New zwift hub one noise + low power readings

New zwift hub one. Makes a grinding noise and the power reads significantly lower than three other power sources.

Noise: It was doing from day one. I have tried now two bikes. Both with newish chains. Both with derailleur hanger checked. Derailleur trimmed to perfection. They both do the same exact noise, that sounds like grinding. It’s fine when my son is pushing with 50 watts and 80 rpm but it’s frustrating when I try to use it. The noise gets worse with power.

Power: I have three power metres (assioma duo, garmin vectors, stages) and a wahoo kickr. They all are read around 3-4% with each other (depending on power and cadence) with the assioma being the most conservative and the stages the most optimistic. ZwiftHub reads 5-6% lower than the lowest in the 300-600 watts range. It is possibly but not likely that zwift hub is correct and all the rest wrong.

Before contacting support on this, has anyone experienced problems like this? My unverified suspicion is that whatever causes the noise, causes lots of friction and losses in the drive train hence the low power readings.

PS: Firmware are up to date. Everything calibrated after decent warm ups or with ZwiftHub I tried to trigger the auto-calibration a few times. Chains cleaned and oiled and BTW, the bike is super quiet on the kickr or riding outdoors.

EDIT: A short recording (the very loud part was to demonstrate what happens when I change to a wrong gear)

I’m seeing similar accuracy issues compared to assiomas and Stages. Both those PM compare near perfect to my kickr. My hub is accurate for about 15-20 minutes then drifts lower through the ride. By 40 minutes in a 300 watt effort is 20-25 watts lower on the Hub. I have had a support ticket open for 30 days with no resolution. I have tried adding a fan directly blowing on the hub and that does not help. I am dual recording every ride for weeks and EVERY ride is like this once the trainer is warmed up.

1 Like

Hi there, did you get an answer to this. I have similar rumbling/grinding noise with new trainer and my watts feel down in the 250 to 500 watt range.
My sprint is fine as is Z2.

Hi Casey, just wondering are you doing an automatic spin-down calibration while riding?
I’m wondering if its doing temperature correction and then not recalibrating once warmed up. I feel like I’m getting similar 20-30watts reduction in power.

Steve, I have tested about everything I can think of. If I deliberately coast to get an auto calibration my results are generally worse. Yesterday at the end of the Fondo I did a 311 watt effort for 2 minutes according to my Assioma’s. This same effort reported as 289 watts on my Hub. I had coasted multiple times long enough to auto-cal around the 60 minute time.

I have also setup a box fan on high blowing directly on my trainer. This offered roughly 1% improvement on my testing.

I tried Erg mode at 300 watts vs free riding at 300 watts and those produced the same results.

Nothing I have found will get me under a 6% drift lower as the Hub heats up. Auto-calibration tends to make it worse and deliberately cooling the hub makes it slightly better.

All of my comparisons over the last 5-6 weeks are on my Zwiftpower Analysis tab.

frustrating, that’s a 7% difference. You would normally see some 3 or 4% due to drivetrain losses.
I seem to be missing 20-30 watts, most noticeable when I do similar 3 to 4 min efforts at threshold. Are you using virtual shifting?

Yes, I specifically bought the hub to get virtual shifting and 10hz updates.

I have another forum post here with more details and have done extensive testing on the hub and did those same tests with a kickr. The results are pretty clear that at power levels > 200 watts the hub drifts down 8-10% from the start of a ride through about 25 minutes. If you ride < 200 watts and don’t let it auto-calibrate it will report within the 2.5%.

Attached is a small snippet of a test. You can clearly see on the 1 minute efforts the blue line is clearly low but the 150 watt “rests” the lines are pretty close. On this test that highest interval was 1 minute avg 351 on the Assiomas and 326 on the Hub (7.8%). This particular test had an auto-calibrate coast done just prior to this power test segment and it was after 90 minutes of a group ride avg 215 watts. This identical test prior to the 90 minute group ride had the 1 minute 350 watt interval report a 1 watt difference from the Hub to the Assiomas.

The most frustrating at this point is I can’t get Zwift to communicate with me at all.

Interesting chart, are you able to replicate this without using virtual shifting? Not sure if you have a cassette you can put on to the trainer and shift manually.
Also was this at similar cadence, just thinking of variables to isolate.

