Adrian! posted:I've actually been asked to write something intelligent here (a first)... so let me try...
Uh so like, PS2, PS3 whatever, you need to first make sure the layout isn't sad. Specifically (1) the amplitude of the DCS waveform is you know, not awful (>10V excursion)... and (2) the waveform shape is sharp and not a sawtooth like thing, ringing like a mess. If the DCS waveform or amplitude is a mess you should fix that first, definitely. Many posts are dedicated to how to do that, and many people here can help you (esp GRJ).
Now beyond "the layout having issues" I have seen that the DCS amplitude measured at the track is vastly different with different locomotives on the track. For example I've seen where a PS2 SD-whatever train knocks the 14V amplitude down to about 4V swing, while my giant DDA40x PS3 trains only drag the same signal down to 10-11V, which implies the impedance loading of different trains is different. The PS boards have something sort of like a Diplexer that separates the 60Hz and 3-12 MHz DCS signals. The 60Hz is sent to the rectifier block, while the DCS is sent to the CDMA decoder. The diplexer basically works by being a frequency selective series impedance to keep the 60Hz out of the digital decoder part, and keeping the 3-12 MHz out of the power rectifier circuitry (and down stream lights and motors).
I suspect what is going on is the diplexer design has been revisited from PS2 to PS3 to greatly reduce how much of the DCS signal power flows into the rectifier system. This also suggests that if you lift off one of the AC input pins of the rectifier bridge on the PS board and slip a choke in series with it, you can improve the diplexer performance further since the choke will raise the impedance at 3-12 MHz, while basically still being a 0 ohm element at 60 Hz.
Thanks for responding to my email. Uh, like, that was pretty intelligent. 
Thanks to GRJ's advice, will shortly try different potentiometer in his TMCC buffer to reduce TMCC v/amplitude more and see if the TMCC signal improvements keep it effective at lower amplitude, while reducing DCS interference problems. Watch this space for a report.
If I later need to address problems with some individual DCS locomotives: For cutting off above-60-cycle input to the rectifier bridge, sounds a simpler assignment than the impossible one of cutting between overlapping TMCC and DCS frequency ranges, as pointed out by GRJ. What KHz or MHz value recommended for peak impedance, if that is pertinent spec? (or which spec to pay attention to?) I get that the choke has to have a 4-6 amp current rating because it will pass the motor and other loads.
I suspect a whole lot of folks gave up on this battle long ago, if they had the same challenges. It would be good to find solutions.
