Bug: ERG mode has intermittent drops in resistance and reported power (Wahoo Kickr)

When I use Zwift in workout mode, I might have, say, a 5 minute interval @ 250w (keeping this simple for illustrative purposes). I set the turbo trainer firmware to “power smoothing/average”. I set Zwift to “3sec power average”. I enable ERG mode in Zwift. I start the workout.

My expectation when I hit the 5minute interval is that effectively I’m making a “bargain” with the trainer - I will keep pedalling and it will give me the correct resistance to keep me at 250w. That isn’t happening (consistently). What actually happens is that the trainer holds a relatively constant resistance (sometimes for up to 5-10 minutes) but then, for no reason that I can fathom, the resistance drops very significantly for one second (two max) before resuming.

When this happens the reported power (both as shown on screen in Zwift and as reported in the fit file data) dips. The odd thing is that it dips to around 66% (2/3) of the value it should be holding. This amount does vary slightly sometimes/there are exceptions but say 90% of the time it dips consistently by the same amount. So if holding 250w, it drops to 160-180w briefly. Less than a second and the normal reported power and resistance resumes.

This is intermittent and happens throughout a ride. I have noticed that it seems to happen more frequently the longer that I ride. So it could potentially be due to the Kickr “overheating” in some way (though its very cold here currently and the room I ride in is extremely well ventilated and this is still occurring).

However my suspicion is that its something about my “pedal dynamics”, cadence or bike position and my pedal motion changing or cadence. I say that because I’ve noticed that if I sit bolt upright to, say, have a drink for a minute and end up checking my phone messages or whatever, when I get back in the hoods almost without exception there’s been no drops in that period. The only difference would be my cadence slightly lower, torque slightly higher and pedal motion more force on the “down” stroke and less “round”/back. I don’t know how kickr works but I read somewhere (DCRainmaker??) that it measures “chain tension” as a way of calculating both power and cadence (somehow) so - and this could be a stretch - I’m wondering if it can’t properly interpret a smooth pedal action and needs the “choppiness” of a strong down motion followed by slack to read power consistently??

My only other hunch is that I can seem to trigger the issue if I spin fast for a few seconds and then reduce cadence - almost like I “get ahead” of the flywheel somehow (the flywheel is speeding faster than I’m pedalling due to inertia) and it thinks I’m applying no power?

Is anyone else seeing the same? And can anyone who is more technical than me and understands the “logic” of how ERG mode works explain this to me, this isn’t correct behaviour right? Specifically I don’t understand

  • If ERG mode is on, the trainer should never stop providing resistance so long as I am pedalling, right?
  • If 3 sec avg is turned on, and pwoer smoothing turned on, why is the power reported not getting smoothed out even with this issue occurring? A series of 250,250,250,250,160,250,250 (power readings each second) shouldn’t be possible with power smoothing/averaging turned on, right?

I’d love to hear if other people are experiencing this, not only from people who have answers/suggestions please.

Technical stuff/point to note:

  • I am currently using an iPad Pro (iPAD OS). Its patched to latest version. But issue occurred on my last device too so is not linked to the device
  • I use a Wahoo Kickr v6 turbo trainer but issue occured on my old Kickr v5 too so doesn’t seem linked to a faulty kickr unit
  • I use wireless connection to link Kickr to iPad, but issue also occurs (no difference) when I use bluetooth BLE instead
  • Latest version of OS and Zwift, although problem has been continuing for over a year across various versions
  • I have tried using the Kickr for cadence detection, as well as a Wahoo dedicated cadence sensor on pedal. Same issue regardless.
  • No modifications to Zwift
  • No modifications/jailbreak or whatever its called to the OS, its as Apple provides it
  • This issue happens even with trainer difficult set to 0%/off so is nothing to do with declines/downhills.

Screenshots to help illustrate:

How it appears in Zwift while riding (in realtime):

How it appears in intervals.icu when its reading the fitfile after ride is uploaded:

A sample of the raw data from the .fit file:

3

To rule out a trainer problem, I suggest doing an ERG mode workout using something other than Zwift such as the Wahoo Fitness app.

