 4MB modules each I could instead reach 10MB (2+8 😎 ), or reach 16MB with 4 modules.
within a couple of months I should get some 256MB CF cards, if they do not create problems at boot I will use them as hd and I should install DOS 5 and Windows 3.0.
soon I will add an ISA sound card, I have to choose one of three available, maybe I think to put one with AZTech chip (V 5.0).
tonight I rebooted worked on the Pentium 60 skt4, I finally managed to get it started from Hd and I ran some diagnostic programs, including HWInfo.
The card has only a couple of problems, the DALLAS DS12887 chip has a low battery, so it must be replaced or modified, the other problem is the 1MB CL GD5434 integrated video card, the card with the connector for the monitor is missing (VGA port ), I could make one with parts recovered from scrap MB, and try to understand how to connect the various PINs, but at the moment I use a 2MB vga pci SIS 6215, the more or less we are there.

Between yesterday and today, I examined a motherboard to be repaired, it is a slot1 with VIA chipset, the model from what we read is MS6199VA, it should therefore be an MSI, the chpset is NB VT82C693A and SB VT82C596B. The fsb from 66 to 133 is supported by the serigraphs, while the multi is between 3x and 6x.
With these values ​​it is possible to install a maximum CPU PIII with a maximum frequency of 800 MHz, while for the Celerons it is possible to reach a maximum of 400 Mhz (FSB 66), then the Celeron 433 Slot 1 and all the others (skt 370 with slot1 adapter) theoretically they would not be supported, but having no jumpers you can try, and could be detected correctly, also because this MB supports Future CPU with VCore of only 1.3V, so there would be no problems for low voltage cpu, unfortunately all those with voltage low are skt370, to mount them you need the adapter that already has on board the jumpers for the voltage of the VCore.
For the tests I would recommend and use a PII 400 or something else around that frequency, later if it works (because I still have to fix it), I could try other CPUs including a pair of VIA C3, even if at the same frequency I think they are much better. Celeron.
Going back to the fault, there are a couple of electrolytic capacitors to replace, I have a couple of recovery practically the same, the problem is that the previous owner has removed one and damaged the copper ring by interrupting the track that connects it to an SMD (it is visible from the images), the problem is that it is located in the upper part, if it were below it would be easier to re-weld and fix.
Unfortunately, having a recovery capacitor, its pins are rather short, if it had long (or if it were new) I would have no difficulty in resoldering it a little raised, and creating a bridge with the tin between the capacitor pin and the track that goes to the SMD, fortunately the PIN that goes to the damaged track is the negative one (ie to ground), so I would make a bridge to the interrupted track and any other ground point in the vicinity. Or maybe I balance one wire at least of the capacitor and then I balance it to the open track that goes to the SMD. Maybe if you have any suggestions for this they would be very useful.
After re-soldering I have to check if it works, that is if it starts with CPU RAM and VGA, if unfortunately it does not go, I should check a couple of tracks in the back if by chance they are interrupted, another that seemed broken was OK, and it was definitely worse than the other two, so I guess they are okay, then of course you can assume anything else, but the answer can only be had after trying the start (and avoiding head bandages).

This morning I finished the soldering of the two electrolytic capacitors, rummaging through the ones I keep aside I found a new one with the same capacity and voltage as the original one, it is only a lot higher, but being new it has long pins that need to be cut in the back after welding.
To connect the pin of the - to the broken trace that goes to the SMD, I soldered from above keeping the capacitor slightly raised, then for safety I also soldered it from below, visually it is not the best but it allows me not to make bridges and solder wires.
With the tester I checked that the solders and the tin bridge were ok (and also in the other capacitor), in fact everything is fine, unfortunately I did not check the lines from the 20 PIN connector for the power supply, we will see between little because it was important.
I gathered the pieces to test the MB, Cpu Slot1 Pentium II 350, a VGA AGP with 4mb trident 3DI9750 chip, a 64MB pc100 RAM.
Connected everything correctly, I install the CR2032 battery for the CMOS and connect the keyboard, I give power to the power supply and I try to start, nothing happens, I remove the switch because it does not work, to start I contact the two PINs with the screwdriver, the POST board marks - - and the LEDs are on (which I could not know by now), after a few moments I hear something frying and I see that they are the twin mosfets near the two capacitors (which I had changed), at that point I turn off but now I think it is too late, checking them later are gone, evidently there is something else broken.
I just hope not to have fried everything that was on-board, I don't know if it happened due to an abnormal voltage of + 5V supplied by the power supply, from a subsequent check everything seems to be normal, even if it is not a quality power supply, not I know if it has an overcurrent protection circuit.
At this point I would say that the board is DOB (Dead on board), with the replacement of the other broken components I do not know if it will be possible to resurrect it, the simplest thing would be to have another same MB.
I don't have it exactly the same, but I have a very similar one, slightly different but with the same configuration, same chipset, same VRM, same slots, and almost the same even in the tracks. However, it needs a total change of capacitors and a BIOS chip (missing), unlike the other one it has some jumpers to manually set the CPU multi, so at the moment this is waiting for spare parts (which I should buy in the future) .
For fried mosfets, instead I should check on some scrap MBs if I have the same ones, and if they are okay, finding them I could think of doing the repair looking for any other faulty components.

