Well, the first thing I see is there's not enough counterpoise. A 6' ground radial isn't a full 1/4 wave and a 5/8 likes to have a full 1/4 wave counterpoise.
If you can replace the 6' radials with 102" or tie several 9' wires to the top of the Imax mounting plate, then bring them down to an insulator & non-metallic line to make as wide a ground plane as possible, you may find it might work better.
Even if you can just add one 102" whip in place of the 6' pointing in the direction of someone with a good meter, you might see improvement.
Power compression? well, antennas, like speakers which use voice coils and lose efficiency with the heat caused by high power, the Imax uses an inductive matching network with a relatively small coil.
As the input power goes up the coil heats up and wastes an increasingly greater amount of power in heat, making the antenna radiate less efficiently than one which passes all the power through the matching network without heating.
NB'r, does that mean that my Starduster, which uses no matching network, is more efficient than any of these antennas with a matching network? Maybe that is why my SD'rs works so well.
A while back I think it was Dave or Oggy in the UK, that posted something about his Shakespeare 5/8 wave with a 8 element radial hub attached and the radials were short like maybe 36" inches.
This is just from memory, but due to the discussion I recall about short radials vs. 1/4 wave radials, I did a model using 36", 72", 102" radials on a 5/8 wave radiator and I think I found the shorter radials worked better than the longer ones. I can't remember the file name or the title I used for the project, and I don't remember if I posted it or not. I also don't know if I had a clue as to why the gain and angle showed to be better with shorter radials either. I also don't know if slanted radials vs. horizontal makes much difference too.
I think maybe you and I might have also discussed the setup inside of the SP antenna as well.
If I can find the file for the model or the paper work, I'll check it out and post it, whether my recall is right or not.
Based on my personal experience with my I-10K, I found shortening the radials not to be a good idea however, but I did not try and re-tune the antenna either. Thus in that case, the I-10K seemed to work best with the full length radials, maybe 103" at 27.205 mhz and I guess on that point you are correct.
However, I think that IDIeselman is working with an Imax in this thread, so I would consider that when giving advice.
I can't explain why these two would be different in this regard, unless the difference in matching, with the Imax using a capacitor to help fix the reactance, is possibly the reason.
Personally I don't think any of the issues discussed so far accounts for the reason that
IDIeselman is having problems.