Great info, thanks 
I do think there is a strong case for Resistance mode for MTI and other low cadence work. It is likely āeasier and saferā in the purest sense than ERG mode. I would recommend RES for most people and for sure new riders in particular. Even with experience, ERG for these types of efforts takes some attention and skill to not get bit from the Spiral. It can be done, but may not be the best choice for most people and uses like this.
Each trainer has its own algorithm for keeping steady power in Erg. I have an elite Direto and find I can ride at 50-60 rpm without entering the spiral of death.
The spiral of death arises when the trainer tries to adjust torque to maintain a steady power. When the crank rotational speed slows down during the pedal stroke, the trainer trIes to dial up the torque to keep steady power. If it responds too quickly with the ādialā, this is when you can enter the spiral of death.
At lower cadence, the risk of the spiral of death increases as the duration of the pedal stroke is longer - giving the trainer more time to turn the dial on torque.
Interestingly, the trainers that are known for fast reaction time (eg the Kickr) seems to have worse spiral of death tendencies than trainers known for slower reaction time ( eg the Direto). This makes sense, as the former trainers turn the torque dial quicker, increasing chances of spiral of death in low cadence riding.
This opinion is based on my understanding of PID control theory that I learned 25 years agoā¦ I hope Iāve recalled things correctly!
Sounds about right and my PID control theory classes were eons ago and I only used it on the job 33 yrs ago to do PLL design.
So I like to think of it as āfriends donāt let friends drive MTI work in Ergā
Are you using the same exact gearing between the trainer modes?
Flywheel speed could be a factor.
Yeah, thats what I suspected. The Neo virtual flywheel is designed to replicate your mass and related speed (assuming you updated it to match in the Tacx app) or use whatever is the default mass (70-75kg?).
The trainer applies that virtually based one the axle/wheel speed. So the large difference in your flywheel speed is a likely reason for your problems.
You could try a test in ERG, with 52x13 and see how it feels. At the same speed, there may still be a bit of different feel. But I am willing to bet your issue with ERG is the really low flywheel speed.
It hurts you since there is less āhelpā around the pedal circle, and the trainer also reacts quicker for resistance changes in that low speed vs higher speed. So you are giving ERG more ways to kill you while in low gearing 
Iād bet itās because the trainer is dialing up the torque too much in the dead spots of the pedal stroke, and your knees wobble because they have to push a high load in those spots to overcome the torque.
My guess is this will help.
I ride in the snow during winter. Some sustained sections of 15%. Riding at about 3 mph and cadence about 40.
I find that a smooth pedal stroke - meaning keeping power down for as much as possible around the Pedal stroke is important.
Partly because of the surface (you can only get so much traction in snow; with a torque spike you can easily spin out).
And partly to keep the bike from slowing down too much during the pedal dead spots (Noon and 6 oāclock) - which means you donāt have to push as hard during the power sections of the pedal stroke (3 and 6) to reaccelerate the bike.
I hadnāt heard of MTI training until this thread. With weight training and enough outside riding at low cadence, I guess Iām already getting enough of this kind of training.
Makes sense in theory. This relies on Erg mode accurately mimicking reality, however, which may not be the case.
Yeah, I wouldnāt necessarily rely on ERG for āreproductionā of outside conditions.
What he can do though: