Troubleshooting a LL GP60 DCC install - SOLVED (and long)

[daniel_leavitt2000]

I have a first generation LL GP60. It is not DCC friendly, so I am using a TCS CN-GP decoder board for it. I followed these instructions:
https://drupal.tcsdcc.com/installation/n-scale/1668

I put it on the track - short.
The motor is close to the frame so I insulate that with kapton tape - short.
I use electrical tape around the motor contacts and areas between the frame halves where the motor wires go - short.
I take trucks, fuel tank, shell and wipers off to test bare frame - short.
I remove and de-solder the motor entirely - short.
I go over the entire decoder installation with a 10x lupe - no defects on the PC boards or solder connections. No metal contact with anything other than the feeder tabs of the PC boards - everything is fine.
I throw a NEW decoder in it - short.
Test decoders outside the frames with alligator clips - works fine.
I throw both "malfunctioning" decoders in Atlas GP9s - works fine.

My gut tells me the short was before the decoder or they would both be fried. The frame halves are not contacting each other. The motor and flywheels are obviously not the issue because it is still shorting with them removed.

What in tarnation am I missing here?

[garethashenden]

Wait, the two bare frame halves aren't touching but are shorting?

[daniel_leavitt2000]

Yes - without any component of the decoder PC boards other than the contact pads touching the frame. It's the damnedest thing I have ever seen.

Frame assembled without decoder? Volt meter says no short. Decoders work fine in other installations.

[peteski]

Ok, so with the decoder removed and motor isolated from the frame halves there is no short?

The decoder is a 2-part affair.  Ok, what happens if you only install the front board in the chassis?  Short? No short?
What if you only install the rear board in the chassis?  Short? No short?

What if you only partially insert the decoder boards into the chassis (not all the way in)?

I also wonder if you accidentally flipped one of the decoder boards and installed it upside down?  Is that even possible?
Scratch that.

[jagged ben]

Please tell me you have a multi-meter that can test for continuity before putting the loco on the track.

Assemble as much if the loco as you like and test the frame halves for continuity.  Try testing the contact pads on the decoder boards to each other for continuity.  Try inserting either half of the decoder as peteski suggests and testing the frame halves again.

Make sure that no components or traces on the boards are touching the frame, other than the large contact pads that are supposed to.  The couple of installs of these decoders I did required some milling and filing of the frame and some Kapton tape for good measure to avoid such unwanted contact.   These are not as plug and play as they might like you to think.

[daniel_leavitt2000]

Please tell me you have a multi-meter that can test for continuity before putting the loco on the track.

Assemble as much if the loco as you like and test the frame halves for continuity.  Try testing the contact pads on the decoder boards to each other for continuity.  Try inserting either half of the decoder as peteski suggests and testing the frame halves again.

Make sure that no components or traces on the boards are touching the frame, other than the large contact pads that are supposed to.  The couple of installs of these decoders I did required some milling and filing of the frame and some Kapton tape for good measure to avoid such unwanted contact.   These are not as plug and play as they might like you to think.

I'll do a full multi-meter test tomorrow.

[peteski]

Please tell me you have a multi-meter that can test for continuity before putting the loco on the track.

LOL!  I assumed (yes, I know) that when Daniel stated "Volt meter says no short." he actually meant that he had a multimeter set on resistance/continuity range and it was either beeping (showing continuity or "short") or zero ohms.

I also  trusted Daniel when he mentioned "I go over the entire decoder installation with a 10x lupe - no defects on the PC boards or solder connections. No metal contact with anything other than the feeder tabs of the PC boards - everything is fine."  that  nothing but the decoder contact pads were contacting any part of the metal frames.

[jagged ben]

I missed Daniels second post.

I wonder if there's something on the bottom of one of the decoder boards that's bridging between the frame halves.  You wouldn't be able to see it underneath the board.  Seems kinda unlikely but it's the best I can think of.

[Steveruger45]

IIRC I think I had a similar issue many years ago on a GP35, long since gone.
Try desoldering the blue and yellow wires to the rear section of the decoder and see if the short  is still there or not.  If no short then reconnect the blue and yellow wires in reverse.

[daniel_leavitt2000]

A bizarre technical problem would have to be caused by an equally bizarre set of circumstances. So let's get into it.

As I had theorized, the short was before the decoder, but in a way -caused- by that very decoder. There was not one, but three shorts.

1: Motor bridging the frame short. LL did not use a motor cage as virtually every other manufacturer does with split frame engines. It uses two small dog bone parts attached to the output shafts to keep everything lined up. These parts plug into voids in the frames. Problem is, the motor can flex and short if the shell is off (the shell keeps the motor from shifting). The tolerance is very small between the motor and frame. Small frame flash was able to cut into the kapton tape and cause a short as the shell was off due to troubleshooting. In this case, had the motor not flexed, the flash wouldn't have had a chance to cut though the tape.
 
But the short persisted with the motor outside the chassis and even disconnected...


2: There is, indeed a flaw with the design of the engine. Let's look at the shaft bearing sockets...
[ Guests cannot view attachments ]
Chassis with electrical tape all over.

[ Guests cannot view attachments ]
See these pins? Almost every model has them. They are used to hold the bearing blocks in place.

[ Guests cannot view attachments ]
And they touch the other frame side in this model! They didn't at the factory. I'll get to that in a minute.

[ Guests cannot view attachments ]
The other side?

[ Guests cannot view attachments ]
Same deal!

So why would this model work fine on DC but short with DCC? It all comes down to the boards! The factory DC boards are a specific wafer width, probably about.5mm.

The TCS boards are just a hair thinner. They need a small blob of solder on each contract plate.

What happened was the solder blobs on the non bearing frame half contact plate was just a fraction of a mm thicker, pushing that frame half up and into the bearing block pin.

3: Missing bearing blocks. When testing the decoder, I removed all bearing blocks and shafts to try and simplify the problem. But in this case, the flange on the bearing block would have prevented this short. Now if I used all the bearing blocks in the completed model, it probably WOULDN'T short. But I use Ron's end blocks on all my models to improve performance. In this case, going without the mid bearing blocks allowed that short to continue even after everything else was assembled.


The solution: I simply shaved 1/5mm from the bottom of each bearing block pin. Then I sanded the flash by the motor. That's it. She works fine now.

[peteski]

Excellent detective work Daniel! Thanks for posting the solution

[eja]

Model Railroading is Fun ....

[jagged ben]

Agree with peteski, excellent work.  Gonna try to file this to memory.

Bizarrely, I just had a similar experience.  I'm using an ESU Lokpilot wired to a Kato lightboard in an Athearn FP45.  (Yes, this is legit.)  Having some performance issues so I try to swap the board to a Kato loco to see if it follows the decoder or is due to the Athearn mech.  Put the board in the Kato, test with multi-meter, all seems well, then put it on the track ... short!  Put it back in the FP45, no short.  Welp, I'm not going to figure it out tonight, but thanks for the reality check that it's probably something I just couldn't see.