Fun with cheap Chinese tantalum capacitors

[peteski]

Reading this makes me wonder about the "end-of-life" failures of the caps that survive the "burn-in" test.

In the way of background, many manufafctured items have what is called a "bath tub curve" shape to their failure rates over time.  That is, the failure rate starts pretty high, but goes down quickly with time-of-use, reaching a low, but  non-zero rate after what is usually selected as the "burn-in" time for a particular type of item.  That low rate may stay almost flat or may gradually increase over time until it starts increasing again at a substantial rate.  The plotted curve looks like a bath tub, hense the name.  The point at which the curve starts to increase at a substantial rate is usually picked as the end of service life, when the part should be replaced pre-emptively to avoid failures in-service.

So far, this discussion has only addressed the burn-in aspect.  What is the failure rate of those caps that have survived the burn-in?  What is the expected service life for 95% of the (successfully burned-in) caps?  Given the failure modes of these caps and the likely effect on a valuable model, it seems like it would be worth knowing these in-service risks before using cheap caps.

You are correct about the bathtub curve.  Here we are just dealing with the "infant mortality" part of the curve.

If you go to any of the electronic distributor's or manufacturer's website and read the technical specs for tantalum caps similar to what we uses, you should see the average life defined.  It will be many thousands of hours.  These caps are used in many consumer electronic devices (including DCC decoders). So that should give you an idea of how long they are expected to last: good enough to be used in those devices and give years of reliable service.

[Maletrain]

Peteski, I understand the specs for electronic devices supplied by manufacturers for their products.  But, I thought this thread was about "cheap" caps supplied by "somebody" in China who makes caps that don't seem to measure-up to the quality of other caps from reputable sources.  I would not expect their caps to necessarily match the failure rates and normal service lives of the manufacturers who are providing such specs.

So, what I was asking is basically whether surviving a burn-in test is adequate insurance that the cheap caps won't still fail inside one of our models and destroy it sometime in our model's life. 

[MK]

So instead of a normal bath tub, we're talking about a bath tub for midgets.  . I would think that's certainly possible as I'm sure the quality is comprised somewhere due to the cheap prices.  But will they still out last MRR life times?

[Point353]

So, what I was asking is basically whether surviving a burn-in test is adequate insurance that the cheap caps won't still fail inside one of our models and destroy it sometime in our model's life.

Describe the burn-in test you are proposing to use.
Include applied voltage, operating temperature and length of time.

[peteski]

Peteski, I understand the specs for electronic devices supplied by manufacturers for their products.  But, I thought this thread was about "cheap" caps supplied by "somebody" in China who makes caps that don't seem to measure-up to the quality of other caps from reputable sources.  I would not expect their caps to necessarily match the failure rates and normal service lives of the manufacturers who are providing such specs.

So, what I was asking is basically whether surviving a burn-in test is adequate insurance that the cheap caps won't still fail inside one of our models and destroy it sometime in our model's life.

Even the "brand name" caps can fail during normal operation.  There could also be external causes of delayed failure (like a voltage spike form the track power).  No cap will be guaranteed not to fail during the model's lifespan. Whether the initial burn-in of the generic caps is adequate or not, is a good question to which I do not have an answer. I'm also not planning on doing any extensive testing. Before I recommended this initial burn-in procedure, modelers (here) were installing those caps without any testing. Performing the burn-in already saved them some headaches.  But there is no guarantee that any burn-in procedure will totally eliminate the possible failures later during operation.  I also suspect that if the material and the production methods are sub-par, those caps will fail more frequently than the better quality caps. 

[Steveruger45]

Well, I got my 100uF x 20v poly tantalum caps and have installed four in my latest project.
See post on loksound in kato C30-7.

As we should keep these caps sealed, they come in a ziplock bag sealed inside an electrostatic bag with moisture card, I will only test ones I am about to use.

I put a household alarm system 12v battery (potential was measured at 12.18v) across the four caps I used and all was fine after 2 hours.  These are from digikey and made by Panasonic. 

Cost of these caps is reasonable at around $1.50 each in lots of 10.

Steve

[Iain]

One of these days I am going to Shenzhen.