Trouble-shooting DCC Power

[GaryHinshaw]

All,

I'm finally getting around to the next stage of track-laying on Tehachapi BC and I've run into a puzzling power issue that I could use some help diagnosing.  The issue I'm having is that one section of track is acting balky: when I try to run a DC loco on channel 00 I get decent performance, but when I try to run a decoder-equipped one, the lights will come on, but when I start the throttle, it hesitates and stalls (and the lights will go out).    Here are some basic facts:

* To date I have 4 separate DCC power busses radiating out from the (Digitrax) command station in a "star" arrangement, where each terminates in an open circuit at the far end of the run.   As of now these are all just hard-wired to the command station without circuit breakers.

* The bus itself is 12-gauge stranded copper, but (perhaps foolishly) I did not twist the wires along the run, rather they are separated by ~3".    The longest bus run is about 35', so I would like to think that inductive losses are manageable (but see below for a question about that).

* The section of track in question is physically isolated from the main pike, but the bus feeding that section also feeds another section of the main pike, and that section performs well.   (I should note that the isolated section has been in place for several months and has been unused, so I should double-check that the track is clean... , but I've never had a problem with dirty track anywhere else, and this section works fine with DC.) 

* A very simplified schematic is shown below.  The main pike is a closed loop served by the 4 busses, and the new section is hanging off the end of one bus, which also serves a short segment of the main loop.  (The isolated section will soon be cut into that main loop in that quadrant.)  It is possible that I don't have a reliable gap in the loop where the question mark is shown, and that could potentially lead to a bus loop.  But if that were a problem, I would have expected to see trouble in the main loop, and I don't.  Decoder equipped locos run beautifully on the rest of the layout, including sections with longer bus runs than the one in question.

* I have double-checked all the wire connections (and re-did a few).  I am 98% sure they are ok.

So, any advice on trouble-shooting steps?  One thing I'd like to check is whether or not my inductive voltage drops are significant or not, but I'm not sure how I would tell.  Is there a standard track measurement I can make that will be definitive?  I assume I want an AC voltage at 10 kHz, but I don't think I have a suitable meter (yet).  Is there another way to tell?  Can I use DC voltage with the throttle cranked on channel 00 and look at variations down the bus?  Any other obvious things to try? 

Thanks in advance!
-Gary

[John]

Re-solder the feeder for that section .. and also try multiple dcc locos ..

[ednadolski]

the lights will come on, but when I start the throttle, it hesitates and stalls (and the lights will go out).   

In my (very limited on DCC) experience,  that sort of symptom was often indicative of dirty track or some other contact issue.   IIRC, you mentioned that you were going to leave the ME factory weathering on the railheads.  If that's the case on this section then perhaps the weathering is degrading the DCC signal?   You would not necessarily see the same symptom with a DC voltage, or even on other sections with weathered rail.

Ed

[peteski]

That behavior of the DCC loco to me seems to indicate that you have mismatch in number of speed steps between decoder and what the command station is set for.  I'm talking about either 14 or 28/128 steps. If you loco (in CV29) is set for 14 speed steps, and the command station is set for 128 steps, it usually results in similar behavior (flashig headlights). I think I read that somewhere in the Digitrax support documentation.

[GaryHinshaw]

Problem solved - it was just dirty track....  A wipe of the rails with a dry rag was all it took to bring it back to life.  Surprisingly, the rag did not pick up much visible grime, so it was probably just some oxidation built up after several months of not running trains on that section.  Ed, you are correct that I have not polished the rails of the ME weathering, but that has never presented an issue on track that see regular service.   Just goes to prove the value of running trains regularly. 

Sorry for wasting people's time,
-Gary

P.S. To satisfy my curiosity, I did measure the DC voltage of the track at several points along the bus with the throttle cranked on address 00 and I found no measurable drop at the far end of the bus, so I think this shows that any inductive losses are negligible.  I'd like to repeat that when there are also lots of decoder-equipped locos running, just to see if that makes any difference.  But things seem to be back on track.

[John]

Oh yea .. and clean the track

[mmyers]

I use an AC meter from Harbor Freight. Exact? Close enough to check for differences in voltage. Lets face it a reading of 13 volts might not be accurate but if the next section reads 9 volts you got a problem that isn' t the meter.

