Stages Or P2M Power Meters?

RChung

meanredspider wrote:

What I haven't understood (and I'll read your post and Robert's posts again) is when accuracy is important: when it's important to know that the power measured in absolute terms is 310W (rather than xxxW+2 ie 2W better than the last measurement). Anyhow, I'll have another read later.

Since you work with medical devices, you'll understand that there are situations where all that matters is relative accuracy and some where absolute accuracy matters. Absolute accuracy will matter when you need to make comparisons across individuals, or for the same individual across different equipment (which can be because you're testing the individual across different times or venues). Absolute accuracy will matter when you're comparing against clinical guidelines that dictate standards of care. Absolute accuracy will matter when treatment dosages can be either ineffective if too low or toxic if too high. So I'm sure you can figure out situations where absolute accuracy in power measurement will matter.

@Robert - thanks, too. Yup, for sure you can do various static checks on the strain gauges. Maybe I'm making it more difficult than it needs to be for a PM but the type of test you talk of simple wouldn't cut it the medical device world even for the simplest things. All you're really doing is a static calibration of the strain gauge array.

No, I'm also talking about dynamic checks, too. Even dynamometers have to be calibrated against known forces. But, in a greater sense, we can swap head units, we can measure rho, we can measure cadence, we can measure temperature to isolate the contribution of the power meter itself.

What I'd like to see, as I said above, is how the various systems (including comms and head units as Alex mentions) perform in a range of conditions of load, cadence, changing temperature (humidity and air pressure would be good), L vs R loading etc As Xav, I think, said - if you're paying a lot of money for an expensive system, you need to know that it's coping with all of this.

I'm pretty aware of the contribution of the cadence. That was one of the initial problems with the Stages.

Ray found in one of his tests that a couple of the meters drifted horribly when the conditions changed. I'd like to know that for my day in the Alps when the temperature started at 0C peaked at 33C and dropped back to 10C my PM was still telling me the truth, for instance.

Sigh. Everyone points that out as if I weren't aware of it. If you re-read Ray's reviews you'll see that I am one of the brain trust who helps him set up his test protocols and analyze the data. Ray hates doing that stuff. In fact, I was the one who did the analysis on that particular test and first spotted the issue (along with the Stages' problems with cadence measurement). The conditions were unusual and extreme, and all it means is that when the temperature varies a great deal you should either coast occasionally in order to trigger the automatic re-zero or else manually re-zero if the terrain precludes it. That's the real take-away message: if you have a power meter that's known to be sensitive to temperature change then re-zero (in fact, even in the situation where you're doing a race up a hill and cannot re-set the torque, once you get to the top and start to coast again you can often spot the difference in torque zero and fix the data post hoc). Unfortunately, there is no way that coasting occasionally or manually stopping for a few seconds re-sets one's bilateral asymmetry, and I don't know of any way to fix that post hoc. That's a problem of an entirely different nature.

slowbike

Grill wrote:

Keep in mind that there's a lot of data out there unpublished because why release your numbers and watch others reap the benefits when you're the one who put the time and money into it. It's rare to see power data from top amateurs as well as pros and rarer still to get a glimpse at their training plan. This holds truer still with aero data.

There are a few (club level) testers I know who will not publish their rides on Strava purely because they don't want their power data published ....

Personally I don't care too much as I think the only detriment to me may be the lack of surprise when (if!) I start next year putting out X% more power than I achieved this year because I will have been pushing all my ride data to Strava over the winter. (Of course, I probably won't be faster and that'll come as no surprise to anyone!)

Grill

Slowbike wrote:

Grill wrote:

Keep in mind that there's a lot of data out there unpublished because why release your numbers and watch others reap the benefits when you're the one who put the time and money into it. It's rare to see power data from top amateurs as well as pros and rarer still to get a glimpse at their training plan. This holds truer still with aero data.

There are a few (club level) testers I know who will not publish their rides on Strava purely because they don't want their power data published ....

Personally I don't care too much as I think the only detriment to me may be the lack of surprise when (if!) I start next year putting out X% more power than I achieved this year because I will have been pushing all my ride data to Strava over the winter. (Of course, I probably won't be faster and that'll come as no surprise to anyone!)