I don’t have another power meter to do a comparison, but this chart confirms my suspicions about what I’m feeling since having my trainer swapped-out.

I have not mounted a cassette to turn off virtual shifting. I probably could test this. I have not deliberately tested cadence but I have done this test protocol with near identical results for nearly every ride since 10/23. The main variables I have tried to chase where 1. With/without auto_calibrating 2. Adding cooling to the trainer (big fan) 3. Using erg mode and not using erg mode. None of these variables got the results anywhere close to acceptable. I did do 2 rides on my kickr with the same protocol and it tested great. This reinforces it is not my comparison device but the Hub that is drifting during the ride.

The base resistance on the flats in zwift is probably sufficient for me to test this in the one gear that is mounted. I could just turn off virtual gearing and try to test with my bike being a single speed. I have doubts this changes anything.

That would be good to try as it might be the firmware for shifting. Or maybe its adding to internal temperature increase over mechanical gearing.

Although, my old hub with virtual shifting was fine for a few weeks - but it developed a calibration fault so Zwift swapped it.

It might be coincidence, but since then the perceived effort to produce 300 watts has got harder. I’ve noticed it more since 1st November, so similar time period.

I’ve got a 9 speed cassette fitted, so I’m also going to try manual shifting and test perceived effort at 300w intervals

Hi Casey, were you able to record with virtual shifting off and did you get the same accuracy drift?

Hi, I did not get the chance to send the support request yet. I will hopefully send it this weekend.

Sorry - Late responding. I did get a response from Zwift support and my hub has been returned for a full refund. I did not do a dual recording with a cassette and no virtual shifting prior to returning the hub.

1 Like

I also have a new Zwift Hub One. When comparing the Zwift Hub One Power Readings with my left crank powermeter, I got lower readings from the Zwift Hub by 8% to 10%. This is quite a lot. I also realized, that cadence readings have been around 4% lower.

Today I added a cadence sensor and configured zwift to use it. Cadence Readings are fine, now. Power readings are a lot more consistent. Zwift Hub readings are now ~2% lower, which is expected.

It seems, that the cadence guessing of the Zwift Hub One is pretty bad, directly affecting the power values. Still fine for me, if adding a cadence sensor solves this problem.

It seems, that the cadence guessing of the Zwift Hub One is pretty bad, directly affecting the power values. Still fine for me, if adding a cadence sensor solves this problem.

That’s very interesting but puzzling. Usually power measuring devices are reporting already computed power over ant+ or ble which means they have done their own internal maths or whatever estimations some trainers do. An external cadence sensor would not affect those. I was also under the impression that trainers don’t have cadence sensors (they can’t directly measure your pedal rotation, only the rear hub/cassette) but they derive cadence from the fluctuation of the torque during the pedal cycle, hence it’s not always very reliable.

I will try to give it a go before I sent it back. Zwift is replacing mine, I presume mostly because of the loud grinding noise and not because of the readings.

I fully agree, that an external cadence sensor should not affect the power measurement. I will do some more rides and some more testing.

Todays ride shows again big differences, even with cadence sensor. Below 250W it’s pretty accurate. At 300W I get ~8% higher readings from my PM, at 400+W it’s getting more than ~12% difference.

Edit: I might try using a cassette instead of virtual shifting. Let’s see.

1 Like

Bad luck :frowning: I am curious if such fluctuation is a result of the auto-calibration feature not working as advertised. Pity they did not include a manual calibration too. In my case, the low reading was consistent from the beginning to the end and of every ride.

I mounted a cassette instead of the cog. As I learned, that using a MTB with Zwift Hub and cassette is clumsy and needs some tweaking (e.g. QZ App), I used the cassette with virtual shifting.

Today I rode a course with four laps, virtual shifting enabled. Each lap I used another gear on the cassette (15-17-19-22). Surprisingly the results are pretty good, difference never exceeds 1.5%. The setup is less noisy as well.

When disassembling the single speed cog, I realized, that the screw for the cog might not have been tight enough. I did not even need a chainwhip to open the screw.

So - next test: remount the cog with a properly tightened screw.

1 Like

Very interesting especially I have also noticed 20 watts power loss when using built in “power meter”.