1. Home
  2. Blog
  3. Linearity Showdown: OG vs 2 vs 2T
  1. Home
  2. Known Issues
  3. Linearity Showdown: OG vs 2 vs 2T

Linearity Showdown: OG vs 2 vs 2T

Tacx announced the release of firmware version .34 for Tacx Neo 2 on January 29th, 2020. That firmware update brought a frequently requested update to Neo 2 linearity that made a significant difference at higher flywheel speed.

Neo 2T received an update for linearity in firmware version .31, but there was never a mention of any such update for Neo OG. This left some wondering how the three generations of Neo compare to each other with respect to linearity.

What Is Linearity?

Simply stated, linearity is the ability for the trainer to resist a power target such that the resulting graph is a straight line regardless of flywheel speed. When we’re talking about a power target in this context, we’re specifically talking about erg mode, where a power target is set, and the expectation is that the trainer will “hold” this power target to relieve the rider of this task. This is a stumbling block for all trainers, but some are better at it than others.

Testing Setup

Trainer Hardware

Before we began the tests, we performed some maintenance to help remove some variability from our testing.

The Neo OG we selected for our testing was using an Edco branded freehub, while the Neo 2 was using a Tacx Shimano freehub that came with the unit.

For both Neo and Neo 2, we removed the existing freehub and simple bearing, cleaned the axle, and replaced the freehub with a brand new Tacx Shimano freehub. The freehub was shared between them so we were testing the same freehub and cassette. The freehub, axle, and simple bearing were greased with Park Tool Polylube PPL-1.

We installed brand new Shimano Ultegra 6700 10 speed 11-28 cassettes on both the Neo/2 freehub and the Neo 2T. The cog sizes for this cassette are 11, 12, 13, 14, 15, 17, 19, 21, 24, 28. Both were torqued to 40Nm with a Norbar 130101 torque wrench.

Trainer Details

Neo OG
ModelT2800
Year2016
Serial Number801617192
Firmware Version0.7.4 / 0.8.4
Neo 2
ModelT2850
Year2018
Serial Number800002959
Firmware Version0.0.34
Neo 2T
ModelT2875
Year2019
Serial Number800048240
Firmware Version0.0.31

Power Meter

For all tests, we concurrently recorded power from Favero Assioma Duo. Both pedals were running firmware version 04.25

Initially, at and when switching between trainers, the pedals were zero offset with the crank arms in vertical position.

Data Collection

For all tests, we used Garmin Edge 520 cycling computers. Both were using version 13.00 software.

One of the 520 computers was dedicated to recording the cadence and power from the Assioma power meter via ANT+.

The other was updated for each trainer to record ANT+ Power from the Neo unit currently being tested, and cadence data over ANT+ from a dedicated RideSense magnetic Speed/Cadence sensor connected to the bike.

Software

For these tests, we decided to use Zwift on MacOS because it has a very simple workout creator interface that can be done within the app, and also allows us to test other trainers outside of the Tacx ecosystem. While we are focusing on the Neo family of trainers, we will (in upcoming tests) show that the same phenomenon exists across all trainers, including Kickr 2018 and Saris H3.

The workout consisted of a single 120 minute interval at 100%. We artificially set our FTP to 100 watts to give us a consistent 100w workout. The cadence reminder was set to 90 rpm.

The Power Source and Controllable Trainer settings were connected via Bluetooth to the Neo unit being tested. The Cadence setting was connected via Bluetooth to the RideSense dedicated cadence sensor.

Bicycle

We used a dedicated trainer bicycle, a Giant TCR Advanced SL0 Di2 10 Speed 50/34.

Before the tests, we cleaned the derailleur cage and replaced the jockey wheels with new Tacx Shimano 11T Standard pulleys and Shimano OEM screws.

We also replaced the chain with a new Shimano 6701 10 speed chain and connected the ends with YBN QRS10 link.

The chain was lubricated with NFS using the 12:12:12 method. We deliberately added a drop of NFS to the QRS10 link.

