For the want of a diode, a PT dies

[Randall]

1966 Princeton Reverb comes in with a dead PT. I had just replaced it about 3 years ago with a Marvel from CEDist. At the time I also replaced filter caps and 6V6s, so I'm wondering why it died again. I installed a Mag Components that I got from Tube Depot. (Both Mag Comp and Marvel are now out of business sadly). It powered up good on my limiter, so I plugged it in the outlet. My B+ was low, and I could smell the tubes. Indeed, the PT was vibrating a bit and getting hot. It was then I realized the tubes were pulling 95mA a side! Yikes.

I pulled the 6V6s, and quickly found that I only had -5v bias voltage, well no wonder. Last time around I changed out the bias cap ( I like 100@100v), and adjusted the 100K resistor down to 82K to get the bias closer to where I wanted it, so I know it was working correctly when left here before. Now comes the puzzler, the 27K, the diode, and the cap all measure good out of circuit. I changed out the cap again, because that is what I suspected first, but it made no difference. I threw in a 1N2007 diode in it, even though the old one tests good, and voila! Now I have too much bias voltage, so I went back to a 100K resistor, which puts it about right for this PT.

So, I wonder, if this diode failed to the point of all but killing the bias supply, how come it tests good?

And boy am I glad I caught this, I would have been very embarrassed when this second PT died, and I am sure it would have.

https://irationaudio.files.wordpress...verb_aa764.gif

[nosaj]

Because you probably tested it with a DVM which I think is just a volt or 2.
Think about it like this You've got a tire that rolls fine when you are going slow. But when you take it up to 55mph the car shimmys like hell until something breaks. Just an analogy whether its good or not is another question.

nosaj

[The Dude]

My guess is that the diode is leaky. If you still have it, check the resistance both ways on a high resistance setting instead of diode check. It should be at least 10 times the resistance reading reverse biased as it is forward biased.

[g1]

Because you probably tested it with a DVM which I think is just a volt or 2.

And it does not apply any reverse voltage test either.

[Helmholtz]

I threw in a 1N2007 diode in it, even though the old one tests good...

Did you mean an 1N4002? If so it has too low a peak inverse voltage and is likely to develop partial reverse breakdown. I would at least use a 1N4005.

[nosaj]

For the price of those there really is no reason to buy anything less than the 1n4007 for most applications.

nosaj

[The Dude]

I was assuming the '2' was a typo and that he meant 1N4007.

[drewl]

Seen that before.
earlier this year I finally used up the 100 4007's I ordered like ten years ago.

So I ordered another 100.

They start failing in older Dual Rectifiers even though they measure good with a DMM.

[Randall]

Sorry, I meant 1N4007.

My meter Fluke is autoranging. The diode measures 4.1 M ohm in one direcrion, and 248K ohm in the other.

[The Dude]

That's more than 10 times the resistance, although both of those numbers seem low. You can still compare those numbers to a known good diode. It could also be that your meter will test it fine and the diode fails in circuit, which you've already proved. At any rate, glad you found it!

Edit: FWIW and out of curiosity, I just checked a 1N4007 reverse biased on my 20M meter scale and got an infinite reading.

[Helmholtz]

The max. reverse current of a 1N4007 is specified as 5µA @25°C. This corresponds to a DCR of 200M at 1000V. At lower voltage the DCR can be expected to be even higher.

I recommend to only use the glass-passivated version which is much more reliable than the standard "plastic rectifier" type.
IIRC, the difference in failure rates is a factor of 20.

To test for reverse leakage, wire the diode in series with a 100k resistor and connect to a DCV source of several hundred volts (reverse direction: cathode to positive), e.g. the B+ of an amp. The reverse current is given by voltage drop across resistor divided by 100k. A voltage drop of 0.5V corresponds to a reverse current of 5µA.
(Principle is the same as testing ecap leakage.)
Of course the applied voltage must not exceed the component's rated voltage.