VOGONS


My new 486-386 junk boards [IT gore]

Topic actions

Reply 21 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

Yeah of course I had a desoldering iron, this would have been nearly IMPOSSIBLE to do that without it (or at least in a rather clean way)

I haven't measured the time it took, the broken one was much tougher to remove (for some reasons, everything that's on that PCChips board is super easy to desolder, like it was made for that exact purpose 🤣). I'd say 30-45min to desolder the "new" socket from the donnor pcchips and 1h-1h30 to remove the one on the receiver board.

Some pins didn't want to get cleared of their solder, in the end I thought "whatever, this thing is broken and has to go" so I took a pair of pliers and a flat-head screwdriver and broke the thing into pieces and removed everything that could come off, few pins were left there of course, I took those off manually with the soldering iron.

Now I still have to clear out some holes, but I can't do that without solder, so I stopped there. As you can see, it's quite clean 😀

(by the way I uploaded a complete photo of the board just before you posted your answer, so it's on the previous page if you haven't noticed 😁 )

Reply 22 of 102, by treeman

User metadata
Rank Oldbie
Rank
Oldbie

yeah I just went back to the start and had a look, nice work!

I really wonder what the motivation was behind the person who did this stuff, how can u break pins on a cpu slot that will accept a cpu in every orientation, or why mess with the voltage regulator, could of clipped it off I guess if can't desolder but yeah... damage is done.

I really enjoy these repair threads especially with pictures of before and after. Think you deserve a good break while wait for new solder. keep us updated

Reply 23 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

I do too 😀

No pins were broken in the socket, it's just that the moving part and the lever were gone, so I can't insert any CPU in there ...

I guess once I'm done with all these boards (at least all those that can be saved) I'll make a "all done" post with all the working boards cleaned and fully tested.

I think I'll also make a photo of the donnor PCChips board : it will soon look like a PCB with just ISA/VLB ports and some components, and nothing else 🤣.

Also, a "new" board might be added to the thread : I just bought a lot of 2 AT computers, one has the exact same pcchips mobo as the donnor one and works (that'll make two working pcchips m912 for me 🤣 ...), but the other one is nearly empty, has no CPU ... and has some weird mark in the CPU socket (burn ?). I'll take a look at it, and if I think it can be fixed, we'll look at this one too !

Reply 24 of 102, by Vynix

User metadata
Rank Member
Rank
Member

Oh my... Talk about Tech gore, it's amazing to see that they still work (after some repairs) despite looking like they're beyond saving...

Bon courage 😊

Proud owner of a Shuttle HOT-555A 430VX motherboard and two wonderful retro laptops, namely a Compaq Armada 1700 [nonfunctional] and a HP Omnibook XE3-GC [fully working :p]

Reply 26 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

Well I have found informations about the 386s myself

So the 386DX is actually a Abit FU3

And the 386SX is ... I don't really know, but same as this one https://aukro.cz/mainboard-386-387-neat-sx-i3 … test-6943623516 (damn, I wish I could ask him to backup the bios ...)

Edit : Well I have found that ebay seller too. According to him it's a ECS board. I'm gonna ask him if he can backup the bios. https://www.ebay.com/itm/ECS-NEAT-SX-Motherbo … &frcectupt=true

I haven't made progress yet as the solder took more time to arrive than expected, sorry !

Edit 2 : oh well I discovered that the few chips left in the banks actually sum up to 2MB ! So all I really need is the BIOS test that board ! 😁

Reply 27 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

Allright, so I've found the bios for the 386SX ! Predator99 uploaded it there !!

80386 BIOS image collection

However ... it still does nothing 🙁 I'm not sure if the ROMs I used are wrong or anything (I used Winbond 27C512 EEPROMs).

Anyways, I've finally received tin, so there will probably be some interesting news for tomorrow 😉

Reply 28 of 102, by treeman

User metadata
Rank Oldbie
Rank
Oldbie

is that a 512kb chip? I read somwhere once not sure if it was vogons or not that if you are using a smaller bios size file on a bigger chip you need to write it twice or more times to fill the whole chip.

so if using a 256k chip for a 128k image you need to write it at the start of the eeprom then at the other half to fill to the end.

Not much info in detail from me but perhaps it can give you a direction to search in

Reply 31 of 102, by Predator99

User metadata
Rank Oldbie
Rank
Oldbie

The ROMs are already 64kb. I read them in my TL866 as 27E512, therefore each image should contain 2 identical 32kb parts. You can flash them to your 27C512 without modification. Did you try to exchange the H and L?

I am also not 100% sure if the image is OK as the board was not working. But most probably it will.

Reply 32 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

Thanks for the infos !

I didn't try to swap them (I made sure they were in the same order you indicated them) but sure I can try to swap them !

At least now I know the roms should work, so the battery damage is probably the reason for it not to do anything at all. (Chipset/CPU don't heat up at all)

Reply 33 of 102, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie
Predator99 wrote:

The ROMs are already 64kb. I read them in my TL866 as 27E512, therefore each image should contain 2 identical 32kb parts. You can flash them to your 27C512 without modification. Did you try to exchange the H and L?

