Tacx Support: 12% loss in power via drivetrain is realistic

Sounds like TACX support is getting ready to start marketing the Ceramicspeed “Driven” drivetrain

This is going to be a huge issue for the next few years I’m betting.

Not that it’s necessarily related, but I have a v5 Kickr and a set of pedals that came in this week and I’ll report back on my drivetrain loss (since the v5 Kickr can’t be calibrated either now)

1 Like

Out of curiosity, are you supplementing the Neo Bike, or making a swap and ditching the Neo Bike?

1 Like

I’m in a different location temporarily so this is a different setup.

I vastly prefer the feel of the Neo Bike to the Kickr trainer + bike, which honestly surprised me. I guess I had gotten so used to it.

Looking forward to throwing the Assiomas onto the Neo Bike though to confirm the power readings there as well.


Cool, interesting to hear that experience, and excited to hear the additional results. :smiley:

My Assiomas give me 10 more watts than either my Gen 1 Neo or my v5 Kickr. I have to admit that they come in handy on Zwift races!

So I raised a ticket with Garmin support supplying them with data from my dual recordings of my T2800 Neo and Neo 2T, measured with the same Garmin Vector 3 pedals. The Neo and pedals line up bang on, but the 2T is 3% lower across the board. Both units are marketed as being >1% accurate. So you’d expect the T2800 and T2 to be at least within 2% of each other, if not closer.

The response was “3% lower would be classed as normal in terms of accuracy”. Now the drivechain loss here is a red herring. I’m comparing two different , highly accurate (and/or precise) trainers to a reference set of pedals and there’s a consistent difference between the Neo’s. They can’t both be >1% accurate if when the pedals are reading 300W, the T2800 reports 300W and the 2T reports 291W. :man_shrugging:

This is a problem for high level Zwift racing where these kind of difference really matter. If everyone is on T2 - fine. But if someone is on a Kickr they have a pay to play advantage.

1 Like

Same bike with the same drive train installed? Fresh lube each time?

Just to add a data point here. I did some comparison testing with my Kickr Core and Assioma Duo pedals. With a moderately clean but dry and ~50% worn chain the power measured by the Kickr (hub) was about 5% lower than the pedals at 150W and about 8% lower at 300W. That testing was done in a gear that gave a nice chainline and with calibrating both power meters. After only lubing the chain and recalibrating again, the power difference dropped to a consistent 2.5-3%, which is as expected.


Yes. Recordings were on subsequent rides. Everything is the same - brand new, freshly waxed chain and new cassette.


I’m measuring two trainers of the same brand, with the same quoted accuracy with the same pedals and the same, new, clean drivechain.

The customer service person gave me some guff about accuracy being greater at lower wheel speeds. That’s not really relevant for racing.

1 Like

Depends on the brand and generation. But modern PMs are better than trainers IMO.

I don’t have data to back that up. Just anecdote with the 100+ devices we have, other athlete at TR and talking to industry folks.


I’ve been through smart trainer customer support myself, and helped a friend as well. Very hard to get them to take accuracy complaints seriously. They will always have an ‘excuse’ as to why the mismatch can be explained in some way that is not their issue. Prepare to be disappointed :frowning:

The whole reason powermatch exists as a feature in trainerroad is that smart trainers and power meters are generally not that accurate, so a workaround is needed. If all these “1%-2%” accuracy claims were true we wouldn’t see the number of complaints we see regarding this. Someone is lying, but without large scale, expensive testing we can’t say who it is :slight_smile:


I’m guessing this comes down to where/how power is determined which drives how accurately it is measured. The neo doesn’t measure the power generated by the rider, it measures speed and voltages inside the trainer. These voltages are likely measured to better than 1% allowing the power to be calculated with an accuracy less than 1%. This doesn’t mean two neos will produce the same power measurement for the same input power, but the power measured internally will be accurate to 1% for that specific unit, even if they are different.

Then surely they should make that clear in their marketing!!! :stuck_out_tongue_winking_eye:

This is exactly what accurately measuring rider power is. If it doesn’t do this within 1%, then it is not meeting the marketing claims, as this is what they are claiming.

