IBM 5162 PC XT 286 Troubleshooting the Grinch

[@Epictronics1]

Support me on patreon.com/Epictronics Buy me a coffee: ko-fi.com/epictronics Join me on Twitter: x.com/epictronics1

[@PixelPipes]

Maybe you'll have better luck Christmas 2025 when you do part 3! This will be the longest running repair series in YT history at this rate!

[@Gectms]

Commodore PC-10 🤔

[@Epictronics1]

This is nothing compared to the PC10 III

[@jasmijndekkers]

Hi Epictronics, You did your best. Maybe later you will fix it. Greetings from Steven from the Netherlands

[@Epictronics1]

Yeah, we'll fix it eventually :)

[@psilimit]

It takes a big heart to try so hard to save a grinch like this.

[@terosaarela4555]

That’s a very high definition green screen monitor on the thumbnail. 😁

[@Epictronics1]

Yepp, those 5151s are 1084p ;)

[@jaimeduncan6167]

Happy Christmas!!

[@f.k.b.16]

The Grinch is a story of redemption and restored faith so...if it really is the Grinch.... It'll one day be better than ever 😂

[@Epictronics1]

That's what I was hoping for too :)

[@antonyd5976]

the 32768 crystal is used for cmos clock because if you divide it down you get 1Hz. that crystal was used for very cheap lcd watches you could buy in the 80's. probably still used for the same purpose...

[@SianaGearz]

Yes yes it's still in every single quartz and digital watch.

[@felixokeefe]

32.768kHz also known as a watch cristal. They're commonplace in most computers for the RTC circuit.

[@autobotjazz1972]

You're a mean one Mr. Grinch.

[@sa230e]

I use to have one of those. It was my first computer. If I would have known it was such a rare machine I would have kept it.

[@Epictronics1]

Lucky you to have one of these as a first computer!

[@g4z-kb7ct]

Swap out all the chips in sockets. There's likely to be a bunch of controller chips like 8259, 8237, 8253, 8255, 8288 etc etc.

[@Epictronics1]

Actually, most of the sockets on this board is my doing. They were installed in a previous repair attempt

[@SergiuszRoszczyk]

By looking at those symptoms I’d check all the buffers between different parts of the system bus. One of the chips might have faulty bit or is not passing information from one side to the other. I had such issue with Hyundai Super16V, where 74LS logic buffer died

[@thebayandurpoghosyanshow]

Merry Christmas, mate :)) I think I remember this Grinch from an earlier video, so it's nice to finally see the next leg of the journey.

[@Epictronics1]

It's been a year. I'll try to be quicker to the next attempt.

[@thebayandurpoghosyanshow]

@@Epictronics1 I'm fine with the prospect of seeing him steal your next Christmas, so no pressure, friend :))

[@Epictronics1]

@@thebayandurpoghosyanshow Maybe we should make this in to a christmas tradition :)

[@thebayandurpoghosyanshow]

@@Epictronics1 :D good idea. I'll be there.

[@g4z-kb7ct]

If the ready signal is not present at the slots trace the signal. It's possible it goes through some other chips first and there may be a broken connection somewhere or maybe it goes into a logic chip but doesn't come out due to a bad logic chip.

[@terosaarela4555]

Good evening and Merry Christmas!

[@Epictronics1]

Merry Christmas!

[@wacholder5690]

15:30 You cannot test the 32K crystal that way. You need a low capacitance probe (below 5pF) and / or 1:10 / 1:100 divider at least. And it has nothing to do with the board. Just the RTC. If it were dead you would get a date / time error.

[@wacholder5690]

The -Ready an active low signal. It got a pullup to +5V and is tied down by the clock generator when it got a stable signal. So it has to come low after power establishes. It was wrong on your first measuring when it stuck at +5V.

[@Epictronics1]

Ok, something seems to be glitchy there. I don't think it reached 5v before it dropped

[@OscarSommerbo]

32,768 kHz crystals are used because you can get very accurate 1 second pulses by just bit shifting it right 15 times.

[@tigheklory]

Maybe you should try a Coleco Adam Computer now! 😊

[@rodneyknaap]

I suggest you test with the MR BIOS generic 286 BIOS which works with this system. The MR BIOS is much more informative in terms of messages and beep codes. Also, when the system reports the first 64k to fail, this also may indicate a complete RAM or other system failure which causes the RAM access to not work. You also can check the refresh logic if this is working on the RAMs which would also be a condition for the system to be able to get further into the POST. I would also measure the data lines on the RAM side to see if anything stands out. The RTC is really tricky and you need to measure the 32k clock on U93 pin 4. If you measure on the crystal the scope may be stopping the oscillation. 12887 and 12885 work however it needs exact battery voltage on the battery pins or the RTC chip will refuse to work... Ready changing is due to reset because ready is pulled low during reset. If ready is doing nothing I would suspect the 82284 being faulty. Pin 11 on the 82284 needs to go down briefly to properly reset the 286 CPU. I don't love the method of letting the PSU apply reset through this input, and could be modified using a RC reset timer which may be more reliable to init the system. I found the 5170 also to be tricky to init and post and the RTC is part of the problem.

[@Epictronics1]

That's useful info. Thanks!

[@craigrenwick9132]

Merry Christmas to you hope Santa delivers a fix.

[@Epictronics1]

He just delivered another broken retro computer lol. Merry Christmas!

[@rlgrlg-oh6cc]

32768 is the timekeeping crystal frequency. You can't probe that circuit without disturbing it, as it's very low power. Some RTC chips have a 1 Hz output that you can probe, which will tell you that the oscillator is running.

