Thermal Faults

Normally I would go around with a can of freezy spray and dob various parts of the board looking for the fault to go away, but sadly in this case it's not possible as the board in question is one of these slot-in types that are inaccessible to investigation when the instrument is under power. :(

Anything.

NT

Decades ago there was always a standard extension board that plugged into the motherboard in place of the questionable one and carried only a socket that the one under investigation could be plugged into to make it accessible, so long as the covers were off...

Mike.

Yes, and the service engineers would no doubt have been issued with them. The service engineers would also no doubt have been issued with duplicate sets of identical boards to swap-out, thereby saving heaps of their very expensive time on each job. This particular instrument is a 10Mhz-22Ghz spectrum analyser (one of two I have made by Hewlett-Packard) so those engineers sent out into the field to fix them would have been very well-supported by HP. Unfortunately, however, I'm not one of them! However, the fault appears to be somewhere in the x-amplifier board and they used *exact* same board for the y-amplifier so at least I can compare them. There's no point now in swapping them over as I now *know* the fault lies somewhere in the x- amp one.

Not much to go on.

Age of board? What does it do? Digital?

Old caps can go leaky losing capacitance with temperature. If they were decoupling something from interference ...

I'd scope out the power supply rails with a DSO, maybe you might catch something. Can you arrange a trigger when the fault occurs?

Lets assume you have the schematics and a bench/lab power supply and you know what goes wrong when the board is warm. Then why not put the board on the bench powered up by your power supply and check the parameters of the circuit whose change can lead to the symptom you see? Check the voltages when just turned on, then heat it up and see what changes.

You don't need an extension card, although they are very handy - I have a dozen or so - however if the card connections are more-or-less standard you can make your own using a plug (hacked from a dead PCB perhaps) and socket and some wire. These will usually work, just usually aren't very good at 100khz or higher frequencies...

It all depends on how badly you want to fix this and what your skills are...

John :-#)#

Quite.

Precisely what I am about to do next!

It's now so recurrent that no special triggering is needed. At the pre- driver stage (the output from the sweep gen stage) I see a nice clean saw- tooth waveform but 3 stages later I see the same trace; same amplitude but frizzled by severe noise and twitching like mad. So I'm closing in on it but am hampered by the fact that the board is inaccessible when powered up since it sits in a slot alongside other boards.

Tack on a bunch of wires to points you'd like to scope. Install board, scope those points, move on. It sounds like you're very close to finding the culprit.

NT

Yup, in fact that's what I had to do to find the excessive noise I mentioned in an earlier post. I hadn't expected to find it; I wasn't looking for noise, but I couldn't very well ignore it given it represented about 50% of the waveform amplitude!

Hopefully. But this unit has multiple issues and this is just one of them. Fortunately for me this is just for fun. If I had to do this for a living I wouldn't survive very long due to the time I take over these things it simply wouldn't be economic.

Anyway, time now to get on and scope those power rails....

As already noted, many things *can* get hot. What is in question is how hea t will affect various components. In no particular order:

a) Carbon comp/film resistors will change value over time, and much more-so of heated. Some types are so awful that they will change in value simply f rom the heat required to install them. Given the alternatives, the only rea son to use them today would be based on audiophoolery.

b) Electrolytic capacitors do not like heat at all. Sure, the come rated to 105 C. and more - but none-the-less, they continue not to like heat. In ge neral, if an electrolytic cap in a piece of solid-state equipment gets warm er than ambient temperatures, replace it!

c) Cold-solder joints will get hot based on the current they may be carryin g. And, in general, there will be some discoloration around the bad joint, or some other visible indication. Most especially if this bad joint is of l ong standing.

d) Broken/lifted/oxidized traces are very often heat related. And in the la st stages before complete failure they may become intermittent, giving you the symptoms you hear/see.

e) If, by any chance there are small IF cans on/near this board, or within this device, silver-mica disease will create thunderstorms and breathing sy mptoms as the equipment heats up.

If you have access to an IR camera, try running that board until the sympto ms are well-established, then pull it and photograph it. A hot component wi ll stand out as a bright white blob. A friend of mine is a hobby photograph er and has IR equipment. One day, he decided to use it to troubleshoot a pi ece of electronics. Within a few minutes he found the problem - that was o therwise entirely invisible under normal light.

Peter Wieck Melrose Park, PA

eat will affect various components. In no particular order:

so of heated. Some types are so awful that they will change in value simply from the heat required to install them. Given the alternatives, the only r eason to use them today would be based on audiophoolery.

We use carbon film aplenty. The reasons are low cost & good availability. Carbon comp is used on occasion for its pulse power tolerance.

NT

In other words, kick the can down the road....

Peter Wieck Melrose Park, PA

it's not clear what issue you're referring to there.

NT

The fact that carbon resistors have a pretty wretched service life and heat tolerance.

Peter Wieck Melrose Park, PA

you're clearly misinformed

heat tolerance.

Vintage is the key word there. Ancient carbon composition are indeed prone to rising in value & going oc. However carbon film Rs have an excellent rel iability record. I've repaired lots of stuff over however many years, and c arbon film Rs are almost never a problem.

I've no doubt your ego will not permit you to get real, so I see little poi nt continuing this. Others who read this will make up their own minds.

NT

Do you even pay attention to what you write?

"Carbon comp is used on occasion for its pulse power tolerance."

Which is the worst possible way of using a carbon composition type of resistor.

Peter told you that. I told you that.

Why is that? This is IMHO not about the carbon film type but about modern massive carbon composition. I've used them for > 40 years in e.g. triac snubber networks and they never failed. All kinds of (same wattage rating) film types did fail. Wirewound is also OK but too expensive.

Arie