Black screen IBM PS/2 Model 70s repair challenge. 8570-121 & A21

-
Support me on patreon.com/Epictronics
Join me on Twitter: / epictronics1
-
Tools I regularly use
DeoxIT D5 Contact Cleaner
Hanstar 861DW Rework Station
Pro'sKit SS-331 Desoldering Station
UNI-T UT61E Auto Ranging Multimeter
UNI-T UT890D Manual Ranging Multimeter
MESR-100 mk2 ESR meeter
PINECIL Soldering Iron
PinePowerPSU
TS-100 Soldering Iron
AMTECH NC-559-ASM Flux
Kester 951 Flux pen
MaAnt Grinding Pen
Multicore 60/40. 0.38mm and 0.5mm solder
TL866 II Plus Programmer
RIGOL DHO800 70MHz four-channel digital scope
Tektronix 2246A 100 MHz four-channel analog scope
InfiRay P2 Pro Thermal Camera
PCBs from PCBWay.com :)
--------------------------------------------------------------------------------------
patreon.com/Epictronics
Join me on Twitter: / epictronics1
Music by Karl Casey @ White Bat Audio

[@DuskHorizon]

No IBM left behind! Rooting for you.

[@Epictronics1]

Thanks!

[@AndyGoth111]

I definitely am interested to see power supply troubleshooting and repair

[@Epictronics1]

You're in luck. Four of my current projects have broken power supplies :)

[@AndyGoth111]

@@Epictronics1 Truly, fortune has smiled upon us!

[@Epictronics1]

@@AndyGoth111 😅

[@jscipione]

22:28 it appears that what you have there is an original 386 before there was a dx and sx variants there was the 386 which ram at 25mhz and is essentially a dx with 32-but bus. Lucky find!

[@Epictronics1]

Yes, I did some reading and I think you're right!

[@nickwallette6201]

Yep, was looking for this. Before there was a DX and SX, there was just "the 386" -- and it was what would later be called DX.

[@gt1man931]

Mistakes and errors illuminate the way. Never give up. 👍

[@jerther_]

I often find that lighter fluid does a really really good job of removing and even dissolving sticker glue residues and such. Been using it for decades. It's cheap and available at the grocery store ;)

[@jwhite5008]

why are you desoldering the logic ICs? not only they are most likely fine but you only need to test the logic - if its a NAND gate that has to output a low check which of the inputs is low and go from there

[@UpLateGeek]

Yep, first rule of electronics repair - thou shalt check voltages. And yes, that includes the power good signal. Between my three model 50s, they had two dead power supplies, and one planar with a shorted power rail. And obviously three dead/stuck ESDI hard drives. Thankfully I could assemble one working unit, and I've got a SCSI card so technically I can get it booting from that. Unfortunately even after many years of sitting on the healing bench, I still haven't had a chance to actually get any further than confirming it POSTS successfully. Although even if I get more than two days of leave this year, I've got plenty of other projects with a higher priority than the model 50, so I'll probably just end up working on those instead. At least I know of a KZbin channel that likes making videos working on old IBMs! 😁

[@Epictronics1]

Yeah, I've checked the power good pin on so many power supplies over the years. It has never failed until this day, when I didn't think to check lol

[@simontay4851]

Just connect the power good pin to 5V. They're both 5V.

[@UpLateGeek]

@@simontay4851 If memory serves, the power good signal is asserted by the power supply to signal to the motherboard that the power rails are stable and it's clear to boot. To bypass this signal, you'd need to disconnect it from the power supply side, since there's some chip or transistor inside the power supply holding it low. Probably OK to do it for testing if you're certain the rails are good, however it's better to figure out why the power supply thinks its rails aren't good since there's evidently some underlying issue with the power supply.

[@Epictronics1]

@@simontay4851 That would probably work, but there might be more to it.

[@nickwallette6201]

@@simontay4851 If the PSU doesn't assert PG, there's a reason. Unless it's just the pin's last transistor that's broken, bypassing PG is unlikely to make _anything_ better, and might make it worse.

[@jasmijndekkers]

Steven knows that type of IBM PC. He works a lot with it. Normally it give a error. Greetings from Steven from the Netherlands

[@patrickbonner3764]

To remove Glue, Stickers - i use WD40 and have nice Results!

