Help finding noise source on old 386 motherboard

[humidbeing]

Hi gang!

I'm trying to fix an old Packard Bell 386 motherboard. It's gives all sorts of random errors. But I think when I find the noise generator I will find the true fault.

This board emits so much noise when turned on that it will jam all FM radios in the house. I have a similar board and it does not do this.

Things I have tried:
Alternate power supplies
recapping board (actually made noise a little worse)
removing RTC chip
removing keyboard controller chip
removing everything from board but vga (on motherboard), ram, and cpu


The noise also shows up on the 5 volt power rail on the motherboard, and it shows up in the vga output of the onboard video chipset.

I'm a noob when it comes to RF. I have a good voltmeter and a scope at my disposal.

Any tips on how I can track down which component is generating all this noise?

[bob91343]

First try to analyze the noise.  Is it random, or are there definite repeatable signals?  If you can identify the frequency that could give a clue.

Another approach would be to wave the scope probe around to see where the radiated noise is strongest.

Perhaps it's power supply related.  You could substitute a different one.  Or try ferrites around the cables from the power supply.

[humidbeing]

Tried with two known good power supplies. No change in noise output.

[humidbeing]

First try to analyze the noise.  Is it random, or are there definite repeatable signals?  If you can identify the frequency that could give a clue.

Another approach would be to wave the scope probe around to see where the radiated noise is strongest.

Perhaps it's power supply related.  You could substitute a different one.  Or try ferrites around the cables from the power supply.

I tried this without much luck. What scope settings should I use? I assume I want ac coupling?

[T3sl4co1l]

Does it show up on ground as well?

A lot of vintage PCs shipped with a shield over the motherboard.  Do you know if that was the case here?  It may simply not be designed for low emissions any other way.

Tim

[james_s]

There's a reason these things were called Packard Hell back in the day. We had one when I was a kid and it had all kinds of problems down the road when we tried to upgrade the hard drive. I knew lots of other people who had them too, they were very common because they were cheap. It may just be bad design.

[humidbeing]

Does it show up on ground as well?

A lot of vintage PCs shipped with a shield over the motherboard.  Do you know if that was the case here?  It may simply not be designed for low emissions any other way.

Tim

No shield. This is definitely not a normal level of noise. It makes the vga output almost unusable, and the waveforms on the ISA bus look terrible and glitchy.

[humidbeing]

There's a reason these things were called Packard Hell back in the day. We had one when I was a kid and it had all kinds of problems down the road when we tried to upgrade the hard drive. I knew lots of other people who had them too, they were very common because they were cheap. It may just be bad design.

I specialize in restoring packard bells. I have never had one make interference like this.

[james_s]

Does it use buck regulators to provide any of the voltages on the motherboard? I don't recall when that became common. Severe jitter in one of the clock sources?

[humidbeing]

Nope. No regulation on the board itself. I checked all the crystals with my scope and they look good.

[humidbeing]

here's a photo of the board, if that helps

https://imgur.com/a/OEdOiwH

[Twoflower]

You said you have a scope. Have you tried to create a loop (maybe few windings) with the scope-probe and scan the board with that loop? Or even just add a short wire to the scope tip and scan over the board. Maybe it can point you to the noise source.

And put a sticker over the EEPROM before it gets a sunburn ;-)

Edit: While removing the keyboard controller it seems you damaged the resistor array between U16 and U21. It might be still OK, but worth a check. And remove the debris located at the array.

Also I noticed north of the ISA-slot there is U18. It looks a bit dis-colored. Could be a reflection.

[T3sl4co1l]

Looks to be built with internal ground plane, and grounded screw holes.  And EMI strip along the connectors.  It's made for a metal enclosure.  Are you just testing it out in the open?

Tim

[james_s]

That takes me back. Looks very much like the motherboard in the PB 386sx-16 my family had back in the early 90s. It had a single ISA slot on the board with a vertical riser that held 4 cards horizontally. That machine had the weirdest issue I've ever seen, it worked fine with the original 40MB WD IDE drive but none of the 3 or 4 larger drives we tried later would work. I could partition and format the drive, copy software to it, move files around but the instant I tried executing any program from the hard drive the machine would hard lock. The same happened while booting, it would hard lock at "Starting MS-DOS". Put the original drive back in and it was fine.

