Dcc is not dc. It is a switched bipolar dc. Switching frequency being in the khz range.
I have a beautiful mehano locomotive that i wanted to convert to dcc. It had great led lighting and a good motor. During this process i had to remove its original pcb as i accidently cut off some of the tracks. So i made my pcb with the NEM connector for easy decoder swapping. The NEM connector pin spacing is 0.1″ and so was easy to make myself. During soldering i used an acid based easy clean flux.
The loco was wired up. When i fired up my controller i immediately ran into some weird issues. Sometimes the motor ran okay. Sometimes it did not. The lighting was all messed up as both the front and rear lights would come on simultaneously. The F1 light used for the Cab was always ON even when the F1 switch on the controller was off. In general the operation was bizarre.
Initially i thought that it must have been the controller. True enough, when i tested the decoder on a testbed, it had the same issues. So i replaced it with a known working Digitrax-DZ125 decoder. Again, same problems. When i retested this decoder on a testbed, it misbehaved again. This meant that loco was actually messing up my decoders. I could not afford to replace any more decoders on the loco !
But again since Digitrax-DZ125 always had overrated motor load handling capacity, i thought i’d try out the ESU decoder. This decoder actually had overload protection and is more rugged. In went the ESU decoder. Loco ran okay for a minute. Then again, same problems, now with the lights. The motor running seemed okay.
Now i was at a loss to explain what was going on. During the initial stages, i checked each NEM connector pad against the other to make sure that there were no shorts. So i knew that DC isolation was not an issue. By this time i had already destroyed 3 decoders. I thought that it must have been the motor. So i tested the loco motor separately in another loco. It ran fine. So motor was not a suspect.
That left only the PCB i made. But i knew that i tested them for shorts using multimeter set in the 1kohm range and everything was okay. I guess this where my 4 college and work experience came in handy. I removed the PCB. Set my multimeter in MegOhm range. Then tested each pad against the other. The results were amazing. The same pads that showed 0-ohms for short circuit testing now showed about 1-megohm between some pads. Then it all made sense.
1: Since the operation was not DC but instead an AC type operation, these pads that were 0-ohms on DC were actually conducting during the DCC operation.
2: I used acid based flux instead of rosin based one. Acid based flux while great for cleaning does have a small conductance. Combine this with the 0.1″ pad spacing and this effect went up.
3: All DCC decoders nowadays use MosFets for the lighting system. The Mosfets are driven open drain (drain floating). Mosfets are voltage driven unlike BJT’s that are current driven. Their gates are quite sensitive to voltage fluctuations. The capacitive coupling of the DCC signals from the track could have played havoc on these Mosfets, thereby driving the LED’s sporadically.
So i re-designed the PCB by removing the NEM connector. No more 0.1″ spacing pads. The pads and tracks on the PCB were wide apart. I cut the NEM connector off the decoder and soldered the decoder directly to the board. This time i cleaned the board thoroughly with rubbing alcohol and an old toothbrush to make sure there was no flux residue. In the photo, the new PCB is mounted on the loco chassis and the culprit PCB is lying below the loco. The loco was wired up again and powered.
Success!!!. The loco now ran great. Great running characteristics, great directional and cab lighting with no flickers or other glitches. So the culprit was the PCB and the flux.
I hope this troubleshooting process helps anyone in a similar boat.