Scenario - original run N Scale Atlas GP40's, non DCC-ready frames, a pair of replacement GP40-2 frames (the same used in DCC-ready GP38/38-2/40/40-2 since the '97 GP40-2) and a TCS AMD4 decoder. Nothing sinister nor unusual. I have something like 20 of these simple PnP decoder swap outs under the belt and every one has worked off the bat except when needing to sort the motor tab contact issue that appears occasionally. The loco was disassembled, the parts being transferred cleaned and lubed where necessary, the bottom motor tab cut short and the wire soldered to it. I even did the Ron Beardon mod of removing the inner bearing blocks and adjusting the play in each worm. When the frame half was populated and the bottom motor tab wire soldered to the decoder, I did a quick check with the trusty Fluke DMM and all seemed in order - no shorts. Screw it together, check again and place on the track with the PowerCab in program mode. Faint buzz. Huh? Programmed ok too! On test took off like a scared rabbit. Ok, strip the loco down and check for any errant wires or flash on the frame. Hmmm... two little protrusions at the front of the decoder on the frame may touch the motor contact pad. Nipped off, reassembled, placed on track, faint buzz. So I pulled the body off a GP38-2 with the same decoder in it and had a quick glance at the frame - it appears the same right down to the ATLAS 1997 on each half. Meter reads around 75k ohm across the frame. Note the difference here is the motor tab at the bottom is far too short on the non DCC-ready (original) motor so a short length of wire is need to be soldered between it and the contact pad on the decoder. Back to the GP40 which measures around 65k across the frame. There were a few solder points very close to the frame halves so I disassembled again (I should note I had given up screwing the frame halves together at this point and relied on the motor cradle to keep things together) and placed a little Kapton tape over these and reassembled. 65k ohms. Faint buzz. :lightning: Ok, time to step back and have a think. There are four circuit tracks that run through the tapered centre of the decoder between the motor contact pods and the rear of the decoder. These have minimal clearance to the frame when installed so I placed a strip of Kapton tape over these along their length and reassembled. Voila! 75k ohms! Placed the offending loco on the program track and no buzz. 20 seconds later it's programmed with number, direction, lighting and three point speed curve and tested. It not clear why this happened with this particular install, except it appears the decoder was partially shorted to either, or both, frame halves even though lacquer coated. I have done probably 15 of these particular framed locos using NCE and TCS decoders with no problems - the rest were the ASD4's fitted Geeps with a different length frame, of course. The difference in this case was a retro-fit of the frames - the same frames! I did note the board was a reasonably tight fit without the need for soldering the frame contact pads or doing any shimming to increase contact tension. This is a v4.8 AMD4. Maybe it was closer to the frame by that bit more? Moral of the story - use Kapton tape from the outset and save the hassle. We live and learn as the Colonel most annoyingly pointed out! Now to do the second unit, this time prepared.