[@minombredepila1580]

The outcome was a failure but the video is amazing. A lot to learn from. Thanks for it !!!

[@Epictronics1]

Thanks!

[@PoloCamper]

If needed I can dig out my model 70 and do some testing. Don't have a scope but only logic probe. Greetz from the Netherlands.

[@Epictronics1]

Thanks! I'll get back to you if it will be needed

[@UpLateGeek]

I bought a bag of those red switches from the UK. They manufacturer is "Arrow England", which I'm guessing was a part of Arrow Electronics, and they're marked "93RP4040B103 K".

[@Epictronics1]

I have a bag of those too :) More on this in an upcoming video...

[@mogwaay]

I've been told D-limonene is the absolute best stuff for cleaning glue and old tape - maybe it would work well with the black goo?

[@tonybardalen]

These videos are golden! Love your work on old IBM`s 😊 Even when you fail to fix it 😁

[@Epictronics1]

Thanks, We'll get there eventually

[@nakfan]

Great filming and editing 👍Per (Bornholm, DK)

[@Epictronics1]

Thanks!

[@BigBadBench]

A Model 70 is what started me learning about computer repair. I should have some IBM 2mb SIMMs, but I can only test in my model 80 on an expansion card.

[@Epictronics1]

I may have a SIMM. I need to check what part numbers are compatible and go through my stuff. I didn't know the Model 70 was so picky with RAM.

[@Mr_Meowingtons]

Ahhhh, I need this. I have the flu real bad, and it's so nice to lay in bed and watch retro pc stuff 😊

[@davidrmcmahon]

Get well soon.

[@jasmijndekkers]

I hope you will get better each days

[@SidneyCritic]

You can try Nifti for cleaning, which is a strong water based ammonia cleaner. As for solvents, general purpose auto thinner works well and is the weakest thinner, but it will remove writing on labels and may damage plastics. My Caig(Deoxit) rosin flux looks like it works way better/faster than that flux, ie, way more aggressive cleaning, it's just that it's so gluey it's hard to clean, ie, after IPA you need Nifti. It's so good that you only need to dip the solder wire in it, ie, a tiny amount, and it's plenty enough to make a shinny joint.

[@Epictronics1]

I think I have some paint thinner. I'll give it a try, thanks

[@tony359]

Came back to watch the part I missed yesterday - thanks for the shoutout! :) Can't wait to see the next episode!

[@AndyGoth111]

29:58 My favorite technique for "cutting" pin header strips is to firmly grip with two pairs of pliers, held snugly alongside each other, one on either side of where I want the break to be, then I can precisely snap the strip at the intended spot

[@Epictronics1]

I'll try that, thanks

[@AndyGoth111]

@@Epictronics1 I haven't had to do it in awhile, but I recall a mild alternating rocking (not twisting) motion working best to induce fatigue into the plastic at its thinnest point

[@GGigabiteM]

CRC Lectra Clean can get off lots of stubborn residue from PCBs and metal. The only problem with it is that it eats and discolors some types of plastics like ABS. It will also etch LCD screens, so you have to be careful when using it. You can hose down an entire PCB with the stuff if you need to, just make sure its completely dry before you put it near any sort of case plastic. It works A LOT better than IPA removing leaked capacitor electrolyte. IPA usually just smears it around and leaves white streaks.

[@envoycdx]

Not a failure video, all good learning :)

[@Epictronics1]

Thanks :)

[@maxtornogood]

Heh, what a saga this is turning out to be!

[@Epictronics1]

I honestly thought this would be a quick restoration lol

[@maxtornogood]

@@Epictronics1 Wishful thinking indeed! 🤪

[@NDR008]

Looking forward to this one.

[@humidbeing]

I bought a Tauntek IC tester kit and built it up. It does more in depth testing for old school 74 series logic chips. Highly recommended over the 866.

[@markdjdeenix6846]

wd 40 removes most glues .it might un stick lables so you know .

[@Epictronics1]

I'll give it try, thanks

[@markdjdeenix6846]

@@Epictronics1 works well with contact adhesive.incidentally if you ever get contact adhesive on your hands wd40 will get it off

[@JMassengill]

Love your work.