For checking voltage drop, I use a RRampmeter and a pair of 8ohm, 10 watt resistors wired in parallel.

[GaryHinshaw]

Copy that Martin, and thanks for the tip about the RRampmeter; I was not aware of that product (I'm still a DCC newb).  Could you explain how you came up with a pair of 8 ohm resistors wired in parallel? 

If I understand things correctly,  I do think a set of relative DC measurements along the bus tells me what I need to know about loss.  I'm assuming that a given throttle setting for address 00 sends the right mixture of 1's and 0's to generate the desired DC voltage, and since the signal is fairly narrow band AC, any losses will affect the DC value in the same proportion as the AC voltage.  But I might be neglecting something important...

-gfh

[ednadolski]

I did measure the DC voltage of the track at several points along the bus with the throttle cranked on address 00 and I found no measurable drop at the far end of the bus, so I think this shows that any inductive losses are negligible.

Are you using a DC voltmeter?  If it is pure DC to the rails, (i.e., 0 Hz, not some equivalent waveform) then you won't see any voltage drop when no current is flowing in the circuit.  You did remove all loads (locos), so you are measuring the open-circuit voltage?

I'm assuming that a given throttle setting for address 00 sends the right mixture of 1's and 0's to generate the desired DC voltage

So it's not pure DC to the rails.  To produce actual DC from a periodic waveform, you would need a rectifier circuit.  These are built into the decoders, so the motor does not see the waveform that is actually on the rails.   To see that, you would need an oscilloscope.   For the specialty DCC tools, the mfr would have to say exactly what they are measuring.

Ed


Edit: DCC decoder block diagram:  http://www.awrr.com/fig6.gif

[mmyers]

Gary,
Two 8 ohm resistors in parallel yield a 4 ohm resistor. We used this to test voltage drop under load for the NTRAK wiring RP. We couldn't find a readily available 4 ohm, 10+ watt resistor. Radio Shack had the 8 ohm, 10 watt on the shelf. An 1156 tail lamp bulb will apply about 2 amps so it could be used as well. We needed something that would remain fairly constant and duplicated in other parts of the country for our tests.

Simply connect the leads from each together at each end. Then connect each end of the resistors to the track and power up. 4ohms at 12 volts will yield a 3 amp load. Then go ahead and take readings along the whole power district comparing as you go. We found that the signal gets unreliable when voltage drops below 10 volts.

Martin Myers

[C855B]

Two 8 ohm resistors in parallel yield a 4 ohm resistor. We used this to test voltage drop under load for the NTRAK wiring RP. We couldn't find a readily available 4 ohm, 10+ watt resistor. Radio Shack had the 8 ohm, 10 watt on the shelf. An 1156 tail lamp bulb will apply about 2 amps so it could be used as well. We needed something that would remain fairly constant and duplicated in other parts of the country for our tests.

Simply connect the leads from each together at each end. Then connect each end of the resistors to the track and power up. 4ohms at 12 volts will yield a 3 amp load. Then go ahead and take readings along the whole power district comparing as you go. We found that the signal gets unreliable when voltage drops below 10 volts.

And don't touch the resistors (HOT!) during the testing, giving them a chance to cool off when done. Also be careful where you set them down. 3A @ 12V = 36W, all being dissipated as heat. I have soldering irons smaller than 36W.

[mmyers]

Forgot to mention the heat. They actually came in handy back then. I did my testing in an unheated garage mid February.

[GaryHinshaw]

Thanks for the tips gents.  Trains are now behaving well in the revived section, but I'll pick up an RRampmeter, if for nothing else to satisfy my curiosity about what's going on in those wires.  I did take some more DC readings last night with channel 00 cranked and with decoder-equipped trains running in the same block. I saw *maybe* a <0.1 V drop as I turned trains on and off in that block.  An imperfect measure for sure, but things are copacetic again.

So it's not pure DC to the rails.  To produce actual DC from a periodic waveform, you would need a rectifier circuit.  These are built into the decoders, so the motor does not see the waveform that is actually on the rails. 

That's correct, it's not pure DC.  My understanding is that channel 00 is reserved for producing a DC offset in the track power for non-decoder-equipped locos to respond to.   The AC component is still present, which causes DC locos to "sing".   I'm not sure what kind of filter my DVM uses when measuring DC voltage, but it's only showing 8.5 V when the throttle is cranked, FWIW.