VOGONS


My new 486-386 junk boards [IT gore]

Topic actions

Reply 40 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

Without CPU the VRM outputs 4.5V when set to 3.45V mode. With CPU it drops below 2V.

I moved the jumpers several times for the CH471A. I even tried to send 5V into the VRM output (there's no regulator installed on the board) and set the jumpers to 3.3V, but still nothing for VCC ...

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 41 of 102, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

What regulator/transistor do you have installed there now? And are you sure you're measuring the CPU voltage in the correct spot - which socket pin did you use?
Rather than measuring voltage you should check the resistance from CPU socket VCC pins to that jumper(s). It should be well below 1 ohm.

Reply 42 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

On the ATC1415, the same one as seen on the photo : LT1587 CT, but I noticed on other people's board, another regulator is used.

On the CH471A, there's none, I just bridged the 5V to the regulator output. When I measure voltage between the VCC pin and +5V there's infinity resistance, but it should be below 1 ohm since I bridged the vcore with 5V.

I measure voltage using the VCC pin the closest to the key pin as it's easy to find. It worked well before, but maybe I'm doing it wrong ?

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 46 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

So I measured the resistance, and it's 100ohm to the +5V (in each case : if I set the jumpers for 5V operation, or if I set the jumpers to VRM (which is bridged to +5V)). I don't know what could be the cause of this. A failing capacitor ?

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 47 of 102, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

Did you check in all 3 socket spots? I'm asking because on 3V3 CPUs one VCC pin is still getting 5V - this is so the I/O can work at 5V still, it's a reference voltage for the pins in other words. And I can't remember which one it is. In fact try to locate it by testing resistance ALL of he VCC pins in the socket to 5V. If you get one that is diffrent value it should be reference voltage. I would expect it to be connected directly to 5V plane so the resistance should be near zero.

I kind od lost track of which mobo we are talking about by now, can you make a photo of the socket and VRM area? Maybe there's a polyfuse there somewhere that has failed?

Reply 48 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

It's about the CH-471A (I'm not doing anything with the ATC1415 right now) now the socket has been replaced (I must update the first post list)

So here are photos of the area
JMtFWBWl.jpg

emnOqQVl.jpg

I haven't cleaned the flux yet, like I said I'll clean all the board once I'm done with all of them

So I came back and measured the resistance on all the VCC pins and it's still near 100 ohm resistance on all of them (sometimes it begins with 89 ohms but then it slowly grows to 95-96 ohms). But now the voltage on these pins is 2.5V and not 1.x V 😕

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 49 of 102, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

Huh. Okay, then test a few VCC pins in the socket against each other, do you get close to zero resistance?
Chances are someone put in a bad CPU (or not correctly) and the copper trace that supplies VCC to socket, and which probably goes through the middle PCB layers, is toast. You simply have to confirm it and then bypass it with a wire on the bottom side. But don't do the bypass until you're sure.

Which jumper(s) control the voltage? JP34? If so, it should be something like this: (1) - +5V, (2) - socket, (3) - regulator output. So if you get good connection on the power in pins, but the middle one that goes to socket shows higher resistance, it's probably a problem with copper being damaged. Weird, you'd normally expect to be some sign of of that. To be sure, figure out which pins you need to test and also check on the bottom of the PCB - just in case it's a soldering issue somewhere or a problem with the socket itself.

Last test, make sure there is no low resistance between socket VCC pins and ground, in case it's a bad capacitor that caused that. If not, go ahead with the wire bypass - if you want to be careful use a thin wire first, no CPU, just measure the voltages. If all looks good replace with thicker wire and try to boot.

EDIT: Actually, for the final repair maybe use several wires to opposite corners of the socket, to improve power delivery. Since we don't know how it's routed and you wouldn't want to end up with a system unstable under heavy load.

Reply 50 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

Jumpers 32, 34 and 35 control the voltage (so the white one and the two blacks underneath it)

All the pins in the socket seem to have 0ohm continuity.

After paying closer attention, I noticed this

jumpers_underneath.jpg
Filename
jumpers_underneath.jpg
File size
363.63 KiB
Views
872 views
File license
Fair use/fair dealing exception

this is the other side of the jumpers that control the voltage. See that black mark ? It's not a shadow, it's something on there.
Also I remember having difficulties removing the jumper on that was on this exact same pin, so it's highly possible the trace inside the board actually burned !

Edit : so I've put a jumper wire (of course I've checked that it wasn't shorted with ground, and it's not) and now the CPU receives 5V and heats up.

