Power ? ? Kurt Kinetic Road machine

(' BTCNJ-Jim-HUBS(D)) #1

Received updates this morning. From yesterday to today the power seemed to vary a lot while on straights and uphills ,
on straights with constant rpm’s power went from 140 to 220 back and forth. Do not know if I am imaging it or something in program has changed . Yesterdays ride I averaged over 18.8 mph 20 mile ride today only 14.8 mph avg 30 mile ride . Granted first couple laps took it easy but seems something is different . Riding Kurt Kinetic Road Machine

(Maurizio Gigliotti) #2

Hi. zwift is possible to know your testing methodooly ?

1- rear tire inflated to X PSI
2- Kurt Kinetic Road Machine rear roller set to X turns
3- weight cyclers

(' BTCNJ-Jim-HUBS(D)) #3

Realize that , had same psi , same 5 and 1/2 full turns . My weight did not change from yesterday to today 189 lbs. Will try again tomorrow . May have just had a bad day . :slight_smile: Thanks for the quick response . I really like what you guys are doing . Cheers

(Maurizio Gigliotti) #4

Sorry I’m a beta tester, I ask them zwift guys for ZPower test protocol.

(' BTCNJ-Jim-HUBS(D)) #5

Now that I re-read your post … Light bulb goes off . got it

(Jason Hurdlow) #6

Kurt Kinetic RnR here… It’s not just you. After today’s update I had really jumpy power too. Having ridden for a couple years with virtual power in TR, I know it wasn’t just in my head. I would be holding a steady power output and the power in Zwift would be jumping all over the place. 173… 131… 215… all within a couple seconds. Also it did seem to average considerably lower than before the update and in my estimation (haven’t tried to see if they’ll run concurrently) lower than I would have seen in TR. I would echo previous inquiries into tire type, pressure, number or turns, etc… I was running 4 full turns with probably (didn’t check) 120lbs in a Vittoria Pro Home Trainer tire. I’m going to try and borrow a friend’s PowerTap and see if I can compare numbers/dial-in the pressure and turns, but the jumpiness needs to get worked out first.

(Steve Valero) #7

I also experienced strange results on my latest ride, glad to see it wasn’t just me.

Also being a long time TR user I am used to off days but this was slightly different. I thought I was just used to the power smoothing (and subsequent slight delay) but I know that my RPE did not match the numbers.

(Phillip Davies) #8

Same issues seen as everyone, power jumping around wildly and a lot less output than usual and felt much harder work that it should have to hold anyone’s wheel, not sure if this is due to the fluctuating power meaning the game thinks i am pedalling erratically instead of a constant higher power

(' BTCNJ-Jim-HUBS(D)) #9

I see in the update release dated 1/15 a major fix was they greatly improved Zpower coefficients for KK and cyclops . Hopefully readdress this issue and move it back up some closer to where it was . It was only slightly off before on the high side.

(Phillip Davies) #10

That’s the update that caused the issues notes above, hopefully they can address them soon

(Andrew Santos) #11

Just did a ride today and after the update the power numbers seem way lower than they should be. Did a steady 20min interval and plugging average speed into the KK formula gives me a higher number.

(Jon Mayfield) #12

If you have jumpy numbers, it might be interesting for you guys to post a link to your strava upload and also let us know what speed sensor you have. We’re actually trying to calculate power a bit more intelligently than just speed->watts, and if you’ve ever ridden with a power meter before you’ll know the numbers jump all over the place within a range. This isn’t your standard virtual power. Seeing your data would be a good way for us to see if what you’re seeing is normal or not.

As for the power numbers maybe being lower than you’re used to, the KK power coefficients supplied by Kurt Kinetic themselves seem a bit high (often 40 watts high!) so those aren’t what we use.

(Andrew Santos) #13

Before the update, the avg power on a given interval seemed about 25-30w too high, now they seem about 25-30w too low, when compared to KK’s virtual power.

