Repairing a tricky L2 cache issue of a SOYO 4SAW2 486 Motherboard

[@pvc988]

Good job finding that ACLK issue. It would certainly be one of the more difficult things to figure out.

[@scrap_computing]

Thanks! Yes you don't usually check the clock signals when the L2 cache isn't working.

[@atheatos]

Nice repair / debugging video :) Thank you for the mention! I measured really fast in circuit the C37 with an LCR meter at 50pf. Probably not a very accurate measurement, but I think you are ok there with 100pf.

[@scrap_computing]

Awesome, thanks! I will try a 50pF one if I see any stability issues :)

[@esc2dos]

@@scrap_computing I tried measuring C37 in circuit with my TC1 tester and the stupid thing says it's two diodes. Glad to see Atheatos checked it. Great information in this video btw. Very educational.

[@scrap_computing]

@@esc2dos Thanks for trying to measure the capacitor! I just tried with my TC1 tester and it shows 2 diodes here as well, so I guess I am using a good capacitor value haha :)

[@esc2dos]

@@scrap_computing Well at least we know the TC1 is consistent :) Glad I could provide some corroborating results. I can't express how much I love that Motherboard, having it now is like going back in time and rewriting history. Back in my 386 days this board would have been mind blowing. Great that you revived it.

[@frankgrudge8823]

My 486 hangs with the same cache issue, you have inspired me to take another look thanks

[@scrap_computing]

I hope you get it working again :)

[@Choralone422]

I really appreciate your tenacity in repairing hardware like this and I wish I had soldering skills like you do!

[@BrassicGamer]

Lucky that the datasheet is available for that chipset really. Not often the case with 486-era hardware.

[@Megatog615]

incredible this board has isa, vlb, and pci

[@scrap_computing]

Yeah, they are not very common, which is why I wanted to rescue this one.

[@izdebo]

If you are looking for , i have a few for sale :)

[@mihalym.6876]

Thank you your post. Presistence and some luck - having datashet - are essential for successfull repair. Nice to drill down to the clocks and showed us the measurements. I suggest to leave the battery charger circuit in place and use the MAX40203 low drop diode with the CR2032 to avoid charging. The diode is tiny, maybe worth to design a little PCB to hold the diode and the battery holder together.

[@scrap_computing]

Thank you for comments. I have designed one of those PCBs that accepts all three: an ideal diode, a throughole diode and a SOT-23. Yeah I may restore it back to the original state.

[@mihalym.6876]

@@scrap_computing Do you mind sharing the project via pcbway? Thank you in advance!

[@Vanessaira-Retro]

Superb work in getting the board semi up and running. Amazing job as always.

[@darthtripedacus1]

Thank you for the amazing videos. Makes me want to get my Pii system up and running again.

[@scrap_computing]

Glad you like them. Yeah you should definitely get it up and running, PIIs are really cool!

[@nicushorul2007]

Great video, subbed! I have to say I was surprised at your decision to replace the unknown capacitor with a random value - although as others have said, the video would've been less interesting if it weren't for this :). I encountered a very similar situations, with an ATC-1425 board from which someone harvested a few SMDs, mostly in the VRM and PLL areas, and in the case of the PLL since the capacitors weren't decoupling but part of a RC network of some sorts I chose to leave their spots empty after seeing the board worked just fine without them. I just wish I had your diagnosing skills...

[@olivedrabcorp3026]

Once again, here is another excellent video, interesting and very well documented. You deserve way more views.

[@Vile-Flesh]

Fantastic video. I learned a lot from this repair. Thank you for uploading.

[@maxtornogood]

Nice troubleshooting clip, I do find these interesting!

[@tomasz89g]

Great work on this one; solving issue after issue to get a winning result in the end. Appreciate the explanations, helping improve the electrical knowledge and debug process!

[@BlackAdder4Ever]

Measured in circuit value of C37 of 80-85pf, using Uni-T UT61E multimeter.

[@scrap_computing]

Thanks!

[@JorgeCarvalho_web_dev]

Hi Sir! Amazing video! Fantastic repair job!

[@pointblank722]

I have exacly same problem with U-board ST1A and tried everything other than looking for signals etc...

[@registrazioniduemillaotton6030]

Very informative video. Thanks for sharing :)

[@adammace935]

Nice work indeed! Great video

[@ayan.debnath]

AWESOME video!

[@Aisflou]

Great repair!

[@collectingretrotech]

Excellent video

[@XtremeKremaTor]

Well done. Mighty impressive

[@shaunclarke94]

Are there any metal parts under the AT or PS2 connectors? Just as you didn't insulate the repaired traces. But great video! Loved the tracing of the cache pins.

[@scrap_computing]

Good point! I probably should have added some clear coating to insulate the traces. These connectors are mostly plastic, so it should be fine :)

[@shaunclarke94]

@@scrap_computing I'm always more aware when I'm about to cover something permanently that I can't easily access again that I haven't forgotten anything that could come back to bite me lol.