Sure. I post all my data for the same reason. If I did pay for a couple hours in the tunnel I'm not sure I'd be terribly keen to just tell people my findings though.

meanredspider

RChung wrote:

Since you work with medical devices, you'll understand that there are situations where all that matters is relative accuracy and some where absolute accuracy matters. Absolute accuracy will matter when you need to make comparisons across individuals, or for the same individual across different equipment (which can be because you're testing the individual across different times or venues). Absolute accuracy will matter when you're comparing against clinical guidelines that dictate standards of care. Absolute accuracy will matter when treatment dosages can be either ineffective if too low or toxic if too high. So I'm sure you can figure out situations where absolute accuracy in power measurement will matter.

Absolutely - in the job I used to do for nearly 10 years, we had 10 million patients per year and 4.5 billion tests per year in my part of the business alone. But it's why I also understand that for most things physiological (and I take power generation by humans as being one of those things), absolute accuracy really doesn't matter in the vast majority of cases. There are far too many variables. Precision, on the other hand, was important.

RChung wrote:

Sigh. Everyone points that out as if I weren't aware of it. If you re-read Ray's reviews you'll see that I am one of the brain trust who helps him set up his test protocols and analyze the data. Ray hates doing that stuff. In fact, I was the one who did the analysis on that particular test and first spotted the issue (along with the Stages' problems with cadence measurement). The conditions were unusual and extreme, and all it means is that when the temperature varies a great deal you should either coast occasionally in order to trigger the automatic re-zero or else manually re-zero if the terrain precludes it. That's the real take-away message: if you have a power meter that's known to be sensitive to temperature change then re-zero (in fact, even in the situation where you're doing a race up a hill and cannot re-set the torque, once you get to the top and start to coast again you can often spot the difference in torque zero and fix the data post hoc). Unfortunately, there is no way that coasting occasionally or manually stopping for a few seconds re-sets one's bilateral asymmetry, and I don't know of any way to fix that post hoc. That's a problem of an entirely different nature.

Yup - I know you're involved in that stuff which is one of the reasons I raised it. My point is that if Ray can "stumble" across "issues" (that I'm not sure users are aware of - Ray's piece suggested he wasn't and he's pretty thorough in understanding a product before he tests it), there just might be other problems. And I don't know that his example is that extreme - my day in the Alps saw a much bigger variation - my commute in Scotland probably did. Even household products are pretty rigorously tested by consumer organisations like Stiva in Germany (look up their water tightness protocol as an example). It would be nice to see our expensive power meters given a harder time than Ray is able to to with his limited means and resources.

Anyhow, I've said enough on this topic. My Stages "can opener" has done all I need of it. I won't ever criticise those who feel they need more. We're all different and have different needs and goals.

RChung

meanredspider wrote:

I don't know that his example is that extreme - my day in the Alps saw a much bigger variation - my commute in Scotland probably did.

I described the condition as extreme and unusual because he took the bike from a warm apartment into the freezing winter air so the difference was large and sudden. Even in the Alps the temperature tends to change more slowly than that, and even in the Alps you occasionally get a chance to coast which would have re-zeroed the torque. We had told him to re-zero after a few minutes and again before he did the "official" test runs (and again after the test runs to be able to see how much drift there had been, if any). Because he had a lot to do, he forgot the first re-zeroing but did the second and third. However when we were laying out the protocol we had asked him to include all the data, warts and all, which is what he did -- even the part before the first re-zeroing.

meanredspider

RChung wrote:

meanredspider wrote:

I don't know that his example is that extreme - my day in the Alps saw a much bigger variation - my commute in Scotland probably did.

I described the condition as extreme and unusual because he took the bike from a warm apartment into the freezing winter air so the difference was large and sudden. Even in the Alps the temperature tends to change more slowly than that, and even in the Alps you occasionally get a chance to coast which would have re-zeroed the torque. We had told him to re-zero after a few minutes and again before he did the "official" test runs (and again after the test runs to be able to see how much drift there had been, if any). Because he had a lot to do, he forgot the first re-zeroing but did the second and third. However when we were laying out the protocol we had asked him to include all the data, warts and all, which is what he did -- even the part before the first re-zeroing.

