Dodgy capacitors time constant circuits beware

[tardishead]

I am building a 670 clone at the moment and I had a nightmare with some distortion in limiting.
After literally months of troubleshooting I identified the problem.
The main time constant cap was very leaky and was letting the program material bleed through into the control voltage. This was particularly bad from bass up to 2khz.
I replaced it and bingo distortion went away
I was literally going to scrap the project after spending so much time and money.
The cap in question was a 1uf MKT 250v yellow poly film.
I never considered a brand new cap was faulty and looked at everything else first.
Big mistake I will never assume this again.

And then I looked at an old federal clone I had built that never worked out. I thought I had a oscillation/motorboating problem with it.
And it also had one of these mkt caps.
I replaced it and couldn't believe how good the unit was when I checked it out.

Wow what a royal p.i.t.a
So can someone explain what exactly this problem is
Also I tried to look for info on what type of caps to use in timing constant circuits. What are the best?
And is there any meter out there that will diagnose leaky capacitors preferably in circuit.

 

[CJ]

congrats on a tough one!  ;D

nothing is wasted as you learn a ton with this type of struggle,

here is a cheap and dirty leak checker, you have to lift the grid side>

 

[soapfoot]

wow, what a PITA!  Congrats on getting it sorted.

Now I'm giving the side-eye to the SA-39b with old wax paper caps that has always distorted since we've had it.

 

[Kingston]

Nice find. Never come across this myself but good to keep in mind.

 

[emrr]

I have seen this in restorations.  Take the original cap out, can't detect a thing wrong with it. 

 

[JohnRoberts]

That's a tough one. Generally you expect new parts to be good but stuff happens. For trouble shooting we need to suspect everything.

JR

 

[tardishead]

yeh forgot to mention I did take it out and measure it at least once maybe twice and it measured 1uf on my multimeter fine

 

[JohnRoberts]

I ran into a faulty production run of electrolytic caps back when I worked at Peavey. It was kind of lucky when one of the warehouse QA inspectors was doing a random test of some parts for tolerance and got an outlier. They were something like 1000uF caps and one measured several nanoFarads. He called me to check if that was OK....    Generally design engineers do not get involved in warehouse process issues, but I spent a lot of time in that factory so when in doubt they would like to ask me. I took several caps apart and noticed that the swage where the foil was attached to one of the cap leads was faulty. Apparently the tooling on the capacitor factory production line was broken and even the caps that measured good were not very well attached and would fail in the field after thermal cycling.

By this time we had already used thousands of these caps in our products that had passed our factory QA and they had gone to the warehouse to ship. We were able to identify which products were built with this one batch of capacitors to rework. It was a common PS cap so used in lots of models. I even had to recall one container that was on it's way out of the country with these time bomb caps in the product power supplies. Luckily I was able to nip it in the bud, Many of those units if allowed to ship would ended up back in service before their time.

JR

 

 

[ruffrecords]

The most common fault in capacitors of these types is faulty marking. It is easy to print 103 instead of 102 but you have now just marked the cap value ten times bigger that it really is and a low value cap would give exactly the symptoms you describe. Did you measure the actual value of the cap you took out?

Cheers

Ian

 

[tardishead]

yes I did mention I measured it and it was fine

 

[emrr]

Every old one in a limiter that was failing tested fine for value.  I can't speak to leakage. 
I have seen other old originals that worked fine, but the value moved constantly, never settling. 
I speak of film caps in these instances.