Been there...

Now you know why some posters here nag about keeping voltage below 75% of the rated value. Also beware antiques (the voltage rating has probably dropped, plus they may contain PCPs), dodgy sources ("untested" = tested and failed) and badly regulated power supplies.

Believe it or not, I have seen this happen in comercial products from respected companies.

And, as per the recommended search, tantalums have their own set of problems.

pqe

WOW

That's scary stuff

easy to see the problem....

"... tantalums have their own set of problems."

The primary failure mode for tantalum capacitors is dielectric breakdown due to over voltage including voltage spikes. If you ever get to look at the raw DC on a decoder you will be surprised by the spikes generated by the DCC and motor drive even when a large capacitor bank (Keep alive) is installed. It is generally ill advised to use tantalums in motor driver and similar circuits.

This is probably what cause this meltdown. If I am reading the markings correctly the parts you used are AVX TAJ A477C parts. The C means that the parts are specified for 16 V which does not provide much over voltage margin in a DCC / motor drive environment.

Fire hot.

Well I won’t be doing this now!   Just bought the stuff to build one.   No way!  

Before some of you swear off home aid keep alives

Use the right components to start with, if you don't know, ask.

Keep alives

I've said this once, but here it goes again.

A single 1000mfd 25volt electrolytic capacitor will do the job in most cases.

You don't need a resistor or diodes, just the capacitor, 

The engine won't run for another 10 seconds without power, but why would you want that?

And if you use a metal can electrolytic they don't burn, they pop. 

More exciting, but no flame.

Clark Bauman

Specs

Indeed, they are 477uF 16V tantalum capacitors. Well, they were ...

At least according to the documentation of the Zimo MX645P22 decoder it should have been ok. They say:

(their emphasis)

An indeed, measuring the pins there I had a 15.3 V max. So at least on the face of it I was within the specs. But I guess mixing 10 cheap ebay pieces is not taken into account by this.

Cheers,

.costin

Re: Specs

Costin,

Not to beat a dead horse here, but please refer back to what Pelsea and Logger01 wrote. Just because you hooked a multimeter to the rails and read 13.5V doesn't mean there wasn't higher voltage spikes on the rails. A multimeter can't measure these high frequency spikes - it never was designed to do so.

I vaguely remember tantalum capacitors being a no-no in circuits from many years ago, but forgot why. This discussion sheds a little more light on it. A Google search as recommended earlier, may help explain it better, but then again, maybe not if you're not well-versed in electronics and electrical subjects.

It sounds like the documentation with the decoder led you astray. Having more voltage margin is always a good thing. 

Regards,

Joe W.

DIY Keep Alives

I definitely do not want to push anyone away from making their own Keep Alives. I just want DIYers to use the appropriate components. In fact I have built and installed dozens of home made power backup capacitor banks for HO, S, O and Large Scale engines, but I also have been designing electronic devices for too many decades.

The appeal of tantalum caps

This is why you need margin on your capacitors

true, but

This decoder is supposed to charge it at under 16V at all times, disconnected from the direct DCC input (the supply is 18.5 anyway, no other boosters, very simple setup). In fact the main board already has a 16V 680uF electrolytic capacitor installed for this purpose but it doesn't manage to cruise long enough in case of glitches. The main board does essentially this (this is the board model, just a different capacitor deployed):

And if you look at their own offering of keep alive components they are all 16V for the 645 model (even 15V for the "supercap"...).

I still think that adding the tantalum capacitors in parallel to the electrolytic one should have worked, baring not enough knowledge of what complications this setup would have.

And guys, don't be negative, how else would we learn anything if we wouldn't experiment? Doing only what you know best how to do wouldn't have many of us around in this hobby (me for sure) After all, what better way of blowing stuff up than testing in production?

Cheers,

.costin

ZIMO Decoders - external energy sources

The ZIMO family of decoders which include the MX645 regulate the charging voltage to external energy source[s] (capcitors) to 16 Volts. This does not leave any margin, but the circuitry does a fairly good job at preventing and or reducing over voltage events. For decoders without the control circuit ZIMO recommends using 25 and 35 Volt capacitors (See page 57-58).

Re: ZIMO Decoders - external energy sources

Hi Ken,

Yes, that is the one I have (MX645), this is why I took for granted the < =16V value.

More searching led me to finding out about the voltage derating concept for tantalum capacitors. There are many resources with various values for it, from 50% to 80% to ~nothing. So while there is no consensus on how much, it's clear to me now that there should be at least some, with more dramatically reducing the failure rates.

And another factor that affects them is heat, my shrink wrapping might have damaged them enough to cause two of them to go up in smoke (the first test was done without it and everything was ok). Ok, next time I'll make sure to get significantly larger voltage ratings than the nominal value I need and be very careful on how I work with them. And no more shrink wrapping

Cheers,

.costin

Normal Design Standards

It's not quite right to say that there is no consensus about the voltage derating of capacitors. Normal US, UK and European electronic design standards state that capacitors should not be operated at greater than 75% of their rated voltage while, in high-reliability situations such as defence and aerospace (where I spent the bulk of my career as a professional design engineer) the specified derating factor is 50%.

I would not consider using capacitors rated at less than 25 volts in a DCC environment - and would prefer 35 volt components where size permitted. I would also recommend using standard electrolytic capacitors for a keep-alive application rather than tantalum which have much poorer thermal and leakage characteristics (but perhaps that is just my professional side talking over the hobbyist's shoulder . . . ).