Login required to started new threads

Login required to post replies

Prev Next
PT P1 Pedals vs Quarq Riken
Quote | Reply
Thought you might like to see my own very unscientific comparison between a pair of P1 pedals and my trusty Quarq Riken:

Run 1:


Run 2:


Some notes I took:
+ the two power meters are very close, giving me tremendous faith in both power meters
+ differences are when I stop pedalling: Quarq repeats values whereas the P1s just go right to zero (**)
**Edit: This is so wrong, as Jim Meyer has pointed out. The power meter does no such thing. *Some* (possibly older) head units pad the silence with repeated power values. Apologies to Quarq and anyone who read this before this edit.
+ I'm happy with both power meters. thank you Quarq and thank you PT for providing us with solid platforms!
+ I will try the P1s while aero testing Hugo Houle (Ag2r-La Mondial rider) next week and will report back the quality of the CdA data

Link to data folder (run029 is "Run 1" and run030 is "Run 2":
https://www.dropbox.com/...-arFf5EXoAIvasa?dl=0

AndyF
bike geek
Last edited by: AndyF: Dec 16, 15 5:39
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
Thanks Andy,

Hard to tell from pictures and I havent seen these in person, but is there any room on the P1 for track straps/zip ties?
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
What the heck is wrong with you STers? You let me post a log-linear plot instead of a linear-linear plot to compare power data! How could this happen? I should have been fried to a crisp... <disappointed>

So here's the proper linear-linear plot of a ride with PT P1 pedals vs a Quarq Riken.


AndyF
bike geek
Last edited by: AndyF: Dec 14, 15 15:31
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
Yeah, surprised no one commented on the all-or-nothing zeros.

AndyF wrote:
I will try the P1s while aero testing Hugo Houle (Ag2r-La Mondial rider) next week and will report back the quality of the CdA data

Any updates there? Significant/nah?
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [Nonojohn] [ In reply to ]
Quote | Reply
Nonojohn wrote:
Yeah, surprised no one commented on the all-or-nothing zeros.

AndyF wrote:
I will try the P1s while aero testing Hugo Houle (Ag2r-La Mondial rider) next week and will report back the quality of the CdA data


Any updates there? Significant/nah?


Those are real zeroes. I just stopped pedalling. The Quarq repeated some non-zero powers (**), whereas the pedals just reported what I was doing.
(**) Edit: This is completely wrong, as Jim Meyer has mentioned. The power DOES NOT repeat power values. Some (possibly older) head units misinterpret the ANT+ protocol and repeat the power when the cadence goes silent.

The CdA data from our aero testing with the P1s was excellent! I give them a big thumbs-up.

AndyF
bike geek
Last edited by: AndyF: Dec 16, 15 5:41
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
So what does that difference mean in terms of calculating your watts on the ride? It seems the Quarq would be overestimating your effort (kJ output) by repeating the values when you stop pedaling. I believe I have read that NP calculations ignore the zero values that the P1's spit out but the Quarq is putting out actual wattage numbers when you arent pedaling which would be counted correct?
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [holograham] [ In reply to ]
Quote | Reply
Unless you are in the habit of hammering 1000watts and then suddenly not pedaling over and over it would be hard for it to significantly skew your watts for a ride. For aero-testing on the other hand, you want to be careful. I discovered this particular quarq quirk a few years back and made sure to modify my testing protocol to never stop pedaling even for any soft-pedaled bits. The quarq seems to basically be repeating power numbers if it hasn't seen the magnet go by. I haven't verified whether the non-magnet based quarqs (with newer firmware) do an equivalent thing when you stop pedaling.
Quote Reply
Post deleted by tgarson [ In reply to ]
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
AndyF wrote:
Nonojohn wrote:
Yeah, surprised no one commented on the all-or-nothing zeros.

AndyF wrote:
I will try the P1s while aero testing Hugo Houle (Ag2r-La Mondial rider) next week and will report back the quality of the CdA data


Any updates there? Significant/nah?


Those are real zeroes. I just stopped pedalling. The Quarq repeated some non-zero powers, whereas the pedals just reported what I was doing.

The CdA data from our aero testing with the P1s was excellent! I give them a big thumbs-up.

Good to know! I was just about to try some field testing with mine. You saved me from having to do a few runs with my Powertap 808.
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [holograham] [ In reply to ]
Quote | Reply
holograham wrote:
So what does that difference mean in terms of calculating your watts on the ride? It seems the Quarq would be overestimating your effort (kJ output) by repeating the values when you stop pedaling. I believe I have read that NP calculations ignore the zero values that the P1's spit out but the Quarq is putting out actual wattage numbers when you arent pedaling which would be counted correct?

The repeated values can have a very large effect on your average (and normalized) power, resulting in significant overestimation of the true results during, e.g., criteriums. This is why those "in the know" still prefer SRMs, esp. older wired ones. It's good to see, however, that PowerTap seems to have been paying attention and fixed this problem, rather than falling into the common ANT+ trap so many others have (not even SRM gets it quite right, as during the download their software removes not only the false repeated values, but also the last real number).

