Thanks for the responses. I discovered that I was simply working too fast and missed a spot where the back of the decoder was touching the frame through the adhesive....supplying constant power on.
My bad.
The decoder survived the shorting and all is working well.
A problem like this is why I always try to use a decoder with the JST 9-pin header and a matching wiring harness. If the decoder is bad, it's very easy to substitute a different one as the first step of troubleshooting. It doesn't happen very often, but I have had bad decoders that do exactly what you're experiencing.
Are you absolutely certain that the motor brushes are totally isolated from the frame? When you put the frame back together, a sharp piece of wire or solder may have poked a hole through electrical tape that should have been placed between the motor brush wipers and frame halves. If you didn't use thick electrical tape there, go back and re-do your installation, and you'll probably have to change the decoder, too.
Well this has never happened to me before, but I have never worked on this old loco type before.
I just put a simple NCE decoder into an older Bachmann split frame HO Dash 8. All connections look good to me.....track pick up to either side of the frame. The motor is isolated and connections are good.
The decoder reads and writes CVs correctly on the programming track, and I reset all values to factory default.
However, the loco begins moving backward when the power is applied to the track, regardless if any or either direction is switched on. The speed control works, but it just moves backwards as soon as power is applied to the track. The loco worked properly as a DC on CV 0.
Perhaps the decoder is faulty...I have never actually sat down to learn how to test them with the Zephyr test leads.
Thanks to anyone with a tip.