I am also not 100% sure if the image is OK as the board was not working. But most probably it will.

Please don't do that - read 32kB '256 parts as 64k '512 ones. The reason is the A15 signal on '512 is in on the same pin as Vpp on '256 - and the '256 expect the Vpp to be at 5V when not being programmed and might not work properly otherwise.

So what happens is the Vpp will be set to logic level low (close to 0V) on lower 32kB, and to logic high (close to 5V) on upper 32kB. Therefore the upper 32kB should be dumper OK since actual power draw on Vpp is very low and any typical digital output close to 5V will suffice to "power" the chip properly. But the lower 32kB might be corrupted - it all depends on the chip. And most people would use the lower 32kB to program true '256 parts and might end up with glitched BIOS.

Reply 35 of 102, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

Do you intend to program these images into '512 parts as well? That would actually work as-is because the socket on the mobo will have a 5V power on Vpp pin so only the upper part of the chip will be addressed, ever - the A15 line will be always kept high. So even if the lower 32k is somehow bad, it'll work. Still, it's better to stick to correct chips and file sizes in case someone will use '256 parts as it was originally in the mobo.

EDIT: Also, if you want to split and rejoin the files, you need to keep the upper half and duplicate that.

Reply 36 of 102, by Predator99

User metadata
Rank Oldbie
Rank
Oldbie
Deunan wrote:
Predator99 wrote:

The ROMs are already 64kb. I read them in my TL866 as 27E512, therefore each image should contain 2 identical 32kb parts. You can flash them to your 27C512 without modification. Did you try to exchange the H and L?

I am also not 100% sure if the image is OK as the board was not working. But most probably it will.

Please don't do that - read 32kB '256 parts as 64k '512 ones. The reason is the A15 signal on '512 is in on the same pin as Vpp on '256 - and the '256 expect the Vpp to be at 5V when not being programmed and might not work properly otherwise.

So what happens is the Vpp will be set to logic level low (close to 0V) on lower 32kB, and to logic high (close to 5V) on upper 32kB. Therefore the upper 32kB should be dumper OK since actual power draw on Vpp is very low and any typical digital output close to 5V will suffice to "power" the chip properly. But the lower 32kB might be corrupted - it all depends on the chip. And most people would use the lower 32kB to program true '256 parts and might end up with glitched BIOS.

Thats intersting and new to me. Never checked if both parts are identical to be honest...

Reason I do that is that you cannot see if a chip is 256 or 512 from the outside if there is a label on it. Therefore I use 512 as standard not to miss anything...

Reply 37 of 102, by Deksor

User metadata
Rank l33t
Rank
l33t

I see. Yeah I plan to install these (at least for now to try to test)

Also : ATC1415 has a new VRM ! I took the same that was soldered there before ... Except it doesn't seem to really work ... I can either have 5v or 1.x/2.x volts, I assumed the one that was there before was the original one, but now I'm doubting ...

Also, the CH471A finally has a new socket ! ... But the board must have another problem, because the CPU voltage is way too low even when precisely set to 5V. Not sure what's wrong here, maybe some tantalum caps are dead ? (Voltages on the PSU are ok even when plugged into the motherboard)

Reply 38 of 102, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie
Predator99 wrote:

Reason I do that is that you cannot see if a chip is 256 or 512 from the outside if there is a label on it. Therefore I use 512 as standard not to miss anything...

That's a problem, yes. You could always check the resulting file and if the lower and upper halves look identical, discard the lower one. I suppose with unpredictable glitching the lower half might end up so corrupted as to look like a valid data - I never had that happen to me but it's not impossible.

At the very least maybe add a small readme.txt with one sentence of explanation that the actual ROM size is unknown so it was dumped as '512 since it's the biggest part with 28 pins. That way whoever finds that file (maybe years after you've dumped it) will have some idea of what to expect. Sorry if I seem annoying but I just had to deal with a bad dump myself.

Reply 39 of 102, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie
Deksor wrote:

Also : ATC1415 has a new VRM ! I took the same that was soldered there before ... Except it doesn't seem to really work ... I can either have 5v or 1.x/2.x volts, I assumed the one that was there before was the original one, but now I'm doubting ...

Also, the CH471A finally has a new socket ! ... But the board must have another problem, because the CPU voltage is way too low even when precisely set to 5V. Not sure what's wrong here, maybe some tantalum caps are dead ? (Voltages on the PSU are ok even when plugged into the motherboard)

ATC1415 - with or wihout CPU? Can you measure the voltages on all the pins of the regulator when set to, say, 3V3?
CH471A - if it's possible to switch between 5V and lower voltages then there should be a jumper that actually selects 5V or regulator output. Several even, one would not be enough for all the current (thought that never stopped mobo manufacturers from trying). If said jumper(s) is corroded, oxidated, or dirty, it could not be passing the 5V properly.