. After going back and looking at your graph, I think we were also talking about “slightly” different things. Your graph shows differences as a chain wears. My data is based solely on the current condition of a chain. 2% being a chain cleaned and lubed. Up to 10% for same chain that needed cleaned/lubed. Some of my results were even recorded on a 12hr+ ride, in which I ran into some rain. And I even noticed the 2% to 10% within that single ride.
I don’t think a linear percentage is possible. At lower powers it is probably a higher percentage. It will also be different depending on the chain speed for the same power output.
I actually keep an eye on the discrepancy between crank and Neo 2. At 300 W and in my usual gears with my usual cadence I see about 4-5W difference. If it goes up I look why. Once some cling film had got in the rear derailleur and it was 10W.
12W I can see as being possible if you really don’t look after your bike. Maybe people on average don’t.
I would add there’s a difference between indoors and outdoors. To see a 10% difference strictly on an indoor bike, I’d agree that may not be possible. But riding outdoors, it is possible within a single ride depending on weather and surface type. It would be rare to see that indoors, unless you do a ride outside in crappy weather then bring your bike indoors without cleaning/lubing.
Trainers aren’t consistent 
.
5 Likes
Also, interns can be inconsistent (as a former intern I know). So there’s a lot of opportunity for inconsistency there.
1 Like
To conclude: turns out the pawl spring in the Neo hub had snapped in two, making the pawls less able to engage with the disc and losing a lot of efficiency. I bought a new spring and the drive train loss is now a more expected 2-3%, tested against two different power meters.
Tacx support stuck to their guns till the end (12pc!!) and actually told me to look up tests done by ceramic speed despite these tests showing that 12 is not realistic.
Got to admire the effort made to close the ticket though.
Appreciate the help, cheers.
4 Likes
Hi,
I buyed tacx neo 2t recently. I feel and believe that it underrate the power in my roadbike. Test it with my mountain bike with quarq xx1, chain new (50kms) + cassette new and ring used. Use an erg mode and power workout in tacx software. The difference in watts was 10%, watt lost is proporcional (k constant) it not a fixed value (30watt less is not correct, 30 watts in 300 watts, 20 watts in 200 w, 40 in 400w).
Very very interesant the analysis, garmin 520plus with quarq xx1 (in mountain bike) vs tacx neo 2t with tacx software in program - workout with erg mode.
10% is … mmm a lot. But is very consistent in all workout. 10% may be in an big cassette (11-40 shimano xt), a heavier chain (sram 11v), an chain ring some wear and full suspension locked (trek 9.7), but derraileur very wear. The test was done in fixed gear (erg mode and cadence regulate the total power).
I need test it in a very pro and clean bike with power meter, 1% accuracy is the tacx neo 2t promise.
hi there your issue is more than likely your pm; i suspect your 4iii is a single sided power meter therefore its ability to read your “true watts” will be flawed; 8 to 10% off is not uncommon; further more i always take drive train loss at about 3% assuming your drive train is in a good state. no idea what tacx are on about saying such high numbers? an easy way to fudge the fact many riders power meter and turbo dont match maybe!!
I dual record a lot of my races on Zwift. I use a Neo which is paired to Zwift and initially Powertap P1S, before switching to Garmin Vector 3. Chain is waxed. Nowhere near a 12% drive chain loss. That’s just utter nonsense.
1 Like
Could you elaborate on that? PleasePleasePlease 
Not every trainer reads the same. 
. Testing one doesn’t validate or invalidate them all.
You could probably get 30 or so and get some numbers that are statically significant.
4 Likes
Now I may have the terms mixed up, but when a manufacturer claims >1% accuracy, I expect all the trainers to read within 1% of each other.
1 Like
1% accuracy means that the powermeter will provide a value that is within +- 1% of the true value.
So at 300w
1% = 297-303
2% = 294-306
When you have different measurement points, you start having a complex set of ranges depending on the scenario.
Pedal > crank arm > crank> hub in terms of general stack rank of power loss.
So if pedal is showing 300, it means the range is 293-306w. If you take a few percentage points from that for an immaculate drivetrain, you could get 290w at the hub, but then apply the range which gets you to 287-293w.
Assume the worst case of high on pedal and low on hub: 287w to 306w which is 6.5% loss which is just due to the combination of power meter error.
So it’s conservative but not out of the realm of possibility that Tacx claims 12%, since all you have to do is bump the real drive train loss up.
Now this doesn’t actually mean that the loss is 12%, its just that the power values being provided back have a potentially larger range and that if its outside the realm of 12% that is their threshold for looking into it from a customer support perspective to replace the unit
1 Like
I always get confused with accuracy and precision!!
If they state accuracy >1%, then all Neo’s should be a <2% difference across the board between all Neo’s?
The number on the trainer is only part of a wider body of evidence. I’ll report back after the next Zwift race which is in 35 mins!!
2 Likes
IIRC the Neo’s are pretty automated from a manufacturing standpoint, so they are all likely calibrated using the same hub based testing rig. You can see it here in @GPLama’s video : https://youtu.be/RGGZujCvEoc?t=242
How that translates to consistency across machines is unknown.
So you’re saying QC is generally better in the realm of power meters when compared to trainers? Do you have an example?
After delving ridiculously deep into this including dual recordings from both Neo1 and 2T, comparing more than one powermeter to Neo1 and back to back testing with the same PM comparing Neo 1 and 2T. I’ve also considered the power numbers as part of a wider body of evidence which includes outdoor rides, comparison with others outdoor data on specific segments on Strava and my own efforts on Strava (including a reference climb I have done over 200 times) with more than one powermeter. Also looked at Xert data from my dual recordings, comparing this to my own RPE.
After all of this my brain is completely frazzled!! I’ll go into detail about my reasoning when I’m a little fresher, but my conclusion is that however you choose to describe the reasons, the Neo 2T reports lower than my Gen 1 Neo by around 3% across the board. I’m not the only person to think this way, but I had read somewhere it was fixed with firmware updates. The Gen 1 Neo was considered the gold standard by which all other trainers were measured. But for those of us on Neo 2T that race at elite level on Zwift where we have to use the trainer as the primary power source, this is a problem.
4 Likes
Looking forward to it! 
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