(Jason Hurdlow) #14

Jon - This isn’t just a slight fluctuation, this is quite bad, and definitely not right. It’s bad enough that last night I did my workout with other software because I just couldn’t bring myself to deal with it (it makes it un-fun). I appreciate your desire to make a better virtual power conversion, but it’s definitely not there yet. The other thing to consider is that users are going to be moving back and forth between Zwift and other software and would like to see consistency between them. I don’t want to have my workout numbers changing between products, it makes them pretty useless. So at a minimum I’d like the option to choose the virtual power conversion between your algorithm and what everyone else is using. If I had to hazard a guess (I’m a software developer) it looks like either some part of the calculation is out of order and a causing floating point precision error, is being done with integers instead of in floating point, or there is a LUT mapping error. I have ridden with a power meter before and never seen anything like this. Here’s my workout in Strava: https://www.strava.com/activities/242032651 Another thing we need to know is what tire pressure and turns on the trainer you are using. I’ve based mine on what others with power meters have found to make TrainerRoad virtual power match their power meters. I’m thinking about ordering another ANT+ dongle and borrowing a friend’s Powertap so I can compare values in real-time.

(Onno Pierik) #15

Last night on the trainer (Rock n Roll) I noticed the same. I also had my garmin Edge 800 connected. The speed on the display of the Edge 800 was very consistent while the power of Zwift varied big time. Wheel speed was the same, it nothing else than a software issue.

I can post both fit files. Is it possible to post them in this topic?

(Jon Mayfield) #16

Jason Hurdlow, the strava activity data looks pretty ok to me - the fluctuations I see are within 50-60 watts, which is a little more than I see when I use a powertap or quarq.

Again, the speed on your trainer is only a part of the equation as far as power goes. If you spun your kurt kinetic up to 40mph and then coast down while slowly pedaling at 50 watts, everybody else’s equations would show you putting out hundreds of watts for a very long time after you let off. Additionally you’d never be able to sprint for the line as your power would be lagged by 4-5 seconds - Zwift would be much less interesting.

I believe I was one of the first to apply realtime virtual power calculations (fall 2010), so I certainly understand how everybody else works regarding this, and we are trying to move beyond that.

Anyway, things will only get better with you guys giving feedback like this, and our power calculations will get smoother. What would be good is if you guys could post up which model of speed sensor you have so we can try and reproduce some of the signal noise issues here.

If you could ignore the noise of the display, I think you’d find the power is very responsive and can hopefully see where we are going with this.

(Onno Pierik) #17

Ah indeed, as a virtual power only user I’ve to see it from a different way (the right way)

To calculate better power you use the speed differences which impacts power a lot.

What about a bad internet connection? (low bandwidth, may be sometimes unstable) Does that impact the power?

And maybe a feature for virtual power. Multiple magnetics in the rear wheel. When using 2 or 4 magnetics you get u higher resolution. Set it up when pairing devices (so Zwift can divide the speed by the amount of magnetics) and zPower becomes more accurate I think.

(' BTCNJ-Jim-HUBS(D)) #18

Jon , I am using a Garmin GSC10 . Which file do you need ?
Would it be more helpful for me to try to run a Zwift training ride in parallel using my Garmin Edge 500 .

(' BTCNJ-Jim-HUBS(D)) #19

Fun Factor …
Jon , I agree with Jason Hardlow’s comment , Took the fun out of it , with new power formula . For Zpower on the KK machine should match TrainerRoad workouts. You made a comment that you thought KK was 40 watts higher . I think if your formula before 1/15/15 included that 40 watts. If you split the difference and added 20 watts back in this would represent my real world riding outside .

(Steve Valero) #20

I have a KK w/ a GSC-10 sensor. Also I have the new style separate units and a separate system that could be used in tandem during a workout if needed to highlight any differences.

I do have to say that I think my ride on Sunday (1/19) was far more stable than the on the morning of 1/18.

It seems that since TR does not have any gradiant calculation the speed and power variation found in the Zwift algorithm would make more sense. I will go against the trend and say whatever change was made between 1/18 and 1/19 seemed to dial the numbers in and stabilize them very well.

Thanks for giving this quick attention :slight_smile: