TSU-750 decoder - sound keeps cutting in and out

[Bart1701]

I am installing a TSU-750 decoder in an N-Gauge Kato FEF 4-8-4. The motor power is still provided by the TCS decoder in the boiler. The TSU-750 is in the tender and is only used for the sound effects.

I have the decoder installed, but have not added a capacitor yet (but will be doing soon). It's a pretty simple installation - plenty of room in this N-Gauge tender!

The two purple wires go to the contacts on the speaker, and the red/black wires are connected to the brass contact strips that run along each side of the tender base. To attach the red/black wires, I had to cut a notch in the plastic plate that sits on top of the contacts so that I could solder the wires to the contact strip.

Since the speaker sits where the lead weight used to be, I am using some press-on weights at the front and back of the tender.

When I set the tender on the track, the sound works just fine...but, a soon as the tender moves (even just a slight bit), the sound just cuts in and out constantly. Actually, what you hear is kind of a somewhat constant "crackling" sound as it moves along.

I have attached a few photos of the install in case that shows something that I am missing (the lead weights may not all be shown in the photos).





Thanks,
Bart

Any ideas what's causing this and how I can resolve it.

[peteski]

That capacitor (you haven't yet installed) keeps the decoder alive during power disconnects lasting fractions of a second.  I suspect that once you add the cap things will be better.  If not, you probably have intermittent power pickup from the rails

[jdcolombo]

Ditto Peteski.

Tsunami decoders are VERY sensitive to even momentary electrical contact dropouts.   You need to add the keeper capacitor; in fact, you might want to think about adding two 20v 220uf Tantalum chip caps in parallel (giving 440uf of total capacitance); two of these will take up about the same room as the round 220uf electrolytic cap that comes with the decoder.

Oh . . . and, clean your tender wheels.  They may look clean, but to be sure, clean them.  I use a wire brush in a Dremel, and take each wheel out, run the wire brush at moderate speed around the tire tread to polish it, then put the wheel back in the tender truck.

John C.

[Santa Fe Guy]

I have found that in some installations these decoders get very hot and an added heat sink solved the problem. Track and wheels were clean.
Rod.

[Bart1701]

Thanks for the advice!

I have some Tantalum caps on their way from Digi-Key. I did not have any more left after doing 2 other steam loco sound decoders.

I didn't realize that the caps came into play so often...I figured it was just when you once in a while would hit a momentary section of dirty or dead track and then the capacitor kicked in to provide power for that moment. I never thought that it would be kicking in as often as it is suggested. As a point of reference, my other sound-equipped steam locos with tantalum caps chug and puff along quite nicely throughout the whole layout without any issues with the sound dropping out. Once I get the caps added to my newest locomotive, I am sure it will be just fine.

...and now, what about the heat sink idea? Any other opinions on this, and can someone explain/show how you would add a heat sink?

Thanks,
Bart

[jdcolombo]

The TSU-750 does occasionally have overheating problems, although I haven't had that be an issue for the last several I bought over the past 2-3 years.  Overheating, however, is not "cutting in and out."  What will happen if a TSU overheats is that it will shut down completely for several minutes until the heat dissipates.   If by "cutting in and out" you mean that there is momentary sound loss, the problem is poor electrical conductivity, which can be cured via the keeper caps.  If the TSU is shutting down for several minutes then coming back on, this is an overheating problem.   

John C.

[peteski]

I didn't realize that the caps came into play so often...I figured it was just when you once in a while would hit a momentary section of dirty or dead track and then the capacitor kicked in to provide power for that moment. I never thought that it would be kicking in as often as it is suggested.

They don't really "kick in" - they are always there filtering the internal power supply in the decoder.  Their total capacitance determines the length of the period of time they will keep supping power during intermittent power pickup period. But we are talking fraction of a second.  If you were to hook up an oscilloscope directly to the power pickups of a locomotive and observe it while the loco was running, you would be really surprised just how "noisy" the electric pickup is.  Those caps filter out that noise.

The caps we are discussing here aren't really the "keep alive" super-caps which store energy to power the loco for few seconds with no power being picked up from the track.  Those circuits are slightly different. 

[Bart1701]

Success!  Thanks for everyone's advice.

The 2 tantalum caps were rec'd and installed. The sound stays on now the entire time. Now, on to programming the sound CV's today.

I did have one issue after the caps were installed with the sound cutting out almost every time the locomotive went through a turnout. I separated the tender from the loco and pulled it around with a diesel pulling it. Still had the same issue so I knew it was localized to the tender. After doing some testing to see if power was being pulled from each set of wheels to the decoder, I discovered that only the rear set of wheels was sending power to the decoder. My install was causing the brass contact strips to not make contact with the front wipers - so no power to the decoder from those front wheels, and also no power to the decoder from the drawbar connection to the loco drivers either. When the tender was going over some Atlas turnouts or a few Insulfrog Pecos, those rear wheels were hitting dead spots and causing the decoder to reset itself when it got power again. There's always a solution to every problem, you just have to figureo ut what the problem is first!

Bart