Fluke 81 tantalum caps issue

[giosif]

Hi,

I have a rather strange issue with my Fluke 81 function generator whereas, after keeping it turned off for a long period of time (e.g. months, sometimes even years), when first turning it back on, a tantalum cap develops issues (usually, it burns).
This happened 3 times so far and, up to now, each time it was a different cap (as in, in different location, on a different board and on a different power rail).
This time, it happened with a cap on the +15V rail where, after a few minutes of running fine, it would start drawing too much current and get really hot.
At that point in time, the display on the FG would turn off, but the fan would stay on.
In this instance, I managed to turn off the FG on time, before the cap had a chance to burn.
I then took the specific board where the cap was located (i.e. the current generator board) out of the FG and powered on just the +15V rail on that board from a bench power supply.
Sure enough, after about 30 seconds or so, the power supply went in constant current mode (current limit was set to 1A, I think) and C36 got really hot.
So, I removed that capacitor and replaced it with a similarly specced electrolytic cap and then installed the board back in the FG.
Now, everything seems to be working fine and the +15V rail is spot on and solid (monitoring with a scope).

Ok, so up to now, this would indicate an issue with the tantalum cap.
However, the interesting part is that, with the cap now removed from the board, I powered just the cap from my bench power supply, up to 20VDC.
And, the cap is not getting hot and seems to be fine (current reading on the power supply at 20VDC was 2mA - not sure if that is indicative of a leakage problem on the cap).
I also checked the cap with my LCR meter and it appears to be good.

So, right now, I am at a loss on what is causing this strange behavior, since the power rails seem to be good and without ripple (I understand tantalums don't like ripple).
Also, checking the larger electrolytic caps for the main power rails, they look fine from a capacitance and ESR perspective (so, again, unlikely to have ripple on the power rails).

Anyone any ideas what might be causing this and/or where to look next?

Thanks!

[bdunham7]

Anyone any ideas what might be causing this and/or where to look next?

The tantalum cap develops an internal microscopic short of some kind and eventually burns up.  You interrrupt the process, and by unsoldering and handling it you also disturb the microscopic flaw that is shorting it and it works again--for now.  I don't know the exact mechanics of the process whereby tantalums go bad like this, but they do.  The thing to do next is to simply replace all of them with higher voltage rating new tantalums or electrolytics.  Small electrolytics will work in place of tantalums in many, but not all, applications.

[floobydust]

Some equipment with dual rails will push a rail to the opposite polarity briefly during power-down or power up.
Example +/-15V rails, during power off each rail falls at it's own rate depending on their load and capacitance present. Some loads are not to GND but from between +15V to -15V. If -15V collapses first to say 0V, current from the +15V rail will push the -15V rail +ve and give reverse polarity to things, just until the 15V rail also collapses to zero.
I had problems with tantalums shorting on RS-232 drivers and traced it down to the power off sequence reverse polarity blip I believe causing damage to the tantalums.

What I do is put a reverse diode at the output of each Vreg to prevent this, very important. It also prevents 7915 latch-up if it comes up late, outmuscled by 7815 the rail can sit there stuck with reverse polarity.

[giosif]

Thank you both for your inputs!

On the tantalums developing internal shorts, that could explain the fact that it tends to happen only after a long period of time of not using the device.
And I am aware I should not replace tantalums with electrolytics willy-nilly; in this case, the cap was between the power rail and ground, so I think it is ok.

On the power-on / power-off timing difference between the different rails, it is something I never thought of, but it makes sense it can happen.
I will try to measure the different rails and see if I spot this type of issue.

[mclute0]

MLCC would be a better choice, but you don't mention the F unless my old eyes missed it. Multiple random tantalums buring without some other cause doesn't smell right to me, IMHO.

This graph says most of what you need to know.



from the sales blabber on...

https://product.tdk.com/en/techlibrary/solutionguide/mlcc_replace-guide.html


the biggest risk is unintended resonances but that is probably easy to measure in your function outputs.

Remember they are "tantalum electrolytic capacitor" and have some of the same issues as aluminum electrolytic capacitors.

You would be best sending an email to Fluke support. The device may be out of warranty, but they are usually willing to answer questions if they reasonably can.

ps: I would be looking at any voltage regulator IC or diodes shorting closed.
 

[bdunham7]

What I do is put a reverse diode at the output of each Vreg to prevent this, very important. It also prevents 7915 latch-up if it comes up late, outmuscled by 7815 the rail can sit there stuck with reverse polarity.

OK, so now I'm rethinking bulk tantalum replacements for power rail decoupling applications.  The typical method is to replace them with higher-voltage parts where possible.  This document was posted in another unrelated discussion:

https://www.navsea.navy.mil/Home/Warfare-Centers/NSWC-Crane/Resources/SD-18/Products/Capacitors/Derating/

They state that dipped tantalums are the most reliable type of capacitor with the caveat that they cannot tolerate more the 2% (of the forward voltage rating) reverse bias.  In many cases, devices already have reverse voltage protection on those power rails, but since 2% of 35V is 0.7V, it would seem that even a diode can't properly protect a tantalum cap with a voltage rating under 35V.  So by your theory of tantalum destruction, the thing to do is always use parts rated for 35V or more and probably replace most of those large capacity low voltage units (i.e. 150uF 6.8V) with aluminum electrolytic. 

[floobydust]

I know tantalums as not being tolerant of any reverse-voltage. Electrolytics I believe up to -1V to -1.5V (short-term) does not affect them.
You could use Schottky clamp-diodes across the rails instead to limit it to -0.3V which is below most half-cell potentials.

For OP, if the main filter capacitors have failed open or low value, the smaller tantalums will experience high ripple current which will quickly heat them up and cause them to fail. I've had lots of tantalums burn up when the SMPS output caps went bad. OP should check the power supply's health.

Analog gear, I like using tantalums on rail decoupling (voltage derated 50%, log for lifetime) because of their moderate ESR which does not add bus ringing unlike high-Q MLCC which can make a mess and are also a rip off you're lucky to get 1/2 their rated capacitance with DC bias and ageing. Electrolytics don't have the HF coverage.

Are these wet tantalums here in the Fluke 81? Or dipped.