Lastly, normalized power includes zeros. Always has, and always will.
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [jbank] [ In reply to ]
Quote | Reply
jbank wrote:
Unless you are in the habit of hammering 1000watts and then suddenly not pedaling over and over it would be hard for it to significantly skew your watts for a ride. For aero-testing on the other hand, you want to be careful. I discovered this particular quarq quirk a few years back and made sure to modify my testing protocol to never stop pedaling even for any soft-pedaled bits. The quarq seems to basically be repeating power numbers if it hasn't seen the magnet go by. I haven't verified whether the non-magnet based quarqs (with newer firmware) do an equivalent thing when you stop pedaling.

Like, say, in a criterium? :)

BTW, it's not just Quarq powermeters that have this flaw.
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [Andrew Coggan] [ In reply to ]
Quote | Reply
Thanks! I'm new to power so excited to learn these nuances.
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [Andrew Coggan] [ In reply to ]
Quote | Reply
Heh, good point on crits. I haven't looked at the actual data, but I would have thought that in the crits I don't stop pedaling completely after hammering, but maybe I do. Usually when that is going on my attention is on not getting dropped and not on whether I'm pedaling.
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
Not sure if it's been posted but...

Any data from tests between PT P1 Pedals and Vector 2 Pedals?

-Alex
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [alexwright009] [ In reply to ]
Quote | Reply
alexwright009 wrote:
Not sure if it's been posted but...

Any data from tests between PT P1 Pedals and Vector 2 Pedals?

-Alex

It's pretty tough to run them both at the same time;)

Hugh

Genetics load the gun, lifestyle pulls the trigger.
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [sciguy] [ In reply to ]
Quote | Reply
And with that, you make a very good point.

(Quietly turns and walks awkwardly away in hopes that no one else is reading that.)
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [Andrew Coggan] [ In reply to ]
Quote | Reply
Good point.

AndyF
bike geek
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [alexwright009] [ In reply to ]
Quote | Reply
alexwright009 wrote:
And with that, you make a very good point.

(Quietly turns and walks awkwardly away in hopes that no one else is reading that.)

Damn it - your self shaming makes it less satisfying to give you a hard time about your post, which made me laugh out loud. It's exactly the kind of thing I'd write!

To the OP - thanks for the information. This makes me feel better about my recent P1 purchase. Of course I'll feel even better when BackCountry actually ships them to me...
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
AndyF wrote:
What the heck is wrong with you STers? You let me post a log-linear plot instead of a linear-linear plot to compare power data! How could this happen? I should have been fried to a crisp... <disappointed>

So here's the proper linear-linear plot of a ride with PT P1 pedals vs a Quarq Riken.


lol

you know who took about 5 seconds to think, "how come Andy is doing so much work near 1kW?", before seeing the scale below the top value.

And the instant power drop with cadence is interesting and at first pass a good thing, and will have most impact on power data accuracy during races like technical crits where surging and on/off pedalling can and does happen.

To verify the reported data, is it reporting zero power when you stop pedalling, or before you stop pedalling? Is there a pedalling rate at which it will decide you are at zero even though you are still moving the cranks?

But overall this removal of repeat values is a good thing.

Repeated power values are managed in various ways by software. I suggest those for whom this scenario matters check carefully instances of repeat power values. I know in the past the removal of such erroneous data from some crits files has ended up reducing reported power by 5 or more percent (I think I saw a 7% drop once). Indeed it's enough to cause one to rethink whether a ride really is in fact an NP buster.


The flip side to this of course is how quickly does a power meter report power when you start or restart pedalling? That's often missed effort, and in races like crits, it's often a hard effort as well.

http://www.cyclecoach.com
http://www.aerocoach.com.au
Last edited by: AlexS: Dec 15, 15 12:41
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AlexS] [ In reply to ]
Quote | Reply
AlexS wrote:


lol

you know who took about 5 seconds to think, "how come Andy is doing so much work near 1kW?", before seeing the scale below the top value.


None taken. ;-)

Quote:

And the instant power drop with cadence is interesting and at first pass a good thing, and will have most impact on power data accuracy during races like technical crits where surging and on/off pedalling can and does happen.

To verify the reported data, is it reporting zero power when you stop pedalling, or before you stop pedalling? Is there a pedalling rate at which it will decide you are at zero even though you are still moving the cranks?


I can't tell when, exactly, it goes to zero. But it seems like it's doing the right thing.

Quote:

But overall this removal of repeat values is a good thing.

Repeated power values are managed in various ways by software. I suggest those for whom this scenario matters check carefully instances of repeat power values. I know in the past the removal of such erroneous data from some crits files has ended up reducing reported power by 5 or more percent (I think I saw a 7% drop once). Indeed it's enough to cause one to rethink whether a ride really is in fact an NP buster.


