At my club's shows, we run DCS and Legacy/TMCC locos on the same loop with no problem so Lionel vs. MTH compatibility is not an issue. Your symptoms are Deja Vu ... my club spent years trying to make DCS reliable on our modular layouts. We tried all of the insulated block/choke/filter/lighted terminals suggestions with inconsistent results. My observation of our DCS testing efforts suggested (to me) is that impedance changes and signal reflection in the layout wiring can weaken the DCS signal. I noted a lot of similarities with trying to balance ArcNet (remember those days?) with 50 Ohm resistors and attempting to balance DCS signal strength with lights, insulators, etc. Impedance and resistance will change with wire junctions, resistance loads (lights, locos) etc. Couple that with electrical loads that roll around the layout and you have an environment that will frustrate balancing efforts.
We found that providing (a) consistent power and ground wire gauges (12G in our case) and (b) providing discreet ground and power PAIRs (the ground is NOT common except at the TIU output) for each loop provides the best possible signal to the tracks. We have drops in every module (4 ft.) connected with terminal blocks. We minimize the length of small gauge wire feeds to the track and solder 12 or 14G feeds directly to the track where possible. We removed all filters, insulators, accessories drawing track power from the main lines, and we are able to reliably operate DCS.
Finally, a TIU may have a weak output. If your TIU cannot achieve a 9 or 10 on a test track, get it repaired or replaced.