[@Epictronics1]

I just probed my 5170 crystal and it worked! Maybe there is a way to probe it somehow?

[@rlgrlg-oh6cc]

One pin of the chip is an output and the other is an input. You can usually probe the output pin, but often times probing the input pin will stop the oscillation, or at least disturb it, due to the capacitance of the oscilloscope probe. There are special low-capacitance probes which have an amplifier located very close to the probe. Probably you could probe it with one of those. You should always use your scope probe in 10:1 mode to reduce the loading on the circuit you are probing.

[@Epictronics1]

@@rlgrlg-oh6cc That's very useful, tanks!

[@cocusar]

you can usually probe the pin that's the output (XO) or the one that shows the inverter buffer output, because it's actively driven and you won't disturb that one. or just use another output of the circuit, because that one used some other inverters for its actual output.

[@g4z-kb7ct]

@@Epictronics1 They are easy to test. Use a 74LS14 logic chip. Wire 5V and gnd to the chip, wire the clock output to pin 1, tie pins 2 and 3 together on the chip and measure pin 4 on the scope....

[@MSThalamus-gj9oi]

"That is just pure evil!" LOL!

[@curiousottman]

RTC circuit won’t affect post except maybe throw an error. Also you cannot test a 32768 that way (inconsequential though). Still, you are awesome and your soldering skills are top notch.

[@jozsefizsak]

That's a stubborn one for sure. My money's on you to beat some cooperation out of it in any case. 😊

[@Epictronics1]

It's stubborn, but I'm stubborn too :)

[@mal2ksc]

My hypothesis on why this machine happened: IBM had moved the AT to 8 MHz but still had a glut of 6 MHz 286 parts, and a glut of XT cases, and said "Hmm..."

[@TruckFan542]

Grinch is still causing his grinchy bloopers..😂

[@Epictronics1]

We'll keep at it until it's fixed :)

[@TruckFan542]

Maybe a SIMM slot is kaboomed?

[@g4z-kb7ct]

Half of the scope shots were unreadable. Put something under the scope to tilt it slightly and the display will be visible on camera.

[@Epictronics1]

Yeah, the anti-reflective film was a fail. I'll see what I can come up with

[@rodneyknaap]

Sorry to hear the many faults with this machine. I'm also puzzled about the choice of making this, maybe Don Estridge liked the original XT case more than the newer AT case. I seem to remember he once commented to especially like the XT. I developed my own ATX design for an AT based on the 5170, however I still would one day like to know the reverse engineered data from the byte conversion core AT logic PAL chip U27 in this system! If I ever come across a 5162 PCB with a PAL on it, I will look into this. I think I will rewire the PAL to reverse it with known methods and reverse the remaining pins separately based on the function of these pins. Many people don't realize the historic importance of this 5162 design and the key to understanding this design which remains locked inside U27 which has never been reverse engineered in human history... I have done this work with U87 on the 5170 as can be seen in my VCF thread.

[@Epictronics1]

I agree. We'll fix it somehow. I'm too stuborn to give up!

[@schnitzelsamy]

Cool

[@intel386DX]

Those EPROMs are without stickers to prevent damage or erase of the data inside . And what is this big one withou socket with window?

[@Xaltar_]

Did you restore the original crystal for the RTC chip? Merry Christmas!

[@Epictronics1]

Yes, at the moment, It has the original clock crystal resoldered. Merry Christmas!

[@THEtechknight]

The original video was far too long ago so i dont remember the details but. Have you checked/swapped the DMA controller? that would hold up the CPU. ALSO!!!! there are two different crystals! the behavior your getting with the new crystal is normal because you inserted a parallel-load crystal in a series circuit. Cant do that. The intrinsic load characteristics are radically different. Put the original crystal back in place, its probably fine. You cant test that circuit with a scope unless you are probing at the DRIVE output of a series-load circuit.

[@thedopplereffect00]

Feels like you were on the wrong path with that 16khz clock. 1/2 of the documented frequency isn't that surprising. Are you sure the oscilloscope was triggering right? Have you tried a full data/address capture? There should be a consistent address/data sequence between identical boards. I wonder if it is possible to step the clock on the CPU and observe each instruction being executed on boot

[@agentkruger]

😘👍

[@rager1969]

32768 should be familiar, as that's 32 * 1024, i.e., 32K. Why would they have a crystal using that as it's frequency? Is it a coincidence or an Easter Egg? I mean, would a round number like 32000 Hz (with some tolerance) be a better design choice?

[@adinnugroho6544]

Grinch? I felt familiar as it was Prank that Aussie KZbinr did 😅 His wife was used Face Mask, but she got pranked as Her Husband put green food dye on Face Mask

[@adinnugroho6544]

Where this man Come from? UK or USA? on about channel, it said USA, but it isn't the US accent I familiar with 😅

[@maxtornogood]

Ahh, so darn annoying!

[@Epictronics1]

What a Grinch :)

[@waldfruchttee]

i hate the youtube ai "translation" of videos... do they think i cant speak english?

[@evefavretto]

Yeah, it's getting on my nerves, but that's KZbin's fault.

[@Epictronics1]

It can be turned off. There's a button down in the right corner of the video

[@TSteffi]

What's worse than the audio track is the auto translated title. It's garbage, and there is no way to disable it

[@KaiMic718]

Ja, klingt gleich nach Werbung für irgend ein Abzockprodukt 😅 ... *ExtraDeutschFürDenÜbersetzer