1 Like

Thanks @Paul_Southworth - I absolutely believe this is a trainer bug yes. I managed to duplicate it using the Wahoo Fitness app. Doing a workout in that wasn’t exactly a barrel of laughs (its just a plain screen) but I could produce the same issue. I have already logged this with Wahoo support (who are not really being that useful/helpful so far).

In terms of cause it seems to be (by my testing, Wahoo support not helping a bit so far) caused by how the Kickr determines if power is being applied, it seems to only work at certain pedal angles of attack or however you would term it. I watched a video by DCRainmaker and GPLama where they explain how the kickr uses “chain tension” to calculate power and cadence. It seems as though you can kind of spin up the flywheel and “get ahead of it” so that even though you’re pedalling, the flywheel is spinning faster than you are pedalling so the chain tension is loose, hence it reads zero power. I think that’s whats happening, anyway. It seems more likely to occur when my cadence changes quickly (either up or down) such as slight changes if I reposition from hoods to drops (or vice versa) or to a bolt-upright position. I also can’t duplicate the issue as easily if I sit riding bolt upright the entire time (but seriously, that’s not exactly a practical workaround! :smiley: )

My question for this forum is mainly

  1. if anyone else is experiencing it since I can’t believe that I’m the only one that’s hit this bug, espcially since it seems related to kickr firmware/flywheel and its maths/calculations rather than a single faulty unit *since it happens on both my kickr v5 and kickv6 and

  2. why the Zwift 3SEC power average setting isn’t working, given that it doesn’t smooth out the reported power as expected when this issue occurs?

What does fit file say about cadence drops at the same time?

3-sec Zwift smoothening is screen only, nothing to do with fit file recording.

Edit: 2 Kickrs in my house (v5 and '18) and I have cadence sensors on both. I’m a mtb’er so all over the place power wise but I can still trick Kickr’s cadence easily. Daughter uses v5 and she is just super smooth.

The fit file shoes no cadence drops. As the original post states, I have also used a bluetooth pedal-mounted cadence sensor to read cadence (instead of the kickr) and the problem remains, unchanged.

Fine, but its not smoothed on the screen either - the power drops are displayed in realtime on the screen, without smoothing/averaging.

Have you tried uploading a log file to zwiftalizer.com to see if it notices a Bluetooth dropout from the trainer? It seems unlikely since you’ve seen this on two trainers, but I don’t have other ideas at the moment. It does seem like Wahoo owes you an answer.

Thanks yes I do not believe it to be a bluetooth dropout given that other bluetooth devices (headphones, HRM) never experience dropouts, and that I can make the problem go away by sitting upright and changing my pedal stroke/pedal dynamics and cadence.

Mostly other devices have no dropouts, but Zwift paired ones have. Zwiftalizer.com is always a good option when having issues.

I am confident this is not a bluetooth issue. I have explained why:

  1. This happens when the kickr v6 is paired by network, not BLE
  2. Other bluetooth devices also paired to zwift (HRM, cadence sensor) never see dropouts
  3. I can make the drops go away by changing my cadence/pedal angle of attack.

Please stop making the same suggestion over and over. Its not bluetooth :sweat_smile:

??? It was my first (and last) suggestion.

I can’t quite parse what you’re saying - are you saying that you see the same issue (power drops) with your kickr? To be clear, my cadence reported stays relatively consistent even when the power drops occur, within reason, its primarily the chain tension an angle of attack on pedal that is changing, as well as flywheel speed.

Nope - never get power drops. I use cadence sensors so that correct cadence is recorded/displayed. Its a minor niggle that did my head in when I first tried the inbuilt cadence. I’d ride without cadence if I didn’t have a couple of external sensors hanging around.

When I asked earlier about fit file recording cadence, it was curiosity around whether it was a data stream issue. If fit file shows no corresponding cadence drop that correlates to the power drop then that would indicate that the data stream is fine from Kickr to Zwift and points to Kickr(s).

