How does erg mode apply resistance during the pedal cycle

This is something I’ve been thinking about and haven’t been able to find an explicit answer to.

On a real bike, force applied to the pedals as a function of pedal angle is non-linear. Ie: applied force peaks at a certain angle and is near zero at another angle. Due to the flywheel effect of your bike moving though, your speed remains essentially constant despite this cyclic rise and fall in applied force throughout the pedal cycle:

Most trainers have a weighted flywheel to replicate this, so that biomechanically, your pedalling on a trainer will follow this same pattern - maximum applied force at a certain angle, then minimal applied force at another pedal angle.

However, what’s going on to the graph of applied force as a function of pedal angle when you’re riding in erg mode?

The only description I can find is that erg mode keeps resistance constant. Does that mean that when you’re in erg mode on a trainer that, even given the flywheel, resistance through the whole pedal cycle is now constant, meaning you need to apply a constant force at all angles of the pedal cycle? This is what erg mode kind of feels like I think.

Basically, in erg mode, is the above graph going to look like a circle instead of a peanut?

Resistance is applied at the flywheel so the answer needs to be approached from that perspective.

The mass of the flywheel defines its inertia, which is the tendency to continue spinning at the same speed despite changes in trainer resistance or power applied at the pedals.

Resistance provided by the trainer is a function of flywheel speed. In a long steady erg interval, stomp on the pedal once. Power will increase as the flywheel speed increases and the trainer hasn’t yet reduced resistance. Then the trainer will “catch up” by lowering resistance. If you then return to your previous cadence you’ll see a dip in power until the trainer adjusts for the lower flywheel speed.

The final part of the answer has to do with smoothing. Trainers will detect the flywheel speed several times per rotation (usually an optical sensor but could be via other means). The trainer will need some number of data points to reliability detect a change in flywheel speed and adjust resistance to achieve the desired power output.

I think all of this means that the pedal stroke power graph is the same. If not, riding in erg mode would feel fundamentally different than in resistance mode or outside, but it does not.


I think your thought process is right, but the unanswered questions are:

  1. How frequently does the trainer measure flywheel speed and what is its sensitivity to react to changes in speed?
  2. over what time interval will the trainer adjust flywheel resistance over?

While it would depend a bit on the exact gear ratio you are in, a single full pedal cycle is going to lead to multiple flywheel rotations on all the trainers I’ve had.

During this one pedal cycle, there would be a small increase in flywheel speed at the ends of the “peanut” and a small decrease in flywheel speed at the indent of the peanut on the force vs angle curve.

If this small change in flywheel speed both lasts long enough within a given pedal cycle to be detected by the trainer, and is large enough to be detected by the trainer, then the trainer could still increase/decrease flywheel resistance within a pedal cycle, making the graph into a circle from a peanut.

Actually I think some power pedals can display force versus rotational angle data. Someone could figure this out for sure by going into erg mode and then graphing applied force as a function of pedal angle at both a very high cadence and a very low cadence. The two cadences would serve to make sure there isn’t a threshold speed whereby due to the sensitivity/frequency of measurement of flywheel movement, the behaviour of the trainer changes.

There will be no difference between ERG and simple resistance mode with regards to pedal force profile.

The feedback loop that sets the resistance is nowhere near fast enough to smooth the small variations during the pedal stroke.


I don’t think so. A simple test is to do a single legged pedaling drill in resistance mode. Notice how it feels. Switch to erg mode and use both legs. Does it feel like you are doing two single legged drills at the same time? Not to me.

If you think about the bio-mechanics aspect, if the trainer were to make the curve a circle you couldn’t pedal well for very long. Part of why the power/angle is peanut shaped is also your ability to deliver torque to the cranks through the rotation. If the trainer resistance curve tried to compensate, you’d be unable to pedal through the weaker portions of your pedal stroke, and/or the stronger portions would feel too easy, or it would be as if the low cadence death spiral happened throughout the pedal stroke.

The trainer tries it’s best to mimic what your bike actually feels like on the road. Response to power input changes are deliberately slow to prevent the trainer from following the variations in your pedal stroke too closely. That sensation would destroy any illusion that you’re on the road.

I would agree with this, at least on the kickr and H3 that I have used. You can easily overshoot the power by increase cadence, so the control loop feedback time seems to be several seconds, and is not adjusting multiple times per pedal stroke. If it were really adjusting as fast as suggested, you would be able to quickly change cadence in ERG mode and still see a smooth/stable power output, and this is not the case.
I think for some trainers like the tacx neo2 that does not have a flywheel, the feedback is likely much faster in order to implement the ‘virtual flywheel’, so I think for trainers without flywheels the situation is likely much different. As I understand it, in order to have the ‘virtual flywheel’, the trainer will need to reduce the resistance during the dead spot in the pedal stroke each revolution to provide the feeling that the flywheel inertia would otherwise provide. Early testing of the neo2 had complaints regarding ERG mode adjustments being ‘too quick’ - even if possible you don’t want the resistance to go from 100 to 300 watts in .01 seconds.
I have not seen any trainer company reveal details of their control loop parameters, as they likely consider this information a trade secret.

It’s a good question and something I’ve often thought about. I’m wondering if the pedal cycle plays a big role in how resistance is applied. For example, I’ve struggled to complete easy tempo intervals in the past when using ERG mode but in resistance mode it hasn’t been a problem (same gearing, same power target, same average watts, same bike / position, same cadence). Due to this, I’ve wondered if a certain pedal stroke suits doing things one way or another, but who knows.

Yeah, I think people with a rougher pedal stroke might have more issues than smooth pedaling. Trainers with smaller flywheels could make that issue worse as well.

For a mid to high level trainer with a clean cadence, I think the power delivery will be quite similar between the modes. The flywheel will smooth out ups & downs (as intended) and unless the trainer is overly fast on resistance adjustment response, there is not a major reason for the modes to differ (in the same gearing).