It's maybe time for high-end power meters to stop taking a once-per-crank revolution approach to power. The pedal stroke generates a sinusoidal torque curve. Sampling once per crank revolution aliases out the interesting part of this action. Let's get rid of those reed switches and replace them with something more modern, why don't we?


Quote:

The flip side to this of course is how quickly does a power meter report power when you start or restart pedalling? That's often missed effort, and in races like crits, it's often a hard effort as well.


Well, power meters whose cadence sensors are more responsive will have both rising- and falling-edge improvement.

AndyF
bike geek
Last edited by: AndyF: Dec 15, 15 15:05
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
AndyF wrote:
It's maybe time for high-end power meters to stop taking a once-per-crank revolution approach to power. The pedal stroke generates a sinusoidal torque curve. Sampling once per crank revolution aliases out the interesting part of this action.


I don't think "aliasing" is the word you really wanted to use.
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [Andrew Coggan] [ In reply to ]
Quote | Reply
For the crit experienced using P1s - any concern about cornering clearance with the P1s ?
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [Andrew Coggan] [ In reply to ]
Quote | Reply
Andrew Coggan wrote:
AndyF wrote:
It's maybe time for high-end power meters to stop taking a once-per-crank revolution approach to power. The pedal stroke generates a sinusoidal torque curve. Sampling once per crank revolution aliases out the interesting part of this action.


I don't think "aliasing" is the word you really wanted to use.


It's exactly the word I want to use. Sampling at a frequency which is less that twice the bandwidth of the signal spectrum.


Edit: to sample at 1Hz, you would have to put a 2-Hz low-pass filter on the analog torque signal to avoid aliasing.

AndyF
bike geek
Last edited by: AndyF: Dec 15, 15 19:26
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
AndyF wrote:
Andrew Coggan wrote:
AndyF wrote:
It's maybe time for high-end power meters to stop taking a once-per-crank revolution approach to power. The pedal stroke generates a sinusoidal torque curve. Sampling once per crank revolution aliases out the interesting part of this action.


I don't think "aliasing" is the word you really wanted to use.


It's exactly the word I want to use. Sampling at a frequency which is less that twice the bandwidth of the signal spectrum.


Edit: to sample at 1Hz, you would have to put a 2-Hz low-pass filter on the analog torque signal to avoid aliasing.

Hi Andy(F)

Not sure I quite follow (and I mix up the terminology at times).

The sampling of crank velocity is done at a variable rate, once per crank revolution (event based sample, not duration based sample). Not sure how that causes aliasing since each rev contains a full deck of torque samples for a complete rev.

Or do you mean what happens to the variable duration data stream when it is converted to a fixed duration (ANT+) data stream? In that sense then I can see how there is a loss of data integrity, as the next fixed duration sample is not updated until a full crank rev has been completed, and the duration of fixed samples will not match the real duration of crank revolutions. Is that aliasing though, or a filtering issue (or something else)?

http://www.cyclecoach.com
http://www.aerocoach.com.au
Quote Reply
Re: PT P1 Pedals vs Quarq Riken [AndyF] [ In reply to ]
Quote | Reply
My dear friend Andy,

TL,DR:

Your Quarq is right, your ANT+ interpreter is wrong. (But don't feel too bad, because many other make the same mistake.)


EXPLAINATION:

Event based ANT+ Crank Torque messages send accumulated values, not rates. So each message is a timestamp, a rev count, and the accumulated torque. This is designed intentionally so that missing messages do not result in lost data. (If you drop a message, then you know the average power during the period.)

----
Example using speed an distance. Which data set contains more information?

time ......1s ........... 2s......... 3s ........... 4s.......... ??...... 6s
speed: ...10 ft/s.....10 ft/s.....10 ft/s...... 0 ft/s...... ???...... 0 ft/s

time .......... 1s ........ 2s ...... 3s........ 4s........ ?? ......... 6s
distance: ...10ft...... 20ft ...... 30ft...... 30ft...... ???...... 30ft
----

When you stop pedaling, then there is no new data to update, so you get the same message as last time. If the crank rev is not incrementing, then you have no new information about cadence or power until the end of the next crank cycle. This is not a flaw in ANT+, this is EXACTLY what is supposed to happen.

Another way to think about it: the ANT+ message you get tells you what happened at the END of the LAST stroke. It does not tell you what is happening now. You need to wait to the end of the next stroke to get that data.

The engineering is both sound and insightful. But there is a problem. Everyone messes this up, over and over and over. Whoever is implementing the ANT+ interpreter needs to RTFM to get it right.

As an engineer, I really want to keep Quarq power meters on the event-based crank torque message as it the technically superior solution. But we can dumb-down the broadcasts to a time-based update, which is easier for others to get right at the interpretation step.

Lets not go there. Lets get your ANT+ stream interpretation working correctly. Mark wrote some test vectors a few years ago to demonstrate these issues. You can find them here: http://opensource.quarq.us/headunit_test/

Andy, what devices and software was used to record and analyze this data?

Jim



James Meyer
Quarq Founder / SRAM
Quote Reply

Prev Next