This unit has an updated RideSense speed/cadence sensor with ANT+ / Bluetooth capability that was used to transmit cadence via Bluetooth to Zwift, and ANT+ to the Garmin Edge 520.

Interval Timing

Intervals were tracked with a Gymboss interval timer set to 1:00.

Test Procedure

  1. Start the interval tracker. During the first minute, engaged Zwift erg mode for the first minute and ramp up to 90 rpm cadence.
  2. At the interval signal, both Garmin Edge computers were started at the same moment.
  3. The bike was pedalled continuously at 90 rpm as closely as possible.
  4. At each additional interval signal, the derailleur was shifted outboard by one cog, consecutively, until each cog had been recorded for exactly one minute.
  5. The “garbage minute” occurs in the middle of each recording where the derailleur was shifted back to the largest rear cog and the front chain ring was switched to the large ring. The cadence and watts were intentionally dropped to zero to help scaling issues in the output graph, and then pedaling resumed to re-engage erg mode in Zwift.
  6. The previous steps 3 & 4 were repeated whilst pedaling in the large chain ring.
  7. Both cycling computers were stopped when all cogs had been recorded for exactly one minute.

Note: because the 28T cog has the benefit of not being shifted into during data collection, it doesn’t have the “shift bump” power spike that all other cogs have. This could be equalized in a future test result by running the cogs from smallest to largest.

Results

How To Read The Graphs

Each test began at a certain “second” in time, so to find any complete value, select from the initial minute:second, and select to the next minute:second.
Start Seconds
Neo OG: 47
Neo 2: 13
Neo 2T: 21

Each graph starts in the small chain ring and largest cog, so left to right, there are 10 minutes of small chain ring. Then there is the “garbage minute” which has been excluded from our results. Then there are 10 more minutes of large chain ring. Any extra data at the trailing end beyond the 1 minute mark has been excluded from our results.

Neo OG

Neo OG vs Assioma.
Full Results: https://analyze.dcrainmaker.com/#/public/5830866a-bb65-4dfc-47ea-47b7d75daab6

There were no surprises in store here. Neo OG has been known to have this defect for quite a long time, and the results are consistent with expectations.

The average difference was 4.06%, with the closest fidelity at 0.21% in the 21T cog and 34T chainring.

Neo OG – 34T Chainring Line Graph

The worst fidelity was a whopping 31.36% in the 50T chainring and 11T cog.

Neo OG – 50T Chainring Line Graph

While the trend lines for these graphs may appear similar, don’t let that fool you. You have to consider the scale as well. In the small chain ring, the largest difference was 13 watts. In the large chain ring, the largest difference was 34 watts.

While it’s interesting to see how the Neo tracked compared to the power meter, the really important metric — and the thing we cared about in the first place — was how closely the Neo and power meter tracked to the initial target of 100 watts. That’s going to tell a similar story, but when you see in expressed in watts, you get a different appreciation for what’s really happening.

Neo OG Distance From Target, 34T Chainring
Neo OG Distance from Target, 50T Chainring

When you look at the Distance From Target graph, you can see right away that Neo is clearly underreporting, since the watts from Neo are very close to the target, but the watts from the power meter are stratospheric.

Neo 2

Neo 2 vs Assioma
Full Results: https://analyze.dcrainmaker.com/#/public/8f1dc7e1-0c47-47b0-42d3-345000127f5f

This was our second effort with the Neo 2, the previous having come the day after the .34 firmware was released. The result is very similar to the first, though not exact, which makes sense because we made changes to the testing apparatus.

The average difference was 3.76%, slightly better than Neo OG. The best fidelity was 0.15% in the 19T cog and 34T chainring.

Neo 2 – 34T Chainring Line Graph

The worst fidelity was 24.47% in the 50T chainring and 11T cog.

Neo 2 – 50T Chainring Line Graph

Looking at the target fidelity, the Neo 2 definitely does have some improvements, holding very near to the target and the power meter in the small chain ring through the middle of the cassette, but again, the Neo underreports its value.