[@Epictronics1]

Thank you :)

[@8antipode9]

Love the video, as always! I have a similar experience with my EEPROM programmer, you can't entirely trust the logic IC test. I'm not ready to buy any of those fancy retro chip testers either...I find if I work at it enough (struggle...) I can usually get it working.

[@Epictronics1]

Thanks. I'll probably buy a retro chip tester someday

[@benjaminwirth5192]

If ipa not works for cleaning i try white spirit. Works good with glue or dry thermal paste for example.

[@pvc988]

Toluene is a strong solvent. It might get rid of that residue.

[@pipschannel1222]

Nice machines you've got there Roman! Love those early 386 beasts. To remove that terrible goo: Try orange terpene (or goo gone / any other sticker remover). It smells really bad but it worked a treat on my Model 60 and 80 👍

[@Epictronics1]

Thanks Pip, I'll give that a try in the Model 80 video

[@eDoc2020]

I was screaming "power good" throughout half the video. Glad you got there in the end. Did you check the PG signal even after putting in the known good PSU?

[@Epictronics1]

lol. It always checks good. Until you forget to check it! No, I ran out of time.

[@SidebandSamurai]

I am pretty sure you looked these codes up: 221 Machine error when copying the BIOS into RAM (8580-111 only). ROM to RAM remap error. An error occurred when the system checked the 128KB that was to accept a copy of the ROM. This is a latching error; once the error is detected, the system will no longer try to move the ROM to RAM until diagnostics (SETUP) are run to clear the error. 225 Some of your memory may be defective or not fast enough for your machine. The 221 maybe a model 80 error but might give you a clue.

[@Epictronics1]

Yes, thanks

[@danniemortensen4217]

I remember Those machines. I bought som Ram from IBM back in 1998 for 4 times The Price for an equel size to a clone pc. And it had to be that brand! It was, if i remember correct, 2 MB fastpage memory. But i newer found out why normal non ibm ram would not work. It just posted an error and that was it. And in 1998 The internet was bad, so No help there.

[@KrissBartlett]

hey at least you got that far very good try hope you can fix it

[@Epictronics1]

We'll get there, thanks

[@simontay4851]

There are so many ICs on that board. The fault could be any of them. Without schematics, where do you even start.

[@86smoke]

I think many old IBMs required RAM with pairity chip and if you manage to find one, they are very picky when it comes to comaptibility (don't ask me how I found it out with my PS/Valuepoint).

[@Epictronics1]

Oh, am I going to get the same issue when it's time to restore the PS/VP...

[@86smoke]

@@Epictronics1 9-chip memory module worked for me - no coincidence it was branded IBM

[@Epictronics1]

@@86smoke Ok, I used IBM branded parity SIMMs for this board, but apparently there only seems to be one IBM P/N that works in the Model 70 :/ I'm hoping there is a hack to use other modules

[@86smoke]

@@Epictronics1 Well, I intended to share an info on PS/VP restoartion, not the board in video. Besides, what model of PS/VP do yo u have? Mine is 325T/S with weird Cyrix 386SX with internal cache

[@Epictronics1]

@@86smoke cool, the weird ones are the most fun. I'm not sure what model I have. It's in storage, so I can't check right now

[@fra4455]

Nice job

[@Epictronics1]

Thanks!

[@fra4455]

@@Epictronics1 nice job

[@drkamilz]

Come on, fix these damned machines!

[@Epictronics1]

We'll get there

[@TheCaitlinlopez]

That is a turbo cache

[@pumelo1]

70s????? intel386 is first made 1985😉

[@Pickle136]

dx chips normally need memory in pairs, is it the same for these ps/2's ?

[@Epictronics1]

I'll double-check, but I think this board will run with one module

[@86smoke]

@@Epictronics1 It depends on module. If it is 36bit parity module, then you are all good.

[@OzzFan1000]

You should only need to match the memory bus width of the CPU. 72-pin SIMMs are 32/36 bit wide (non-parity / parity). A single 72-pin SIMM should match the 32/36-bit width of the CPU memory bus.