[humidbeing]

Looks to be built with internal ground plane, and grounded screw holes.  And EMI strip along the connectors.  It's made for a metal enclosure.  Are you just testing it out in the open?

Tim

Yes. But it does the same thing whether it's in the chassis or not.

[humidbeing]

Here's how the noise manifests itself coming through the VGA signal

[T3sl4co1l]

Interesting.  Well, I'd have to have a close look at it, with instruments.

Tim

[Haenk]

There are several quartzes on the board, so a lot of frequencies are generated at the same time. Scoping around really might show something, you might be ableto match the frequency to a quartz.

[Ian.M]

Back in the 90's Packard Bell PCs had a reputation for being very low quality, (usually worse than what the average back-street cash only clone builder of the era was building) and were cost conscious to the point of getting sued several times for using recycled parts sold as new!

I'm seeing a board full of ICs with Q1 1992 date codes.  Although its too early to have been from the 'capacitor plague' era, all the electrolytic capacitors on that board are twenty eight years old, and were 'economy grade' when new.  Odds are many or all of them have dried up, have high ESR and aren't doing much decoupling.

If you've still got the noise problem after fully recapping the board, you may have to dig deeper, though if one of the custom chips is bad, its probably BER.

[drussell]

Indeed, those things were absolute junk right from the factory.

You need to replace every decoupling capacitor for a start, if you want it to work right.

Personally, I would just throw anything from that era with the Packard Bell name on it as far as possible and run away, then pull out a real motherboard, but then again, I have dozens of proper, decent quality 386 boards here.   

[ChristofferB]

-are you sure the noise causes the problems, and not the other way around?

A gate in a gate array left with floating input (from ESD, faulty chip etc.) could concievably oscillate and create lots of noise.

I agree with the others, though. Unless the potential noise source is findable with the tools at hand, I'd toss the motherboard. They can't be expensive to replace via Ebay, if needed.

I was going to say 'get one of those cheap chinese 35-4400 MHz spectrum analyzers off of ebay' but for a one-off problem that's probably more expensive than replacing the board.

[james_s]

I agree with the others, though. Unless the potential noise source is findable with the tools at hand, I'd toss the motherboard. They can't be expensive to replace via Ebay, if needed.

I guess it depends on your definition of "expensive". I haven't shopped for 386 boards but I've sold a couple of untested as-is 486 boards for ~$75 over the past couple of years and those are easier to find than 386 boards. PCs of this era are pretty rare these days, especially the more proprietary stuff that people often junked instead of fixing or upgrading.

[humidbeing]

-are you sure the noise causes the problems, and not the other way around?

A gate in a gate array left with floating input (from ESD, faulty chip etc.) could concievably oscillate and create lots of noise.

I agree with the others, though. Unless the potential noise source is findable with the tools at hand, I'd toss the motherboard. They can't be expensive to replace via Ebay, if needed.

I was going to say 'get one of those cheap chinese 35-4400 MHz spectrum analyzers off of ebay' but for a one-off problem that's probably more expensive than replacing the board.

these motherboards are actually very rare and worth a good bit of money

[humidbeing]

Back in the 90's Packard Bell PCs had a reputation for being very low quality, (usually worse than what the average back-street cash only clone builder of the era was building) and were cost conscious to the point of getting sued several times for using recycled parts sold as new!

I'm seeing a board full of ICs with Q1 1992 date codes.  Although its too early to have been from the 'capacitor plague' era, all the electrolytic capacitors on that board are twenty eight years old, and were 'economy grade' when new.  Odds are many or all of them have dried up, have high ESR and aren't doing much decoupling.

If you've still got the noise problem after fully recapping the board, you may have to dig deeper, though if one of the custom chips is bad, its probably BER.

no offense, but if you would have read my first post you would have seen I replaced all the caps already. It actually exacerbated the noise slightly.

[Ian.M]

Sorry I missed that.

How much experience do you have recapping motherboards?  Was there any evidence of electrolyte leakage?

Use the wrong caps (e.g. too high ESR, or insufficient voltage rating), or damage a PTH barrel's connection to an inner trace and you can make the problem worse. 

I never was a fan of Oak Technology VGA cards.   Do you have enough info to know how to disable on-board video, (which I'd bet was by jumpers if its possible), so you can try a known good VGA card?