Neo 2 Distance From Target, 34T Chainring

Here again, we see the issue in terms of watts, where Neo reports a value below 7 watts at its maximum distance from target while the power meter climbed up to nearly 31 watts. While this is still a vast improvement over the original Neo 2 with pre-.34 firmware, it’s only slightly better than Neo OG. In fact, in the small chain ring, it’s actually a little worse, as we’ll see shortly.

Neo 2 Distance From Target, 50T Chainring

Neo 2T

Neo 2T vs Assioma
Full Results: https://analyze.dcrainmaker.com/#/public/949c238f-6623-4f8d-4f73-25006dbbca5c

The Neo 2T we tested was still using .31 version of firmware, which includes the original linearity fix. The firmware updates in .32 and .34 do not affect the outcome of this test (or shouldn’t!) so we did not update the Neo 2T before testing.

While the minimum difference between Neo 2T and the Assioma pedals was very similar to both the Neo 2, that is where the similarities end. In the small chain ring, the closest fidelity was 0.67% in the 17T cog and 34T chainring, but the average difference was a mere 0.24%! That is because for some portion of the time, Assioma read lower that Neo 2T, netting a very low average. Why that happened, and to this extent, is not yet understood. Even so, the maximum difference was 3.04% which is dramatically different from OG or Neo 2 — and that was in the small chain ring!

The result from the small chainring was so surprising that we’re going to have to run it again or a few times. There’s nothing to say that it can’t be like this, but it’s definitely not what we were expecting.

Neo 2T – 34T Line Graph

Moving on to the large chain ring, we see another round of differences between how the Neo 2T behaves compared to Neo OG or Neo 2. The values in the large chain ring tracked so closely that you can barely see daylight between the digits. That’s not because Neo 2T was significantly better at holding the requested resistance (it wasn’t); the highest average value reported by Neo 2T was 126.6, just a small step behind Neo 2 at 130.8.

The difference is that Neo 2T reported the output more realistically., creating a situation where the average watts reported by Neo 2T was higher than the average watts reported by either OG or Neo 2. In essence, Neo 2T is simply more truthful about its inability to resist low watts at high flywheel speeds, reporting the watts more faithfully than Neo OG or Neo 2.

On paper, that may cause fidelity to appear more reasonable with Neo 2T, and while the values were definitely lower, the behavior is effectively the same as with Neo OG and Neo 2. The grass is not that much greener.

Neo 2T – 50T Line Graph

When we look at Neo 2T from the Distance From Target perspective, you get a very keen idea why I said initially that Neo 2T really shined. While I feel that Neo 2T slightly over reported watts with Neo 2T, the fidelity to the target was such that the training impact of such differences would be very difficult to measure. This is a very, very nice result in the 34T chainring

Neo 2T Distance From Target, 34T Chainring

Where Neo 2T really loses linearity at higher flywheel speeds in the 50T chainring, it at least maintains credibility by reporting watts much nearer to what our power meter was reporting. The actual watts themselves were closer to the target as well, so overall.

Neo 2T Distance From Target, 50T Chainring

Summary

To really get an appreciation for just how good Neo 2T is, you have to see the trend lines in watts between all of the trainers side by side. In the 34T Chainring, Neo 2T is far and away the most linear. When you examine the trend lines and notice that the Neo 2T is the only one that’s even close to horizontal, there is a very clear leader.

Trend lines from the power meter. Since we know Neo and Neo 2 produced unreliable results, we have to rely on the power meter itself to compare the trainers. Neo 2T is nearly horizontal, with an average of 1.5 watts, which is a very, very nice result! Note also that Neo OG outperformed Neo 2 in this sample.

When it comes to the big chainring, though, Neo 2T has its achilles heel nicked. Although the numbers are lower overall than Neo OG or Neo 2, the trend is still similar. In this case, Neo 2 and Neo OG again reverse roles, with Neo 2 edging out the Neo OG, and the Neo 2T edging out both. In the big chain ring, linearity is just an issue that will probably not be resolved any time soon.

Trend lines from the power meter showing how the trainers performed in the big chain ring. The results are pretty much what’s expected, with Neo 2T holding on well through the 14T cog before letting go.