Couple of fyi’s - you likely know this stuff, but if not…

  • When Kickr’s are in ERG/workout mode, they are sending your power data back to you/Zwift smoothed if ERG smoothing is set that way on Wahoo app.
  • People were having data issues on the V6 with wifi until a firmware sorted them fairly promptly. They seemed more severe than yours so just mentioning it

So a couple of things you could try - do a workout without Wahoo’s ERG smoothing on (your onscreen will look a bit choppy but you may still be able to determine if drops are occurring), and then repeat similar intervals as a free-ride.

Yes I also have a pedal-based cadence sensor, and I get the same issue when I set Zwift to read cadence from this rather than the Kickr, no change to the issue seen.

The fit file shows no cadence drop at the time, no.

ERG smoothing is set on the Wahoo app yes. But the power does not appear smoothed, and the 3sec avg does not seem to work. If it did then sequences of “250,250,250,170,250,250” should not be visible in the fit file (but they are).

The firmware is patched to the latest via the Wahoo app. But the issue is present on both by Kickr v6 and also my Kickr 5 and I’ve had the former over 18 months on a number of different firmwares, with the issue present througout.

I’ve already tried the workout without ERG smoothing on (several times) and the data is so choppy its hard to tell if the issue is present.

yes it should, the 3 second average is only for the display while you’re riding not the actual power that is recorded, so the actual power would be in the fit file not the average figure.

But what I’m saying is that the same sequence shown on screen matches the fit file. I see that 170 flash up on the screen for a second, even with smoothin on in Wahoo app and “3sec avg” turned on in Zwift.

Take a look at my very first post - you can see the drops (blue dips) in the workout graph which is shown on-screen, in real-time in Zwift - and they match the dips in the fit file (as shown by the intervals.icu screenshot).

What is just as irking (and maybe needs a separate bug report) is that even in ERG mode when you’re not in control of the power you are nagged to “increase power” by Zwift. Every. Single. Time. the power drops. Like - OK - tell me how I’m supposed to do that exactly, Zwift!

So I think I may be getting to the root of the issue after some experiments today.

The problem disappears almost entirely if I ride in “small-big” gearing while in ERG mode, i.e. in the little ring at the front (e.g. 34T), and a big cog at the back (e.g. 28T). There’s an element of cross-chaining so its not great as a long-term solution, but I can’t seem to reproduce the issue when riding like that.

I’m not technical enough to understand what reason there is for this, but hoping someone might be able to explain?

Below is an interval from today, riding in small-big (as described above). I saw perhaps 3-4 drops throughout the entire 4+hour ride, versus something like 50+ drops for the same duration when riding in my normal gear.

Confirms that its diddly squat to do with bluetooth! :smiley:

It would be great if someone else with a Kickr could do a ride with the below parameters and see if they can reproduce the issue:

  1. Set up a workout with a long interval in ERG mode (say Z3, 20 minutes)
  2. Put the bike (on Kickr) in big-small (50T front, 11-12T rear)
  3. Ride in the hoods/drops, but occasionally sitting upright and then dropping back down again
  4. Try and vary cadence, speeding up to 110+rpm, then dropping cadence to 60rpm, then back up again a few times, ideally while also changing ride position.

I’m pretty confident you’ll be able to reproduce the issue seen - the power will drop low by 30-100% and the interval go “red” with a warning to “increase power”. Can someone do this and verify?

Can you post the links to the zwiftalizer files for the two different rides.

Hi @Gerrie_Delport_ODZ I am happy to share information necessary for debugging, but my privacy is important to me. I’m not posting a fit file on a public forum since my understanding is that these can contain PII, correct? I am happy to post screenshots of any relevant information you require from Zwiftaliser for the two ride files, or to share a zwiftaliser link with official Zwift support staff only.

Have a look at one and see if there’s anything that concerns you. It seems pretty benign to me. It doesn’t include the fit file data.

EDIT: To me the most important thing is whether YOU see any anomalies in the Zwiftalizer results, or anything that coincides with the trainer behavior you reported. If not then it’s probably not relevant.

what is your normal gear?

for erg mode i go for small ring at the front then about 3 down at the back so the chain is in line.
bigger gears can make erg mode less reliable as smaller changes in cadence or pedalling a bit harder on one stroke has a bigger impact and it can try to overcorrect