Digitrax SDN136PS decoders overheating

[bobthebear]

Hi. I have several of these decoders working well. I have recently installed 3 in Con-cor U50 chassis, fitted with 8ohm sugar cube speakers and 4 tantalum 22omf 16V capacitors. Stall ampage was checked at an average of 0.6amps. I have had all 3 get VERY hot, at completely different times, and then fail. The latest one to fail ran on our club layout all day, then for 5 minutes at home, before it died. It's not one particular chassis or it would point to a short, and they have been thoroughly checked anyway.
Each time one blows, I replace it with a non sound decoder and it's fine. Any suggestions?
Cheers, Bob.

[nscaler711]

I wonder if its because you are using the wrong ohm speaker... Seen that happen in a cell phone when I replaced speakers in headphones....

[bobthebear]

Hi.  8ohm is correct, as the Digitrax fitted one is 8ohm.
Bob.

[C855B]

I don't know this particular decoder, Bob, but I do know the U50 chassis pretty well, for obvious reasons. What is clear is there has to be an issue with the installation technique common to the three.

First, are they unmodified? That is, given the "three" you mention, they don't happen to be James' C855s, do they? The reason I ask is with the chassis splice, if you haven't jumpered the front and back halves together at the chassis, you could be dealing with intermittents stressing the keep-alive. Is it possible the keep-alive capacitor is (electrically) touching the chassis?

Is it possible the different speaker than Digitrax' is electrically touching the chassis? IOW, an unexpected ground? Kapton tape between the speaker and chassis is worth a try.

How about the cam (gray) lead? Is it cut and insulated?

Generally, I would make sure that nothing about the decoder is in direct contact with the chassis. Not even the shrink tubing around the decoder module.

[bobthebear]

Hi Mike.
Yes, you've guessed correctly. They are the first C855 set.
But.......they have jumper cables, the speaker faces upwards, so the plactic box sits on the chassis, the grey lead has been removed completely, the decoder sits on a kapton bed, but floating for ventilation and the capacitors are covered in Kapton tape.
I could understand a problem if it was just one chassis, but all 3 worked, then at different times, the decoders gave out. I know sound decoders run hot, but these blistered my finger!
I have done well over a hundred installs with no problems. This has me beaten.

[C855B]

That's just bizarre, Bob. These are just too expensive to be failing like that after all the known and appropriate cares have been undertaken.

Will the decoders work with the sound harness unplugged? At this point it's time to throw caution to the wind and isolate it to sound vs. motor control.

Of course there's the obvious - we've installed these in model Alcos, and have selected the 251C library. The software is so good it's trying to emulate Alco locomotives' propensity to smoke.   

[bobthebear]

Can't beat a bit (or lot) of Alco smoke!
Well, for now I have 2 units with sound, and one with just a standard decoder. I was just too late to remove the speaker plug before the last one gave up the ghost, but I may give that a try.
Keep smiling!

[peteski]

How about you leave the shell off one of the misbehaving units and run it on the layout monitoring the temperature of the decoder.  If it gets hot (but before the decoder fries), let it cool down and mute the sound (or turn the volume down), then run it again.  if it still gets hot, disconnect the motor, turn the sound on again and let the loco sit on the track and "run" it with just the sound blasting.  See if that makes it hot too.

[C855B]

Hey, Bob... James told me you added several additional keep-alive capacitors beyond what was supplied with the encoder. Interesting idea and I fully understand the logic, but keeping the charge topped off on four high-value capacitors without current limiting might be more than the power management in the decoder can handle. Cut it back to "stock", then add the others one by one until the heating problem recurs.

You might also think about a diode and resistor (value = ???, probably > ~2K) in parallel to limit the inrush current for the extra capacitance while the diode allows unrestricted discharge in the opposite direction. That is, limit the charging but no limit on discharge, letting the decoder draw on its own terms.

[bobthebear]

Thanks Peteski and Mike.
I had cut the new tantalums down to 3 from the original 4 and had removed the stock capacitor anyway. I will remove another tantalum now. I will do some more tests when I return from my tour of Scotland.
Cheers, Bob.

[peteski]

This is the circuit used by companies like TCS in their keep-alive modules (more info in https://www.therailwire.net/forum/index.php?topic=37327.msg447874#msg447874 ).


But this is for the super-caps which have capacitance in hundreds of thousands of micro Farads.

For smaller caps (thousand micro-Farads or less), ZIMO decoder manuals simply install a 100 uH choke in series with the capacitor. The choke naturally impedes fast changes in current, therefore limiting the surge current going to  or leaving the capacitor.  That seems to be a simplest solution for limiting current.