If you’re going to use erg mode for your power based workouts, we still recommend using the small chain ring and a straight chain line, mid cassette, and occasionally vary the rear cog to distribute the wear more evenly.

Start in the 21T cog and conduct your workout. Next workout, shift to the 19T cog. The next workout, shift to the 17T cog. Then go back to the 21T cog and repeat for subsequent workouts.

This will keep your watts in a reasonable range while adding longevity to your indoor drivetrain.

Full Results

https://docs.google.com/spreadsheets/d/1XDAhATRmKOPEU3WHyWLa4Yr8194THEyAwHboO1zAIhs/edit?usp=sharing

Updated on February 3, 2020

Was this article helpful?

Related Articles

Comments

  1. On all 3 trainers, you hit an issue not related to linearity on the large chainring – you are clearly hitting the trainer’s resistance floor, hence it becomes unable to maintain the requested 100W against the Assioma measurements, and then both trainer-measured and pedal-measured power go up. This happens because you chose a power level (100W) that is not very realist for most use, and you drive this low power to very high spindle speeds.

    I would suggest doing the test series around 200W, and throwing out all results where the trainer is unable to meet the power request.

    1. Strongly disagree. While this test was 100w, this issue was previously demonstrated at 200w, 250w, and 300w. Furthermore, the wattage floor of Neo is 30w. 100w is 333% higher than the wattage floor. To suggest that 100w is not typical use is extremely “you” centric. There are many small female cyclists for whom 100w is sweet spot work.

      I would suggest doing your own tests and let us know how it goes. 🙂 We have plenty to do already, but we’d like to see your results.

      1. I did something similar this afternoon with my Neo OG. Instead of staying at 100W for the entire workout, I created a Zwift workout that repeated these 16 times (8 times in 36T and 8 times in 52T): Free ride for 30 seconds, 100W for 1 minute, 200W for 1 minute, 300W for 1 minute. The test went through these eight cogs on the cassette: 25T, 23T, 21T, 19T, 17T, 15T, 14T and 13T.

        Note: ANT+ dropout during the 52T – 19T 200W test so there’s no data for that.

        Testing setup:
        Tacx NEO OG firmware level 0.2.1/0.7.4/0.8.4
        Cassette on NEO OG: Shimano Ultegra 6800 11-speed 11-28
        Quarq DFour91 power meter firmware level 7 (latest is 10, as I found out after the test). Power meter was “zeroed” prior to the start of the test.
        Bike: 2019 Specialized Venge Pro with Dura-Ace 52-36 chain ring
        Bike computer: Edge 530 (firmware 5.34 beta) paired with Quarq DFour91 (cadence provided by Quarq)
        Bike computer: Edge 520 (firmware 13.00) paired with NEO OG power meter (cadence provided by NEO OG)
        MacBook Air running Zwift. NEO OG was paired with the Zwift app running on the MacBook Air as smart trainer.

        Results:
        The 100W intervals in both 36T and 52T seem to match up quite OK with your results. When the chain was at 23T and 21T, power data from NEO OG and Quarq DFour91 matched very well in both big and small rings. Just like in your charts, as the chain moved toward smaller cogs, the power discrepancy between the Quarq power meter and Neo OG grew.

        However, for the 200W and 300W intervals, when the chain was on the small ring (36T), the power discrepancy actually narrowed as the chain moved toward the smaller cogs. This behavior is different when the chain in on the big ring (52T). The smallest difference between the two devices happened when the chain was on 17T.

