Tant trauma (again)

So a circuit that had been tested. And now failed. The tant. (35V 10 uF Kemet) was in the right way (and a now a dead short). 15V power supply. It's the back half of an LC low pass.. 100uH. Could I be getting some ringing on the inductor that over voltages the cap. (I'll have to look.) There's 100+ of these in the field and no reported failures. Maybe one marginal cap? Should I just use 50V tants on all the 15V power rails?

Grumble, George H.

Reply to
George Herold
Loading thread data ...

There's a little ringing, peaks at ~22V for ~300us, Q ~2 GH

Reply to
George Herold

Hot plugging or no?

Q > 1 sounds like a poor supply filter design. What did your ESR calculation look like? Did you consider ceramic + R instead?

Good old electrolytics can be very reliable, if not ran hot. They don't fail shorted under surge conditions, or self-heal in transients.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

I had a tantalum (orange drop) go bad on a power supply, but there were three or four sections all with tantalums, only one failed (after a year); probably it's not a design issue, just a component's random failure.

The silver-tantalum mil spec variants are said to be better, and there's a small surplus stash of those in a drawer somewhere, but price prevents me ever using 'em. The horror, the horror: two of 'em that I've employed to save space in a power supply have replacement cost $83 each. Comes down to $78 in tensies...

Reply to
whit3rd

Does your tantalum usage have the series resistance that limits inrush? If not, it should be redesigned before you build any more. A mix of ceramics and aluminums?

About 2 years ago I had a Tek 7A26 fail - one of its tantalums shorted many years after it left the factory. I don't recall what the rating was; I think it was on the 15V rail.

Regarding what's in the field: this is hard. Is power limited enough to be confident that there's no safety or fire hazard if one of these fails?

-F

Reply to
Frank Miles

That's a peak current around 0.75 amps. Tantalums don't like current: it detonates them.

I don't use tantalums on power rails if I can help it. If so, derate voltage 3:1.

Sometimes a tantalum has just the right ESR.

They are erratic too, in batches.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

They can have longterm acid leaks too.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Toggle switch

Design? I'm afraid I did no calc's beyond the LC corner freq. Most likely I looked at the same step response and figured 'good enough', The Q of 2 adds some gain near the corner*...

I mostly wanted something to knock down SPMS spikes. ~100k Hz, I recall first trying open ended inductors and being appalled at the pickup.

Right, I should see if we can fit in some Al electro's. I guess I've always been (perhaps needlessly) worried about the lifetime 'drying out' but if used with hardly any ripple, will they last ~20 years.

George H.

*airing dirty laundry... This was for a low noise power supply, my big error was picking the wrong voltage reference. But I was younger then, and knew less of noise.
Reply to
George Herold

At that price I'd get out a roll of paper & ali foil. Some chip fat ought to keep the damp out :)

NT

Reply to
tabbypurr

Huh, OK. I didn't know. (I'm mostly an idiot, I'm sometimes amazed that I can make things work. :^)

OK so derating is good in the short term. but I wonder if after a while the suppliers learn of your derating, and turn down the pass knob. (A few months ago, I repaired an old RF amp we made years ago. Power rail was a dead short and I traced it down to one 16V tant on a 15V rail. There's still ~10 16V tants left on the thing.

Reply to
George Herold

formatting link

Cheers

Phil Hobbs

(Bored in Incheon Airport)

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Right I read that, there's ~no ripple here, but it does get blasted when turned on. (Incheon? long flight home.)

George h.

Reply to
George Herold

Interesting read, thank you.

--
Reinhardt
Reply to
Reinhardt Behm

ESR is critical, and if you want good HF attenuation too, a smaller low-ESR cap in parallel is needed. Indeed, the bulk cap serves more to dampen things, than to attenuate things (unless you don't need much attenuation to start with, in which case that's fine).

Nice thing about tants is they're sold in every combination of C and ESR. They can be low ESR, but are usually modest.

Polymers too can be medium or high ESR, but are usually low. There's a lot of space to shop around!

An Fc of 5kHz is really low just to "knock down" spikes -- also, if it's the spikes you're worried about, the frequency content will be in the many MHz, and may include common mode that's not going to be fixed with the normal-mode filter.

A higher Fc may be desirable so as not to affect the controller -- the filter's resonance tugs on the controller, in your case probably putting a dip in its passband. That might not be a problem, but it's... awkward?

Cheers, Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

OK cancel all of that! Operator error. The power plug was wired backwards, and that blew the tant.. funny enough. The rest of the day was wasted as I tracked down the wrong problem.

George H.

Reply to
George Herold

HaHaHa! The tantalum saved the rest of the circuit by self-destructing.

To avoid this problem in the future, add a diode in series with the positive lead. Anytime you connect to something external, you need some kind of protection. ESD, miswires, improper supply voltage, and so on. Trust but verify.

Reply to
Steve Wilson

Right I've got a little plug (led, R and zener) that I use to test all power plugs. I forgot to do that. I'm guessing I won't forget again.

George H.

Reply to
George Herold

The problem is if you give customers access to the internal circuitry, you give them the ability to destroy the device. Some well-intentioned but ignorant clutz can say "I can fix that" and proceed to wreak havoc on your design. Another example is someone stepping on a cable and ripping a connector out of a pcb. The answer is to have a bulkhead connector to take the strain, with a separate interface cable to the pcb. The bulkhead connector is cheaper than the pcb and less likely to be destroyed.

I learned early in my career that things that are not supposed to happen are guaranteed to occur.

But you already know this.

Reply to
Steve Wilson

We usually have a tanszorb diode on power rails that come from external warts, in case the voltage or polarity are wrong. Worst case, the transzorb dies shorted, and saves the downstream stuff.

Polyfuses suck. Efuses seem to suck. A real fuse and a transzorb seems to be the best self-recovering protection.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

PolyZens rock. They cost a buck, but they work great.

Cheers

Phil hobbs

Reply to
pcdhobbs

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.