Kurt virtual power issue, power too low...


(Joe Kipfer) #1

I am using my ibike speed/cadence sensor to use the zpower while riding zwift on a K road machine.   I’ve noticed that my power numbers are pretty consistently 15 to 20% lower on zwift compared to what my ibike is reading…also I ran side by side with trainerroad and it’s showing the same thing, when I’m pushing 240 watts on ibike virtual power or trainerroad virtual power I’m only pushing 205 or so on zwift.  I always felt kind of slower/sluggish on zwift and finally decided to check it.  I thought maybe it was my ibike but I downloaded trainer road just to see and sure enough it’s the same thing.  Now I know I could buy another ant+ stick and rebroadcast TR but I’d rather not spend the money on a second stick and spend money on a trainer road subscription just to get my power numbers correct, is there anything I can do at this point?


(Christopher Pallotta) #2

Without a power meter (with strain gauges) that measures power vs estimates it, you’ll never know which is more accurate. For what it’s worth, I’ve tested my Kurt Kinetic in Zwift using virtual power while also recoding the rides with my SRM/Garmin.

At first I had slippage warnings almost constantly and the difference was about +50 watts from Zwift/KK vs the SRM actual power. 

On on the next trial, I moved the ANT+ sensor much closer to my speed sensor and fixed the slippage warning issue. After many laps, the estimated power was remarkably close, all things considered. Overall, though, it still measured about +8 watts average over several laps. 

So, in my experience, virtual power with a Kirt Kinetic on Zwift is fairly accurate, but with a bit of over-estimation. 


(Ron Skinner) #3

it would have to be that the Zwift has a numeric error in the formula for translating the wheel speed to power curve.  Technically in Zwift there is no way to alter the power curve, so the power you see will be biased until Zwift decide to change that. Until they allow bluetooth reception the KK in ride pod is not going to help you either.  So maybe the best thing is to send in a request straight to Zwift tech and see if this is something they know about, since there must be numerous people with the same resistance unit and the same problem, particularly since you have 2 other preferences that show the issue.  

Do you have the iBike Ant+ power out option?  If your iBike is the iBike+ maybe it will transmit out the power reading as if to a Garmin? Just a thought not certain it would work.  Maybe check with Velocomp?

 


(Joe Kipfer) #4

Hmm, I wasn’t aware ibike had a power out option, that could easily be my solution if I could figure that out, I’ll check into it, thanks!  

 


(Christopher Pallotta) #5

I’m sure Zwift is simply using the power curve that Kurt has researched and developed to be most accurate. I don’t see how they could improve on that and things are likely to be as accurate now as they’re ever going to be for estimated power without a spindown calibration option.

I agree, though, that when using the KK pod which allows for a spindown, more accurate results could be achieved as this will largely zero-out factors such as tire inflation, tire rolling resistance and roller tension. If Zwift were to be able to read from the pod, results should be better. 

I know some have had good results with the ibike, but it’s important to remember that it’s still just estimated power. Based on my experience, I’d say the OPs numbers from Zwift are *probably* more accurate as my tests show that, if anything, using a KK with Zwift will result in higher wattages than actual. 

I should note that I have great confidence in my SRM numbers. It’s recently been calibrated, it’s extremely consistent and it closely matches another Stages power meter I own.


(Ron Skinner) #6

Hi Christpher, though I tend to agree, my point was 2 applications that both use virtual power curves from the KK road machine in the OP were consistent while the Zwift number was off.  This suggested that it might be the Zwift calculation.  On reflection it might also be a time lag issue.  In the TR and iBike cases the power is done locally, iBike reads speed and translates in real time, as well TR is working resident on the OP’s computer.  Though I am not 100% certain since the Zwift app seems to not be working in a local mode, maybe the difference is the time it takes to go to the Zwift server and back.  That could account for a reading difference, but it would be scattered both +/- and not biased to one side or the other, in other words, if the final average power for all three were the same then it might be a time lag issue.  If the average power is lower for Zwift vs the other 2 apps then it is likely something off with the Zwift application of the power conversion calculation, or the OPs wheel diameter is not being translated the same for speed at the Zwift side?  I know I have played around with wheel diameter as a work around before when the virtual power curve was not right on.


(Christopher Pallotta) #7

I see where you’re coming from. In my experience there is no time delay to speak of comparing local vs. Zwift.

Wheel circumference could absolutely affect the power estimates, though, as the power curve estimate is only accurate if the captured speed is accurate. Wheel circumference will be a factor. 