        I have included the Dropbox links to the Excel spreadsheet (sorry, I don’t have access to DCR’s analyzer), the graphs (red lines in the chart are Quarq DFour91, green lines are NEO OG), and a screenshot of the Zwift workout.
        https://www.dropbox.com/s/hezdvd2njytr8l2/NEO%20OG%20DFOUR91.xlsx?dl=0
        https://www.dropbox.com/s/qda5ppam7gyx9ge/NEO%20OG%20Zwift%20workout.png?dl=0
        https://www.dropbox.com/s/315vj51lmp8f7ak/NEO%20OG%20DFOUR91%2036T.pdf?dl=0
        https://www.dropbox.com/s/wmm6ayofgagm01u/NEO%20OG%20DFOUR91%2052T.pdf?dl=0

        I plan to repeat this test again, with a modified workout and with my just updated Quarq DFour91 power meter. This is my first time using workout in Zwift and I had assumed during the “free ride” interval, there would be very little resistance. Turned out, it just meant the ERG is off during free ride. That created an issue during the test when the chain was on the smaller cogs as it became increasingly difficult to recover from the 300W effort 🙂 So next time, I will replace all the free ride part of the workout with a 50W interval.

        1. Cool stuff, thanks for sharing. If you feel like, drop us a note with your workout files and we’ll send you back a link to the analysis.

          If you decide to test against a 50w interval, be aware that that’s getting close to the wattage floor, and also that you will need to keep your cadence pretty high so that you don’t fall below the 12 kph threshold at which Neo intentionally reduces resistance.

          -Drew

          1. @Drew: I’ll do that after my next test. Thanks!

          2. I repeated my test last week using a slightly updated Zwift workout. I replaced the 30-second free ride intervals with 50W intervals. I also updated the firmware on the Quarq Dfour91 to the latest.

            I made a mistake during the test as I was staring at the screen, forgetting to change gear during the 30-second 50W recovery interval, from 36-19 to 36-17. So part of the 100W effort meant for 36-17 was actually done in 36-19.
            These are gear combinations involved during the test: 36-25, 36-23, 36-21, 36-19, 36-17, 36-15, 36-14, 36-13, 52-25, 52-23, 52-21, 52-19, 52-17, 52-15, 52-13. The interval for each gear combination are: 30 seconds at 50W (recovery, change gear), 1 minute at 100W, 1-minute at 200W and 1 minute at 300W.

            Here’s the link to the DCR analyzed data:
            https://analyze.dcrainmaker.com/#/public/9c4132f4-27b7-4ab8-5e13-d683a70466c8

            I am also providing the gear shifts during the test:
            http://di2stats.com/rides/view/62246

            If you look at the power comparison graph, it’s pretty clear there does not appear to be a perfect gear combination (using 52-32 chain rings and 11-28T cassette) that allows NEO OG to hold power across a range, even for the same gear combination.

            At lower power (100W), 36-25, 36-23, 36-21, and 36-19 seem have the least differential. But these gear combinations result in bigger differential at higher power output (300W).

            On the other hand, the gear combinations in the small ring 36-14, 36-13 seem to have the smallest differential at higher power output (200W and 300W) but the accuracy suffers at lower power output.

            In the big ring, the NEO OG performs reliably well across the various cogs in higher power output, but the lower power and mid power suffer toward the smaller cogs.

            For my setup, if I have to choose the best gear combination for my TrainerRoad workouts, it’s probably 36-15. I don’t mind the inaccuracy at lower output since my sweet spot is between 200W – 300W. I just can’t remember how noisy this gear combination is 😆

          3. I accidentally left out the combination 52-14 in my previous comment. It’s in the test.

  2. Some of my own testing with the 2T and Assiomas matches these results. While the recommendation to conduct erg workouts in the small ring and middle of the cassette still holds, it looks like with this particular setup that the big ring, middle of the cassette is also a decent, if occasional, option. Thanks for the work.

  3. Hi Drew. The tests I have done so far seem to indicate that some gear combinations have better fidelity than others across a range of targeted power. I have a few questions:

    1. Can we assume that the gear combination that has the best fidelity is the same for different models of trainers and different brands of trainers?

    2. If we cannot make that assumption, how do you think one should validate the ERG mode of a trainer as part of a review? If for example 36-19 has the best fidelity for Trainer A while 36-21 is the one for Trainer B, the ERG mode test for Trainer A will have a better outcome than Trainer B if the reviewer uses 36-19 for all his tests. I posted similar questions to a couple of the most popular and respected sport gadget sites/YouTubers and so far, I have heard nothing back.

Leave a Comment