HP 3312A cap replacement

[petieken]

Hello,

Last week my HP 3312A function generator died while I was using it to test some filters I'm learning about. At first I thought I goofed up and overloaded the output stage.

After opening up the device today I checked the voltages and all where ok except the -17V rail which was sitting at -0.6V. The LM320 regulator was burning hot, so obviously there was a short somewhere on the -17V rail to ground. Finger test didn't give me any clues as no other components were hot or warm. Then I did a diode test for all the semiconductors connected to -17V rail, all of them were ok. Next step was checking the caps which were connected directly between -17V and GND by lifting one leg. Luckily after lifting 3 legs I found the shorted tantalum.

My question is about the replacement capacitor, the original HP part is an axial 2.2uF +/-10% 20V. What kind of capacitor would be recommended as a replacement? Should I use tantalum again? What was the reason for HP to choose tantalum over normal electrolytic caps when they designed this FG?

Cheers

[FlyingHacker]

I would guess they used tantalum due to how stable they are in capacitance given some AC component in the circuit. Maybe to limit microphonic effects too (vs. ceramic)? 

[SeanB]

Put in the same value, 35V unit. Or use a 4u7 35V unit. Tantalum used because it is reasonably stable, low ESR and self resonance is very broad and not a very sharp peak. Also will not leak like an electrolytic. If you are not happy using Ta, use a 105C 2u2 or 4u7 100V low ESR electrolytic, Panasonic or Nichicon, in place of it. The capacitor will probably do another few decades there.

[petieken]

Thanks guys.
I was wondering about the tantalum choice because I sometimes hear Dave complaining about them in his teardowns. But now I can see the reasoning why HP chose Ta in their design. I'll order a couple Ta caps.

I was wondering if it's common for tantalum to fail as a short?

[Cubdriver]

My understanding is that tantalum caps are very sensitive to overvoltage and that it's best to have a considerable safety factor as far as the voltage rating goes - in the neighborhood of a cap voltage rating of 2x the expected working voltage seems to be the rule of thumb from what I've seen and read.  With that in mind, a 20V cap on a 17V rail is cutting it pretty close, and could well have contributed to its demise.

ETA - yes, their common failure mode is to stop being a capacitor and instead become a wire (and possibly a match shortly thereafter).

-Pat

[petieken]

I was also thinking that 20V is really close, usually when I replace (electrolytic) caps I always make sure I have a safe margin on the voltage rating. The FG is ~35 years old I guess, so I can't complain.

I've ordered some 35V tantalums to replace the broken one. I'm happy it isn't one of those TO-99 packages that died.

[icy.dev]

RESURRECTION ALERT

Had same issue with a failing -17v rail.

Unit had some issues and prior owner hacked the rails while troubleshooting...only to put one of the 2.2uf Tantalums in BACKWARDS! (C232)...and like an idiot I blew 2 caps following their lead instead of the schematic!!

2¢ advice:  USE THE F$&@ING SCHEMATIC when replacing parts on a used piece of gear!!!