That actually brings up a good point I hadn’t thought about. When I did my tests with estimated power on Zwift I don’t recall having the option to change wheel circumference. That would mean Zwift is making an assumption, probably for a typical 700c wheel. Perhaps the wheel circumference Joe entered into TR or the ibike is different that what Zwift is using. 


(Joe Kipfer) #8

My wheel circumference for both my ibike profile and trainer road profile is the standard 2096mm…I just re-verified.  I wish there was a way to adjust that in zwift.  I really don’t have any other explanation other than something weird with zwifts power conversion of my speed sensor.  It’s so strange.  I’m still looking in to using my ibike to transmit ant+ to zwift.  Looks like you can but i may need to purchase an upgrade.  I originally just trusted the zwift numbers, but then when i seemed to go a bit slower than my actual outside speed I chalked it up to just the off season…then after a month and a half of hard zwifting it was enough for me to check it, then when i cross referenced with another KK road machine power curve in trainer road and I got the same thing I now believe the zwift numbers are wrong here for me.  


(Christopher Pallotta) #9

Joe, definitely open a support ticket with Zwift on this if you haven’t already. That’s the best way to get an answer. Curious what they’d have to say. 2096 would be the correct number in most cases, I’d say. 


(Joe Kipfer) #10

Opened a ticket, will keep you all posted on what they say.  Thanks to all of you for your help!


(Nigel Van de Velde) #11

Hi,

In previous discussions about the kk powerprofile one of the zwift employees mentioned that they didn’t use the power curve formula provided by kinetic.

Back then they explained to me that they needed a different formula to deal with accelerations. Your rear wheel speed doesn’t correspond immediately with the power you put on the pedals during short accelerations. I get that and love the fact that they correct for this. Yet this should only be for accelerations. For steady pace they should stick with kinetic’s formula… Now it seems low indeed. In real life I am a decent cyclist, whereas on zwift I just suck in comparison to others.


(Joe Kipfer) #12

Nigel,

That’s really interesting…I’m sure the accelerations would be even harder if I was riding the pro fly wheel.  It kind of makes sense, i guess.  Sorry you’re also feeling my pain.  On group rides I was getting dropped after 4 or 5 laps.  My FTP is 217 (3.2 wpk) and in actuality I’m riding 240w or so just to get it to register 200 watts on zwift, if i lose focus and lose the draft for a split second it’s over because I’ve already been over threshold and LTHR for like 35 to 40 minutes and have nothing more to give to chase.  It can definitely be demotivating, but then again, I’m a beast right now outside compared to what I normally am this time of year.  :)  


(Joe Kipfer) #13

Well I upgraded my ibike to include the ant+ power out feature, but neither zwift nor trainer road are picking it up.  They will still pick up the speed/cadence sensors, but not the power meter.  Bummer.  Should have just gotten a second ant+ stick for the money I spent on the upgrade.  In the mean time, I’ve been going down the list of trainers in zwift to see if any of the other trainers virtual power closely matches what i’m getting from the kurt power curve.  Some are close, but none are close enough for me to ride with integrity…most push out way too many watts.  Looks like the only way to do it is for zwift to use the kurt curve or to rebroadcast from trainer road with a second stick.


(Christopher Pallotta) #14

Joe, I think it’s important to remember that everyone riding a Kurt Kinetic on Zwift is subject to the same power calculation as you. And again, my results show that the Kurt Kinetic estimated power is an inflation of actual power as recorded by my SRM and Stages power meters. 

Yes, I can understand your frustration that your power is not matching across different software platforms. That can make tracking and setting your power zones difficult. 

But, when you suggest that you’re frustrated because you’re getting dropped after a few laps, again, remember everyone else on the island is using Zwift power profiles. I don’t think it’s correct to assume that you’re at any disadvantage. From where I sit, using my SRM and Stages power meters, you’ve got an advantage. 


(Allen McGregor) #15

I’ve been using a Kurt Kinetic Road Machine (2010 model) with a PowerTap hub on Zwift. I’ll be taking the trainer tire off the PowerTap wheel soon to use it outdoors, and was curious how close zPower would be.

I did one lap with a Garmin Edge 500 reading the PowerTap speed and wattage, and Zwift reading speed from a Garmin GSC 10.  I uploaded the Edge and Zwift files to Strava:

https://www.strava.com/activities/273641979 

https://www.strava.com/activities/273639768 

The Zwift/zPower average wattage is 227 compared to 243 for the Edge/PowerTap in my case. About 7% lower.