By doing various checks on the Mosfet of the MSI, one of which is the one that has fried, I looked for other MBs that had the same CEB703AL Mosfet, by chance it is present in another faulty MB, some time ago I had started to do some checks, to see if I could fix the fault.
It is a 6ABX2V VER: 1.2, from memory I think the manufacturer is Lucky Star, in essence it is an ATX BX Slot1, the problem is that it starts up but the POST card shows - -, while to turn off you have to disconnect the power , holding down the button connected to the PWRBT PINs nothing happens.
There is also the CPU fan socket (FAN1) faulty, the 12v coming directly from the Power Supply are missing, also in FAN2 the 12V is missing, while in FAN3 there are, so having a long enough wire the CPU fan can be connected there.
These are the voltages detected on the mosfets:
Q3 CEB6030L PIN1 + 4.52V | PIN2 + 5.01V | PIN3 + 2.06V
Q5 CEB703AL PIN1 + 0.66V | PIN2 + 2.06V | PIN3 + 0.06V
Q6 AMS1085CM PIN1 + 0.10V | PIN2 + 0.30V | PIN3 + 4.97V
From the measured values ​​it seems that the problem is in the Q6, apart from the PIN3 the voltages are very low or almost 0V.
Testing on-board the mosfets (or whatever they are) do not result in shorting, so it could be a problem due to something else, for example some electrolytic capacitor, the problem of the FAN is still strange, that could be due to an interrupted track in the back , but seeing them they look OK, I don't know if maybe there is a fault in that circuit, and the MB protects the CPU by preventing it from starting.
Later I would check if there is power to the BIOS chip and the RAM, unfortunately having no indication of codes from the POST card, you can not do anything but attempts, and try to check the area around looking for something that is KO.
It would be useful to have a spare AMS1085CM, replace it and see if any voltage returns, if I'm not mistaken pin1 (should it be ground ???), pin2 the Vout ?, and pin3 the Vin.

Yeah, Q3&5 look ok:

1                 5V
2123            |- D
3GDS  4.5V   G -|
4               |- S
5                 2V
6
7                 2V
8123            |- D
9GDS  0.7V   G -|
10               |- S
11                 GND

Probably Vcore.

I think Q6 is an adjustable regulator and doesn't look right. Pin 1 is adjust (0.1V), Pin 2 is Vout (0.3V) and Pin 3 is Vin (5V). Think it might be a cache voltage supply if slot 1 is similar to slot A. So Vout is very low. Output voltage should be roughly Vout=1.25V*(1+R2/R1). So Vout shouldn't be able to drop beneath 1.25V. Regulator could be broken, or something could be shorting Vout to Gnd.

[edit: oh, and on the fans, don't some boards have them switched through a FET so they can be controlled?]

snufkin wrote on 2021-11-15, 21:57:

Yeah, Q3&5 look ok: […]
Show full quote

Yeah, Q3&5 look ok:

Copy code to clipboard

1                 5V
2123            |- D
3GDS  4.5V   G -|
4               |- S
5                 2V
6
7                 2V
8123            |- D
9GDS  0.7V   G -|
10               |- S
11                 GND

Probably Vcore.

I think Q6 is an adjustable regulator and doesn't look right. Pin 1 is adjust (0.1V), Pin 2 is Vout (0.3V) and Pin 3 is Vin (5V). Think it might be a cache voltage supply if slot 1 is similar to slot A. So Vout is very low. Output voltage should be roughly Vout=1.25V*(1+R2/R1). So Vout shouldn't be able to drop beneath 1.25V. Regulator could be broken, or something could be shorting Vout to Gnd.

[edit: oh, and on the fans, don't some boards have them switched through a FET so they can be controlled?]

