Tacx Neo 2 – How accurate is it really?
I own a Tacx Neo 2 and I have believed that it had been under reading for some time now. I bought it a few months ago and started using it when the Canadian winter hit in December. I decided to test it against a few of my power meters by running different bikes during each session to see what the power differences were. Each session came back with the Tacx reading lower, between 15-18w depending on which unit I used between my three SRM’s, my Infocrank and my Quarq.
To take a more scientific approach I decided to do a proper comparison with a structure ramp test. Each test was exactly 40 minutes long. To make sure the test was fair, I tested the Neo 2 against two power meters, an SRM and an Infocrank. If the results were inconclusive, I had the option of using a Quarq and another SRM of mine to settle the odds. You will see later on that the results were exactly the same, both units show the Neo 2 under reading by more than the 1% accuracy that the manufacturers promise. From these results, I felt that it was unnecessary to do anyb further comparative testing.
The two tests were run using a Garmin 1000 head unit. The head unit controlled the Neo 2 using a structured workout which was downloaded from TrainingPeaks. The workout included a warmup, a ramp test and then a warm down. The entire workout was in erg mode with no gear changes.
This workout was designed to see how different the power figures were between the two units when looking at the ramp test intervals which started just after 20min. Looking at an overall average over the entire workout would not be an accurate depiction of the Neo 2’s accuracy. The ramps started at 160w and worked up to 375w. The reason I chose this range was that most of my training is done between 200w and 420w.
Like I mentioned before, you can’t compare results overall as usually you will find that at certain power bands there are differences that don’t correlate with the other power bands. However, if you scroll down you will see that the difference between each ramp is very stable throughout both tests.
Ramp by Ramp
You can see that from both tests below, the slopes are almost identical. It’s the offset that’s out. In the first test it's out by roughly 15w, and in the second around 18w.
Side by side comparison
From the graphs and tables above, you can see that the Tacx is under reading considerably, but consistently, by the exact wattage. Even though it isn’t good news that the Tacx us under reading, that fact that it is consistent is positive. This means that that if I push 350w on the Tacx today, it should coinsistently be 350w tomorrow and the next day. In conclusion, the Tacx is consistently out.
Would I buy this unit again? Probably not. Here is why: I’m meticulous about my power meters and their accuracy. If I jump on any of my bikes, I like to know that the numbers I am pushing are all comparable. Secondly, to be reading below by 15-18w throughout the power range of 160w – 400w is huge, especially when you're on the lower side of that range in the L1 or L2 zone.
I like to do a few Zwift races and that difference would seriously hamper performance. Lastly, this is a really expensive unit with a claimed accuracy of 1%, which isn’t the case. It would make more sense to buy a cheaper direct drive, with a lower claimed accuracy and run an accurate power meter as the power source instead.
It’s a pity there is no calibration option, as this would likely sort out that zero offset issue.
Cleaning the data - Steps taken:
Uploaded all data into WKO
Garmin devices are notorious for dropouts when reading from indoor trainers. I therefore went through each file to see if there were any data dropouts, and if there were, I replaced the missing values with the 5 second moving average of the prior data to the drop out. To fix the files I used the “Data Spike ID and Fix Pack” Module in WKO.
Example: If you look below, there was a 4 second dropout on the SRM data from 5:21 to 5:26 and 7 seconds later on in the test.
Test 1: Total time worth of dropouts per unit:
Garmin 1000 – Recorded Tacx data – 26 seconds worth of data dropout corrected
Garmin 520 Plus – Recorded SRM data – 11 seconds worth of data dropout corrected
Test 2: Total time worth of dropouts per unit:
Garmin 1000 – Recorded Tacx data – 6+4+3+7 seconds worth of data dropout corrected
Garmin 520 Plus – Recorded Infocrank data – xx seconds worth of data dropout corrected
Follow my progress on strava or zwiftpower.
Be a slave to speed.
2018 Cape Pioneer Podium Finisher - 2nd Solo Cat 3x sub vet Gauteng ITT Champ 4x sub vet SA ITT Bronze Medalist 2x Cradle Traverse Winner ABCC - Level 3 Certified Cycling Coach