New Power Match Questions

I’ve been playing with Power Match and am finding that I have quite a few questions about it. Overall it is working well, so some of this is just my curiosity but also wanting to make sure I understand the tool correctly.

I use a Tacx Neo (og) and Garmin Vector 2S left side pm, and Android app.

Q1) Where are the power match calculations taking place? I recall Nate saying in a podcast that new Power Match uses machine learning of some kind, which by my interpretation means there is some kind of calculation being done somewhere between my PM and my turbo. I use Android app. Is the Vector sending power data to the app, which is then calculating the differential to send to the turbo trainer?

Q1.1) If Q1 = “yes” is there any advantage in calculation time (i.e. the time it takes for the app to literally do the power differential calculation, which in turn theoretically makes the turbo respond quicker) if I were to use an old weak phone vs. something brand new top of the line with a shiny fast CPU in it? Would going to PC further improve this time? Or is it just negligible?

Q1.2) If Q1 = “no” where does the calc take place? Is data being live-streamed back to a server somewhere for the calc? Or some other method?

Q2) Power Match smoothness… I understand that going to max smoothness means a slower response on the turbo, and max responsiveness means a faster response on the turbo. I understand that a faster response will mean a lot of “surging” which will hurt the legs. Why would max smoothness be not great? Wouldn’t it help to keep peaks and valleys at a minimum in my actual power reading? Or would it just keep those peaks and valleys happening longer while the turbo corrected?

Q3) Is Power Match better if the power meter measures both legs? Not asking about whether a total power measurement is better to train/race with - clearly that is a “yes” - but instead wondering if the tool itself works differently on a one sided vs two sided meter?

Q4) Does Power Meter accuracy matter any more than it typically would in a power match environment? Thinking both on the turbo and non-turbo sides.

Thanks. I like the feature. Just want to be sure I’m getting out of it what it is designed for. And I may be looking for an excuse to by a dual sided device.

Edit: a few days with no responses… maybe @Nate can assist if this isn’t too “secret sauce”

This is how I think it works…

TR app is constantly reading the power from PM and based on the number it will add or remove trainer resistance to more or less keep the power. Thats why you can ride at different cadence without affecting much the power reading. TR can be use offline (well, you need to load the wo first) , so I think is all done in app…

Smoothness will be hard since you are changing power constantly. Its will be very hard to keep power on flat line. The best way to accomplish real smoothness is to flat your cadence and make sure you are not mashing the pedals unevenly. If you cadence is kept at a consistent rate, then power will be consistent.

TR just read total, So there is no real difference between L/R. At the end it will be added and send as one… in the case of single it will give 2x the reading power.

I wouldn’t say it matter more, but it matter about the same. As long as you use the same power meter on all rides you should be fine. I would try to avoid using the trainer PM and then riding out with a pedal PM.


I am also having a rough time with power match

Before I moan - I want to say a big thank you to the TR Support team who a top and great.

I have a Tacx wheel off trainer and a 4iiii power meter, I have updated the firmware on both and on a daily basis carry out a calibration after ever warm up section of each workout

I have tired Ant+ and BT I have disabled all other devices in the room… I have built and updated a new Laptop (win10)

I have tried placing the Laptop next to the Turbo, its currently 8" from it now

I find that it is constantly below the required power and the lag is awful I have played with the power smoothing turning it on and off increasing it and reducing it, increasing the intensity level also does not help, it seem to ramp up to around 70 - 90% then get “stuck”, i have monitor my cadence to see if i was spinning up before the increase and causing the Turbo to get confused but nothing i have done so far gives constant results.

I was hoping that on the longer steady intervals it would be able to track and stay close, I understand the sprint ones will be hard to ramp and down, but the 20mins ones…

I also find that the rest sets where i should be riding around 130Watt i am stuck at 170… not so restful

I have contact the support team who have been great, but i seem to remember a few years back i don’t have this problem and each new version of the App is as bad as the one before in this respect.

I love TR and credit for helping myself and family get thought the locks in the UK but they are rides where i just wish it would be better… i have tried the other App (zwift) and it seems to report better now this maybe them simple manipulating the date better, but the lag dose not seem so bad…

Please fix this Please

Thanks for lettings get this off my chest i feel better now

I am brand new to PM. I currently have a new set of Garmin Rally XC200s. I use them both on a KICKR Core and TACX Neo (train in two locations). It seems that they sort of automatically set themselves up once added to my TR devices (iPad version, BLE). I do not see any option to have them in power match mode.

These are what my settings look like. Does all this look correct?

The TR app, wherever you run it, is doing the power match magic. It is effectively reading the power from both your power meter and trainer, running those (plus possibly other things like rate of change of power from both sources - think power velocity, rate of change of rate of change - think power acceleration, etc.) through a control algorithm to generate power commands that it sends to your Neo

On your question about source: I don’t know how sophisticated the TR power match algorithm is, but from building control systems utilizing really low power processors, I highly doubt any device that can run TR is CPU limited for its control algorithm. Plus, both the Ant+ and BLT specs are pretty low data rate protocols - at least for trainer control - by the standard of today’s devices. So TR is limited by the protocol data rates

The problem with maximizing the smoothness parameter is that it will slow down the response of the trainer to changes in wattage. Imagine for example that you are doing a workout that is sets of 20x20 (20 seconds on, 20 seconds off), so if max smoothness slowed the response down so it took 15 seconds for the trainer to hit the requested power (I’m making up the 15 seconds), then in each interval you would only be doing the interval power for 5 of the 20 seconds - not what the workout is intending.

Theoretically yes. It depends upon how different you are with left / right power, how this varies with power, and fatigue

Actually it does. It’s not so much accuracy, as stability and slope of the error.

  • Power match can be affected by the Accuracy - measurement error %. If this is large, then it could create an oscillation simply because of a perceived change in power that isn’t real. Imagine that you need to put out 200 watts, but your power meter is only accurate to +/- 10%. Then TR could think you are putting out anywhere between 180 - 220 watts when you are really putting out 200. So this would induce an oscillation as TR tries to compensate for this measurement error
  • Stability: over a period of time, it the “true” power is 200 watts, will your power meter always return the same value? Or does this drift over time (e.g., due to temperature changes)?
  • Slope of the error: Think back to high school math: y (measured power) = m(slope)*x(true power) + b (offset). If the slop <>1, then the error between the true power and the measured power will vary with the power. Depending upon how far from 1 the slope is, this could induce an oscillation

Power Match is only a trainer feature, which you are setup to use.

Effectively what Power Match is is a more complex way of controlling your smart trainer instead of just sending the trainer requested power. Probably by looking at things like what is the rate that the measured power from your power meter and your trainer are diverging / converging, and factoring this into the command sent to the trainer.

1 Like