Training mode - constant speed or speed based intensity adjustment?

[fellrnr]

Would it be possible to implement constant speed, or isokinetic training? This would reduce resistance if your speed drops, or increase it if your speed increases.

Jonathan

[amtriathlon]

No supported file format and no smart trainer has these feature, it could be implemented via extensions and closed loop control, but GC doesn’t have have closed loop control, not even for power training, see #1789.

[fellrnr]

The feedback loop wouldn't have to be fast. Having some level of latitude would be appropriate. If you had a target speed, then +/- 10% would result in no change, but exceeding that would result in a percentage change to the intensity with changes ever few seconds, similar to hitting the up or down buttons on the training bar. The trainer won't react faster than a few seconds anyway.

Does that make sense?

[amtriathlon]

Does that make sense?

Not for me, but I don't even understand what's the point of this feature.

[fellrnr]

That's a good point. Let me describe the problem and the solutions I've found so far. I'll post this on the user forum in case someone has a different approach that I'm missing.

Problems with maximum intensity “all out” training on a bike trainer
Using a bike trainer for “all out” training is tricky as the intensity will vary day to day, and even within the same interval. If the “3 minute all out test” type of interval is used, the power output will drop from a brief peak down to critical power.

1. ERG mode obviously won’t work as you can’t predict the required power output. Set the power output too high and a positive feedback “death spiral” will occur with resistance increasing as cadence reduces until a stall occurs.
2. Incline against distance is better, but if the incline is harder than expected, the speed is slower so the distance takes longer to cover. Likewise, an incline that is too easy is also too short. Another positive feedback, though not as failure prone as ERG mode.
3. Incline against time solves some of the problems, but no standard software supports this. I modified Golden Chetah to implement incline against time, mapping watts to incline to give approximately the right watts/incline from an ERG file.
4. Ideally, the training software would adjust the incline so that speed (and therefore cadence) will stay in a reasonable range. This is especially valuable in the “3 minute all out test”. This wouldn’t need to be a precise, “real-time” adjustment, but enough to keep the cadence from dropping too low as W’ Balance is exhausted.

[rgeorge1985]

Hi Jonathan

Curious why you are relying on the 3MAOT? Its a) unpleasant b) has a learning curve, difficult to dial down intensity required and c) modeled values maybe higher than what you can sustain.

Go for simple 3, 7 and 12 or 15 min constant time trials.

[fellrnr]

Sorry for the lack of clarity. I'm not using 3AOT as any kind of test, but as a training approach. Even if I'm not doing the 3AOT specifically, I'm typically doing HIIT at close to all-out intensity. (I believe the 3AOT provides some unusual benefits - https://fellrnr.com/wiki/3AOT)

Thanks! Jonathan

[amtriathlon]

I still don't understand the rationale for this, provided:

1. It is well known the 3min all out test was developed for and validated on fixed resistance ergometers, to vary resistance will invalidate the protocol.
2. It is supposed the users have gears available and know how to use them in slope mode, just like in real hilly rides.

[fellrnr]

1. As noted above, I'm not using 3AOT as a test.
2. I find that at maximal intensity, changing gears is awkward, partly due to reduced SpO2 and partly because I struggle with gear change when I'm pulling on the bars hard. I figure software can manage it far better.

(On a complete tangent, I disagree with assertion concerning the use of fixed resistance for 3AOT. Some studies did, but some used isokinetic which seems to be somewhat better. There's also evidence that CP changes with cadence, see "Human critical power-oxygen uptake relationship at different pedalling frequencies")

[rgeorge1985]

I still don't understand the rationale for this, provided:

1. It is well known the 3min all out test was developed for and validated on fixed resistance ergometers, to vary resistance will invalidate the protocol.
2. It is supposed the users have gears available and know how to use them in slope mode, just like in real hilly rides.

Ale,
I'm not sure if isokinetic is totally outlandish in indoor training. I still own a Computrainer and if I recall correctly, the original Racermate software had an option to program bots to hold a fixed speed so you could draft them and hold that constant speed. If you then imagined there were no bots and simply a UX asking the user what speed they wanted, you could satisfy a need to ride at constant speed by varying resistance as OP wanted , couldn't you? Since this is ERG mode, I suppose any creative way to manipulate resistance is ok.

[amtriathlon]

I find that at maximal intensity, changing gears is awkward, partly due to reduced SpO2 and partly because I struggle with gear change when I'm pulling on the bars hard.

It gets better with practice, and it is a good skill to have for cycling in the real world.

[fellrnr]

I think I understand the disconnect. I suspect most GC users are aiming to mimic outdoor cycling, where gears are your only option for changing intensity, and slope/distance is the nature of hills. I'm trying to optimize exercise on a smart trainer, and I don't cycle outside. To me, changing gears to modify the resistance on a smart trainer seems bizarre, It's like having a microwave that only does a fixed time and you have to modify your portion sizes to change the heating. As a software engineer, I find the ability to creatively change the exercise fascinating, and I tend to prefer changing the UX rather than changing the human.

[amtriathlon]

Yes, I also see the disconnect, and it is not exclusive of this thread, there are other users who want the software controls the cadence, or does the pacing for them, like #4142.
Effectively Train in GC is a tool to help cyclists and triathletes and, in my opinion, removing the elements which require to use the supercomputer between the ears is not a step in the right direction.
But this is an open source project, you can experiment with the features you want in your fork and, if you contribute them some can be merged, provided they follows the contributing guidelines and the scope of the project.