[@johng.1703]

Some of the boards from SOYO. We’re actually quite good, but those were pentium boards. They even did a half decent pentium 2 board too.

[@scrap_computing]

Yeah SOYO was pretty common in PII boards, but I think they were a well respected board manufacturer.

[@johng.1703]

@@scrap_computing I used to work at a distributor in the UK, the HX and TX boards were very good Pentium boards, the TX being the later and better board.

[@rasz]

and then PCChips bough them ahahhaha, fate worse than death

[@Constantin314]

mega awesome repair. i see you as an Einstein, not kidding :)

[@charonunderground8596]

Very good job !

[@SachinKumar-wx3up]

Which Epsom programmer are you using

[@GigAHerZ64]

ASUS PVI-486SP3 has capacitor C34 right next to the chipset itself. Unfortunately i was unable to measure it in-circuit. :(

[@scrap_computing]

Thanks for trying to measure the capacitor value. @atheatos , who owns this exact board, says it is 50pF :)

[@JohnSmith-iu8cj]

Why solder a ps/2 connector for AT cases that don’t have an opening there? Or have they?

[@rasz]

because drills are a thing

[@SianaGearz]

You don't have to install an AT board into an AT case! You can just install it into an ATX case, much more abundant too. Then you have two courses of action to make it pretty. The classic one is to take an AT-to-ATX IO shield (standard) and add a hole to it for the PS2 mouse port. The other is to take a universal mesh IOshield and punch out the ports that you need. I'm sure you can think of more ways to go about it.

[@JohnSmith-iu8cj]

Makes sense. Though I prefer time correct AT cases when possible. With a ps/2 slot bracket.

[@friendlywizard6462]

Some of the last made AT cases did have a smaller opening for a PS/2 mouse port next to the bigger 5 pin din keyboard port. I remember having a AT case that had that layout.

[@laudennn]

great vid, thanks

[@DarkennSlots]

very enjoyaable to watch

[@johng.1703]

Yeah you would get a PS/2 cable on a back plate in the box with those back in the day.

[@myne00]

Not sure if you know this, but using a dishwasher with no soap is the lazy way to clean boards. Rinse in isoproply/metho if you're cheap after and it'll look new.

[@Mr_Meowingtons]

i have never had any good luck with 486 and PCI slots..

[@scrap_computing]

As far as I know PCI in 486 boards is not very well implemented so it is not compatible with all PCI cards. I guess it is similar to AGP slots in socket 7 boards.

[@dorinxtg]

A M A Z I NG!

[@jozsiolah1435]

The L2 cache seemed to be nothing earlier, a 128 kb or 512 kb had no difference. If there was no cache or was cache, it made a difference. With Win 7 activated windows, Sec Essentials regularly opened the tray, asking for the Win dvd to be inserted. Windows does it yearly, even offline. The use of the cache later appeared to be true, at 50% load, 1 Ghz the windows is as fast, as at 2.1 Ghz full load. My theory is, that the Cpu uses 0-5% during startup, because the chipset uses the cache for everything, instead of asking the cpu to work. My cpu is only loaded when I play, watch video or streaming. The cache is important to keep the fan at minimum rpm, especially when someone takes a laptop into an office, the high speed fan is disturbing. For 486, the cache is able to switch the cpu to passive cooling, and the fan may not be needed. Cache frees up the cpu from loading it, the chipset uses it. Similar to the difference between the pio, and ultra DMA mode, but that’s about the hdd. Inactive, unregistered windows won’t use the cache, and the cpu will heat up. Win 9x has a reg trick, you can no longer register it. Receive any A4 fax successfully.

[@SianaGearz]

No.

[@watchmakerful]

[2:36] Duke???

[@scrap_computing]

Duke!

[@ctiborkoza8944]

Great Joob 👍

[@aspinx]

Replacing a capacitor with some random value... what could possibly go wrong. That's why it's worth checking if it's at least a filter cap or not, so you won't have to replace cache, desolder/replace 7408, check the board with oscilloscope and going through chipset datasheet. On the other side - the video won't be so interesting then!

[@scrap_computing]

Haha yeah, I just assumed it would be a decoupling capacitor, but didn't bother checking :)

[@cosmefulanito5933]

If you are removing the solder mask and the joints are good, there is no need to tin coat the traces. What you should do is cover the tracks with mask. What you didn't do. In other words, you did what you shouldn't and you didn't do what you should do. You should NEVER remove the battery charging circuit. Use a compatible battery. Please be a good technician and do things correctly.

[@petarmajic9415]

I agree that he should have applied a protective insulating layer to the traces, but reinforcing the acid-weakened traces is a must! A trace that is too thin can cause the malfunction, or even can burn under voltage (I have witnessed both situations). Therefore, I cannot completely agree with what you wrote.

[@kemencegaming]

Why you sound like Kermit the frog