But still no sign of life 🙁

Maybe the power part of the mobo wasn't the only part to be damaged inside, but I have nothing to check the signals, etc ...

Unless you have an idea, I think I'm gonna give up on that one (I haven't put cache, do you think not having cache even though it's configured to have some could prevent the CPU from starting up ?)

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 51 of 102, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

Ah, so it is burned out after all. I did notice the darker area but I thought it was also flux residue you mentioned.

These mobos start with no cache and no RAM, just CPU fetching code from BIOS ROM. So just the CPU, BIOS and test card should give you codes, first for the missing RAM and then maybe due to missing cache chips. If you get no codes at all it's not going to be an easy fix.

Without a decent oscilloscope you really can't do much in situation like this. I mean there are ways if you have time and patience but that's still looking for a needle in a haystack.

Reply 52 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

Yeah that's what I thought ...

Well if anybody wants it, I can give it to you (just pay for shipping), else it'll end up as a donnor board too.

Moving on to the board with a burn mark in the socket (I have to add it to the list) I've received recently. The burn mark doesn't seem to prevent it from starting up (even though it doesn't look good), I have POST codes, but there is a little bit of corrosion so I'm gonna treat that right now)

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 53 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

New board photos added, see the first post. It's the "TK-82C491/GP-4N-D28B1"

Totally unknown board apparently.

I got it to POST (I fixed a dead trace going under the keyboard connector) and swapped the KBC (seems like the old one wasn't working properly somehow), but I can't do much beyond that :

I can access the BIOS setup and do stuff there,but nothing more, as for some reason, the board decides to reboot during the "WAIT" message of the AMI bios.

Also, time doesn't tick in the BIOS, so perhaps the 32.768KHz crystal is dead ? And maybe that's the reason why the board is stuck in a loop.

I'll have to investigate that.

Edit : well i've swapped the crystal, it still doesn't tick ...

I noticed two pins in two of the VLB slots get stuck, I managed to get the board to boot on medias several times, but it's very very hard to get the HDD controller working for some reasons (and when it does the partition of my SD card is totally corrupted). I tried to put several controllers in the slot where there is no pin that get stuck, and it's still not working very well ... The slots look good, but maybe they're dusty, I might try the contact cleaner there.

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 54 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

Moving on back to the AT1415.

Well I tested another VRM (the one from the PCChips), and now I can get 4.0V and 5.0V. That's better, and that tells me the VRM that was on there was probably salvaged from somewhere else and thus was incorrect.

... sadly I damaged the vias in the process of removing the one I've put on there 🙁 Can't really say it's all my fault because as you saw they've been through quite some stress already. I'm gonna buy the same regulator as the one present on treeman's ATC1415.

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 55 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

So I'm about to buy the TIP127A transistor ... but all I can find are "TIP127" (without the "A") transistors. Are these the same or not ?

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 58 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

Moving on to the TK-82C491/GP-4N-D28B1 (at POST it shows "SNB-M022"). So as I mentionned, the clock didn't tick and the board was also stuck in a boot loop. The other day, I threw contact cleaner in the VLB slots and tested the board further. The time was still not moving and the VLB seemed to be very flaky (especially IDE which I never managed to work properly).

Today, I fired it up once again, and the IDE is now working fine 😕.

The clock still didn't tick though, but I took the opportunity to run quake benchmark on it, and I tweaked some settings to improve the performance (despite not having cache, that board with a 486DX2 66 manages to reach 7.0FPS in quake 😲 )

Last time I rebooted it to tweak some more settings, I went back into the bios section where you set the time and the disk drives and ...

The clock was ticking 😕 😕 😕

I don't know what has happened there, but the board "fixed itself" twice today. Perhaps some old capacitor wasn't doing its job ? Or that the RTC doesn't like not receiving power from the battery ? I doubt this is due to broken traces since the board was still in its case when I received it (except if the CPU that has burned in the socket has damaged traces internally, but it doesn't look like it did, especially when the only really broken thing was the clock, which has nothing to do with the CPU ...)

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 59 of 102, by retardware

User metadata
Rank Oldbie
Rank
Oldbie
Deksor wrote:

See that black mark ? It's not a shadow, it's something on there.
Also I remember having difficulties removing the jumper on that was on this exact same pin, so it's highly possible the trace inside the board actually burned !

The PCB layer delamination is clearly visible.
And the solder tin is opaque, while it shines on other places.
Guess this must have been some poof sound and magic smoke 😀