Thank you so much for your valuable suggestions.
Regarding the Q6, some time ago I too thought that the problem was in the PIN2, but I thought that the problem was in the Mosfet (or regulator), double-checking what is around I would have identified a capacitor that seems to be shorted (I have to remove it and retest it) , it is a 470uF 6.3V electrolytic (EC21).
Following the track of the PIN2 of the Q6, we arrive at the + of the EC21 capacitor, so far it seems to be all right, but by checking (in diode test) the PIN + and - of the capacitor results in a short.
Instead for the FAN, I looked for if I could find any broken tracks, there is a scratch but it looks OK.
I'll see if I can tomorrow, to change the condenserino, maybe while I'm also changing the one next to it, even if it seems fine.
I hope that's the only problem with this MB, but it's interesting if I can solve it, it could come in handy with other MBs that I have to repair, about this next week another six MBs arrive, according to the seller maybe they are all working, but from what I can see almost all of them have swollen capacitors to replace.

PC@LIVE wrote on 2021-11-16, 20:36:

Regarding the Q6, some time ago I too thought that the problem was in the PIN2, but I thought that the problem was in the Mosfet (or regulator), double-checking what is around I would have identified a capacitor that seems to be shorted (I have to remove it and retest it) , it is a 470uF 6.3V electrolytic (EC21).
Following the track of the PIN2 of the Q6, we arrive at the + of the EC21 capacitor, so far it seems to be all right, but by checking (in diode test) the PIN + and - of the capacitor results in a short.

I think it'd be quite unusual for an electrolytic to fail short. More likely to be one of the small surface mount ceramic ones. There are probably quite a few components on that output, so testing in circuit will just tell you that there is a component somewhere that's broken. It does sound like there's a short to ground on Vout of Q6, could be worth double checking by measuring the resistance from pin 2 to anything Ground on the motherboard, like the shields on connectors. If there is a short then after you find the cause you'll probably have to replace the regulator.

OK thanks.
In fact I thought of removing the capacitor to retest it, sometimes it happened that they are broken even if visibly intact.
But after I take it off, I will still check if there is anything else in the area that could create short.
For the moment I do not have a spare, I have searched among the scrap MBs but they are different, I will eventually search online if it is found, and in case I order a couple.
I will try to do as you say, measuring the resistance from the PIN2, I think there is not much to check, as for ceramic SMDs, I don't have a suitable tester to check them, the only thing that
I can do is use the Tester in diode test.

PC@LIVE wrote on 2021-11-16, 21:39:

as for ceramic SMDs, I don't have a suitable tester to check them, the only thing that I can do is use the Tester in diode test.

It won't check if the capacitance is the right value, but if you've got a multimeter you can put it in resistance mode and measure that across any capacitor. If you get anything reading near 0 ohms (say under 2 ohms) across a capacitor then you can be pretty sure there's a problem somewhere, but it's not necessarily the component you're measuring across. Normally there will be many components in parallel across a power supply and what you measure is the combination of all the resistances. If any one of them goes short then that whole circuit will measure as short. It often means having to keep taking off individual components and testing them out of circuit until you can find the one causing the problem.

snufkin wrote on 2021-11-17, 11:50:

PC@LIVE wrote on 2021-11-16, 21:39:

as for ceramic SMDs, I don't have a suitable tester to check them, the only thing that I can do is use the Tester in diode test.

It won't check if the capacitance is the right value, but if you've got a multimeter you can put it in resistance mode and measure that across any capacitor. If you get anything reading near 0 ohms (say under 2 ohms) across a capacitor then you can be pretty sure there's a problem somewhere, but it's not necessarily the component you're measuring across. Normally there will be many components in parallel across a power supply and what you measure is the combination of all the resistances. If any one of them goes short then that whole circuit will measure as short. It often means having to keep taking off individual components and testing them out of circuit until you can find the one causing the problem.

Thanks for your valuable suggestions, I'll try to do other tests, the difficult part of the SMDs is to remove them without damaging the tracks. Today I removed the capacitor that seemed to be shorted, in reality by testing it it was found to be working, I did not do further checks, since once removed it is not shorted.
Removing it you can see the tracks below, the - goes to ground, while the + goes to the PIN2 of the AMS 1085CM, two SMDs are connected to the same track, and both seem to be shorted, being more problematic for me to remove them, I check the + and the - of the capacitor removed, and again surprise it is shorted.
So I check the track of the - and the PINs of the power supply collector (20 PIN), surprisingly I find the + 3.3V shorted, doing further investigations, I find why, in JP1 there are bridges between the PINs 1-2 3-4 and 5-6, basically the - of the capacitor is connected to + 3.3V, this of course should be normal, since bridges have always been like this.
A further check performed with the diode test of the digital tester, shows a reading of about 20 between PIN 1 and 2 of the AMS 1085CM, even by inverting the leads, I don't know if it is an anomalous value, but most likely there is a fault, unfortunately I don't have another one with which to check or compare.