I used 2 full turns on my Kurt Kinetic knob after tire contact. Pressure was around 80 psi.

I’ve had my PowerTap manually verified for accuracy. Could there be significant manufacturing differences in the Kurt Kinetic over the years leading to these varying results? 


(Christopher Pallotta) #16

Allen, great test. As I wrote above, I did a similar one with my SRM vs Kurt Kinetic and my results showed about 7 watts average higher for the Kurt. My tire pressure was about 115, though, vs your 80. I just read somewhere that Zwift recommends tire pressure be no more than 100 psi. Maybe 100 psi is the magic number?


(Nigel Van de Velde) #17

Hi Allen,

Thanks for your input. I use 3.5 full turns and 100psi on the Kinetic. 2 full turns would definitely give me tyre slippage with my Vittoria Zaffiro Trainer Tyre.

I don’t have a power meter to do comparisons, so it’s somewhat hard for me to prove that Z-power is rather low. I do have the benefit that my outside rides are generally totally flat and without much traffic. So I do have a good idea where my power output should be at certain heart rates, just based on speed to power conversions.

 

 

Yesterday for example, I did a steady ride, solo and at low intensity.

 https://www.strava.com/activities/273307511

Avg heart rate = 129bpm ; avg speed = 28,9 km/h ; Reali life elevation = 5m (one bridge over a highway). Wind was stable at 3 beaufort. So headwind and tailwind were equal during the ride. My aerobic fitness is quite high, so generally I don’t have any decoupling towards the end of easy rides. No hard corners and I pedalled for 96.2% of the ride.

According to speed to power calculators that you can find online (e.g. http://bikecalculator.com/index.html or http://www.cyclingpowerlab.com/PowerComponents.aspx), both calculations correspond to a power output of 160-165W. This is a value that I’ve seen for previous rides as well, so I guess a heartrate between 130-135 bpm corresponds with a power output of +/- 160W for me.

 

When I compare this to values I get with z-power, than generally I notice power is quite a bit lower. Following ride has low intensity intervals between 130-135 bpm and all of them result in an average power output between 135 and 140W when using z-power.

https://www.strava.com/activities/273932749

In real life this power output would have given me an average speed of only 27km/h on a flat ride.

 

When I compare with Kinetic’s virtual power for the same ride (recorded with a second ANT+ stick), than the low intensity intervals give me an average power output between 155 and 160W.

https://www.strava.com/activities/267036264

In this case I do get the expected correlation between heart rate and power output.

 

For my setup, I think Kinetics virtual power is more realistic than z-power. And I guess that’s only logical. They’ve been experimenting with there resistance units and curves for years…

Kind regards!


(Nigel Van de Velde) #18

That’s indeed the most annoying factor. To be honest I don’t care about the numbers in comparison to others, however I do want to compare to my own data from previous seasons. Now it’s just too complicated: Having 2 ANT+ sticks, using 2 programs, open and start recording in a specific sequence to get the data in both programs, merging the files afterwards before uploding to strava,… I almost spend more time on all of this than on the actual training itself. 


(Joe Kipfer) #19

Got a response on my ticket.  

 

"

Crystal H (Zwift)
Mar 24, 18:49

Hi Joe,

Thanks for writing in and great question.

Trainerroad reads the power reading directly from your iBike which is why they are showing the same number, not because those two programs are coming up with a more accurate number than Zwift.

Zwift calculates Zpower based on an algorithm factoring in your height, weight, the specific trainer you are on and drag of the resistance unit, we also factor in the virtual terrain and virtual wind resistance.

On the road, the Newton’s “Physics Engine” converts air pressure, accelerometer and speed measurements into opposing wind, hill slope, acceleration, frictional forces. When you are riding indoors there is no accelerometer, speed measurements, opposing wind, hill slope or frictional forces which will make your indoor power reading inaccurate on the iBike.

Sometimes you will feel sluggish on Zwift because you are climbing a virtual hill, or hitting a virtual headwind.

Running Trainerroad at the same time as Zwift can cause an interruption in Zwift being able to perform and is not recommended or supported.

I hope this answers your question. Please let me know if you have any other questions or need clarification on anything.

Ride on.

Crystal H"

 

 


(Nigel Van de Velde) #20

Hmmm, seems like a standard answer form on any type of question about zPower. Not impressed with it.

I can’t see why power output would be influenced by weight or heigh. Weight and height should have an influence on your speed, not on your power.

Why don’t they just use the same formula as Kinetic provides us with and correct this for short accelerations. Could be so simple and so good…