I expect power meters to generally be much more accurate than smart trainers, as at least in the case of spider (or rear hub) based power meters the measurements that the power is based on are much ‘simpler’, and closer to the actual power generation (ie our legs) that we want to measure. In trainers like the kickr with a belt and pulleys, in addition to whatever bearings are the trainer, there is a lot of loss that needs to be estimated out of the equation. We are not interested in how much power makes it to way down in the device to actually be measured, we want to know how much we are putting out.

I have not seen any kind of testing protocol described by any manufacturer regarding how they get their accuracy claims. 1% - over what time period, what cadence range, what ‘speed’. Some trainers have been shown to be off by way more than their claimed accuracy just across different flywheel speeds. Unless they limit their claims, saying a trainer has 1% accuracy, I would expect that from about 50 to 120 rpm cadence, from gearing from 34/28 to 53/11. These are reasonable cadence and gearing ranges. My take is that these accuracy claims (especially/mostly on the smart trainer side) are hand-wavy guesses for a very limited best-case test conditions.
Maybe the rise of zwift and e-sports will drive some kind of 3rd party accuracy testing, but I think instead zwift will just make their own hardware and ‘certify’ that. They have to do something with all their free VC money :slight_smile:


They might be claiming it, but no trainer will ever measure the “rider power” to less than 1% accuracy given the uncertainty in loss in the bike drive train and coupling to the trainer. Surprise, surprise the marketing team didn’t listen to the engineers when they wrote the marketing material. Or the marketing material isn’t an engineering document with error bars and methodology.

1 Like

Like cars, there is no standard definition to what exactly “power” means from a measurement location perspective. Manufacturers in cars quote crank horsepower, but as any car nut knows, power at the wheels is another story, even among similar vehicles.

For smart trainers, they are calibrated around power at the hub and they absolutely will read lower than power at the crank spider, versus arms, versus pedals. There is no reason why a hub based power meter in a smart trainer can’t be +/- 1% accurate to its calibration in practice given the right manufacturing quality/software/etc but its measuring against power at the hub and will never be able to account for losses happening at the drivetrain.

Normally, this isn’t a big deal, because you can have internal consistency from a training perspective because you probably aren’t swapping between power meters all the time and can track progress with the same device. Similarly, it doesn’t really matter for real life racing because what actually matters is how fast you go in real life and the power number is just a point in time measurement of the real world strain. You don’t go faster or slower up a climb just because your power meter is reading low that day.

For Zwift and esports specifically, they don’t distinguish or adjust between “where” on the drivetrain that the power is being measured as it relates to how the avatar moves forward, so as a result it benefits power meters placed at the pedals and potentially dings people who use the hub based power meters. They are trying to get around this by forcing auto-calibrating trainers that can be plugged in direct like the Wahoo Kickr v5 (or whatever replaces the Neo2T) but as we can see now, it just introduces a whole other issue where folks are going to seek out the “good” trainers that read “accurately” or even a little bit high.

TLDR this is an esports problem or a problem if you specifically need to swap between different power sources. Otherwise, just use the same ruler every time you measure :slight_smile:


Did your read this, @Mone ? @Nate_Pearson said we have to buy PMs for our indoor bikes! :man_shrugging:

Nail. Head.

For elite eSports events we are required to use smart trainer as the primary power source to Zwift. Going from a Neo T2800 to Neo 2T means you lose 3% of your power. Is that why the Neo is no longer the most popular high end trainer for elite eSports athletes?

For the UCI World eSports Championships, the UCI have partnered with Garmin and are shipping every athlete competing a Neo 2T. This is a step forward as everyone will be on the same trainer.

There’s no transparency in the way manufacturers deliver their claims of accuracy (or are the actually talking about precision?), so all the talk of drive chain loss has to be a bit a misnomer. How can we be certain that in those trainers that estimate power rather than measuring it with a strain gauge don’t account for that loss in their algorithms? If you want to measure drivechain loss you need the same calibrated strain gauge on the pedal and in the hub.

For training and getting faster, it doesn’t matter what the number is as long as the trainer reports the same number every time. Hell, I used to use TR virtual power and had an FTP of 485 “watts”. It didn’t matter because it was consistent. It gave me a benchmark to set my efforts. For eSports it’s imperative that trainer report the same number not only every time, but between trainers, ie we need accuracy AND precision.