In my case, my 3/8 coil (from stout tanks) was coiled quite tightly and in a small radius. In essence, it divided my HLT fluid into two zones (inside the coil and outside the coil) that didn't interact. I added a tee to split the recirc ingress into two paths, so some went to the zone inside the coil and some went into the zone outside the coil. This made a pretty big difference. It is possible the same thing is happening with you, although with only 20' I wouldn't think you have enough coil to cause this to happen.
The best thing you can do is insulate the MLT. None of that reflectix stuff, their R-factor is negligible for our application unless you also build an air gap into it. There was actually some sort of litigation over their claimed R-factor, as their testing method produced some deceiving results. You can tell the numbers are fishy, as on he website I think it states that doubling up the material doesn't have a large impact on the R-factor. So, all it does is reflect radiant heat transfer and create an air barrier to limit natural convection. Neoprene is a lot better, if you can buy / fabricate something that fits well. I'm sure there are other insulation material options too.
If you insulate the MLT, the coil has less work to do to add back the lost energy. It will track the HLT much more closely, and ramp up temperatures faster. A better coil (and flow rate) will help, but you have to look out for grain bed compaction if you recirc too fast. The max recirc rate will depend on your MLT dimensions, grain bill, and lautering mechanism. I think a false bottom is much more forgiving than something like a manifold or stainless braid. I'd love to see somebody do an experiment on maximum recirc rates.
One option that I would like to explore myself is utilizing an external heat exchanger for the HERMS system. Something like a copper or stainless CFC. Recirc the water at full speed through the outer shell, recirc the wort through the inner shell as fast as the grain bed allows, and maybe insulate the outside of the heat exchanger. I expect some pretty great heat transfer between the two fluids given their turbulence. With an in-tank coil, no matter how much you churn the HLT up, there's going to be a thin layer of cooler water next to the coil. A CFC would eliminate this problem mostly. I'm waiting on some other things to do this upgrade, but I plan on writing the results up on HBT.
Sorry, I rambled on too much in this post... its an interesting topic that I've spent a lot of time thinking about.