The impossible happens every day

[g1]

the drive channel didn't sound as distorted as it should on just one occasion when the footswitch was operated. The owner reports it never did it again but wants a full investigation to make sure it never happens in the future.

Sometimes with those sorts of issues I will simulate "brown out" conditions with the variac. Should the fault recur, at least you can say "sorry, can't guarantee that you won't run into poor AC conditions again"
When the first 6100 series "midi Marshall's" came out, guys were having problems with the channel switching locking up occasionally. Of course they always worked flawlessly in the shop until we started playing with the AC line voltage.
Much later folks came up with mods to improve the 5V supply, but originally we would tell the owners "what do you expect, it's got a computer in it, buy a real amp .
Not really, we'd tell them they needed UPS systems, like everyone else running digital gear that was locking up from brownouts.

[Jazz P Bass]

My heart sinks when I get those jobs. I have one right now - A Hotrod Deville where the drive channel didn't sound as distorted as it should on just one occasion when the footswitch was operated. The owner reports it never did it again but wants a full investigation to make sure it never happens in the future.

Hit it, poke it, push, pull, bang it on the bench, gentle pressure, vibration, inspection, measure, test, observe, play quiet, play loud. Nothing.

I bet the next time out it will go wrong.

Never Say Never.

[olddawg]

Crazy stuff happens all of the time. I have an A/B/Y switch imputing my pedal board because I use a talkbox frequently. The output to the pedal chain will only work without severely attenuating the signal with only one short cable I have. I don't know why, I can't measure or observe any difference with my other cables. At this point I say screw it, it works. Lol. If a customer had the same complaint I would be very skeptical. If it quits working I MAY figure it out!

[nickb]

An Impossible Thing Happened Today

The scenario. Marshal JVM215C making a very load assortment of popping, whooshing and burping type sounds.

Traced to the hi-voltage MOSFET based gyrator in the 'resonance' circuit. There's only half a dozen components in it and the FET is the likely culprit. Change FET and noises are still there. Bummer. Rather than keep yanking the PCB and since a few passives are as cheap as chips opt to replace all of them. The noises, of course, steadfastly remain. OK, isolate the gyrator from everything thing else. I can still see the &*(_+ noises using the 'scope, in particular I can see them on the gate. Replace the two gate bias resistors with 1% MF low noise jobs since I checked the zeners a moment before. Still the same. In desperation I change the FET again and there you go - peace and quiet.

I guess not impossible but the chances of getting a bad brand new FET has to be pretty darn slim...

[g1]

I think defective new parts is possibly the worst scenario there is.

[The Dude]

There's nothing more frustrating than chasing your tail over new parts.

[Enzo]

yet in tube circuits we preach all the time that "new" is not a guarantee of "good". it applies to all parts.

[g1]

Just that the sockets make it so much easier.
Stuff like transistors, Fet's, IC's are much more difficult to "try a known good one", plus it is more unlikely that anyone will accept a return.
While I see your point, I think with tubes a lot of it is because the quality control has gone down so much. I expect a much lower rate of "duds" with any other type of components (excluding counterfeit stuff).

[Enzo]

I wasn't thinking that so much as when you install a new part, and the circuit doesn't work, do not assume the PART is good just because it is new. ALWAYS consider that the new part COULD be involved in the lack of function. It is a corollary to my "never think up reasons not to test something" mantra.

[g1]

I hear ya . But it makes it so much easier if you can eliminate suspects. And some things can't really be tested, like this noisy Fet for instance. But I guess you mean test the circuit.
So I guess as far as the troubleshooting decision tree, you proceed like Nick did, on to the other parts of the circuit, then start over again.
Lather, rinse, repeat.

[nickb]

I think defective new parts is possibly the worst scenario there is.

Yes, it's bad. But intermittent's I despise/fear more that anything. Isn't rule #2 of debugging being able to reproduce the problem "make it fail" http://www.debuggingrules.com/debuggingrules.jpg

What's your strategy?

[Mick Bailey]

#1 Understand exactly what the problem is and the environment in which the fault occurs.
#2 Where possible, get the customer to replicate the fault.

[Tom Phillips]

A task of the season brings up another "impossible" situation.
When troubleshooting a 50 bulb Christmas tree light string I found no less than 9 bad bulbs before the string was made fully functional.
It took a while but I didn't let it beat me.

[Enzo]

perseverance is an important aspect.


I was watching Star Talk, with N.D.Tyson, and he was talking to Penn and Teller. Penn says magic is about working harder than someone thinks a trick would be worth.

[km6xz]

New part failures are more common than ever because so many difficult or expensive ICs and transistors are bought on eBay by techs who think they are getting a deal. Often it isn't and those $28 high power MOSFETS you got for $12 on eBay turn out to be remarked $2 MOSFETs. I got a "good deal" on a tray of ON Semi MJ15024 and MJ15025 and found out their were actually low power dies in a normal TO-3 case marked very convincing as the real thing. I seldom buy on eBay so it only happened to me once but I know shops which have been burned many times, the price is just too addicting to give up entirely.
Other than the forgeries, parts now are remarkably reliable as new to meet spec....except vacuum tubes which are worse than in the 50s and 60s.

Intermittent problems are the biggest time wasters for shops and most difficult to prove there is or is not a problem. Just by changing the policy in my shop when I was in California, of how units incoming were written up and verified as having a defect, resulted in about $4000 more /month in net profit. I instructed the girls at the counter to ask the owner to demonstrate the problem and set up a universal test cart with circuit breakers, variac, signal source etc. By watching the customer set up and use their unit, often they could spot user errors, particularly with keyboards and midi gear. If the customer could not duplicate it, by knowing exactly how he expected it to be set up, there was an excellent chance that there was no problem with the unit he brought in, might be elsewhere in the system. That was particularly true for anything related to home studios which has a lot of things in the signal path.
Often a user will say, "Check it out" as the only complaint when they are actually referring to a very specific set of routing or settings. This is really common with mixers. To completely test every combinations of possible routing can take a very long time, when what the customer really is complaining about "when channel 3 and 7 send pre-fader, Aux 3 is slightly lower than Aux 4". By eliminating ambiguous complaints or unverified intermittents from reaching the tech's benches, productivity increased dramatically. A unit on the bench with an unverified intermittent is the hardest hit on tech productivity there is because it takes a great deal of time to "prove" it works when it might not.