Fair enough - I've been back and re-read the piece and the part about the warm apartment isn't at all clear so I suppose it's not surprising that people raise it often. Even the Garmin temperature plot only starts at 56F/13C and I know from my unit that it starts all the rides at the temperature of the house. It's not unusual to get a significant drop in temperature in a big climb and not many have somewhere to coast least of all in a competitive event. It's surprising that Quarq hadn't build temperature compensation into their device. Anyhow, now I know.

colinsmith123

Wow, reading all of that was like listening to an A Level Physics Debating Society.

I chose the Quarq, over Stages for some of the reasons mentioned in this thread (I think). Morevover, because my friendly LBS could give me a greater discount on the Quarq than the Stages.

I do glaze over a little with the technical side of things. I wish I had the patience / intelligence / time to understand it all better.

What did spark my interest was the conversation regarding temperature compensation. What I can't get my head around is why the technology sitting after the strain gauges cannot automatically compensate for temperature changes. As a point of reference, it does the term PM in cycling parlance refers to the device doing the sensing, ie. Stages; Quarq; P2M etc. and yet it is the head unit that is displaying and recording the measurement, this is the meter. So why is it the likes of Garmin head units do not automatically compensate for temperature changes within its algorithms?

My bike lives in the shed, the Garmin in the house. Doing a 'calibration' or as I think it really is, a zero offset when the temperature difference between inside and out can be 10-25C difference seems pointless. I try to remember to backpedal around 15-20 mins in to the ride and again a couple of times on a , but don't always remember. Again, a simple alarm on the Garmin cannot be that hard to write in to the software. Don't start me on the firmware / software argument I have to deal with that at work.

Thanks for an interesting post.

slowbike

Grill wrote:

Sure. I post all my data for the same reason. If I did pay for a couple hours in the tunnel I'm not sure I'd be terribly keen to just tell people my findings though.

Not sure why you'd want to publicly post any non-road data anyway - I usually set my turbo sessions (when I can be arsed to do them) to private ...

RChung

meanredspider wrote:

It's not unusual to get a significant drop in temperature in a big climb and not many have somewhere to coast least of all in a competitive event. It's surprising that Quarq hadn't build temperature compensation into their device. Anyhow, now I know.

Yeah, I mentioned race situations where you may not have a chance to trigger automatic re-zeroing. But I also mentioned that in those instances when you get to the top you often coast enough to trigger it and if you want to (that is, if you're desperate enough to) you can often correct the data for a torque error post hoc.

colinsmith123 wrote:

So why is it the likes of Garmin head units do not automatically compensate for temperature changes within its algorithms?

At the moment, there are power meters that live in the rear hub, the crank spider, the bottom bracket, the left crank, and the pedals. Each of them is built a little differently, with strain gages embedded in different places. As long as the power meter transmits using the ANT+ protocol, any ANT+ head unit can receive what's being sent. So the Garmin head units are agnostic about which PM they're connected to. FWIW, I think the newest generation of Quarqs has better temperature compensation than when that test was done but I haven't directly done another test to make sure. I suppose I could shove one in the freezer or under a heat lamp for a few minutes then pull it out and go for a ride but from a practical point of view I just tell people that if they experience large temperature changes under conditions when they won't trigger the auto-zero they should manually zero.

meanredspider

colinsmith123 wrote:

Wow, reading all of that was like listening to an A Level Physics Debating Society.

A Level? I feel flattered

colinsmith123 wrote:

Thanks for an interesting post.

Ah, that's better

frisbee

colinsmith123 wrote:

What did spark my interest was the conversation regarding temperature compensation. What I can't get my head around is why the technology sitting after the strain gauges cannot automatically compensate for temperature changes. As a point of reference, it does the term PM in cycling parlance refers to the device doing the sensing, ie. Stages; Quarq; P2M etc. and yet it is the head unit that is displaying and recording the measurement, this is the meter. So why is it the likes of Garmin head units do not automatically compensate for temperature changes within its algorithms?

The Garmin doesn't know how the strain readings change with temperature. It'll depend on the strain gauge, the wiring to the gauge and material the gauges are mounted to.

The best way of compensating for temperature is to design compensation into the strain measuring circuit itself, on the force measuring systems I worked on we used dummy gauges which cancelled out the temperature effects on the gauges that measured force. A pretty common approach, I would be surprised if some of the power meter makers don't do the same.