The Powermeter Thread

[swuzzlebubble]

and it's not even a power meter just a power estimator. The good news is that the competition is bound to continue to lower prices and drive the increase in feature development.

Hugh

iBike Newton +

The Newton refutes the thesis that power should be measured directly and cannot be determined exactly by indirect measurement. Thanks to its wind and acceleration sensors, incredibly accurate power values can be displayed.

How power is calculated inside the Newton remains a well-kept secret, but it has been improved considerably by the manufacturer since 2004.

The Newton’s results are surprising. The Newton and combined speed/cadence sensor can be mounted quickly. After a ten minute out-and-back test ride the device is ready and shows, amazingly,
almost exactly the same values as the other power meters tested.

RENNRAD: POWER TO GO

 

[Alex Simmons/RST]

RENNRAD: POWER TO GO

That's some pretty lame testing.

 

[swuzzlebubble]

That's some pretty lame testing.

Par for course for a bike mag tech article

 

[swuzzlebubble]

Yes, I have a G3 and manually dialed the Newton+ in over a few rides. If you can borrow a DFPM and do that it helps considerably I have to admit. It takes a little knowledge to be able to look at sections of a ride to see what needs taking and giving to get the results you want. Issac has a DFPM analyzer but I got better results manually studying the data and tweaking. Like building a wheel it needs a human's touch. But after those tweakings I haven't touched it. I only check it occasionally because I'm coming from a Gen3 iAero which I had to tilt and check all the time. The Newton has proved much more reliable as I haven't touched it other than the wind scaling.

To be honest, the Newton with the newer OS is way easier to deal with than a gen3, it's just better everywhere. I never had the Gen3 Iaero give NP scores like the compared DFPM even though the avg. watts were some-what close. I like to use NP as a true guide when riding especially in groups. This really lets you see how it's dealing with all the minor and major accelerations, the newton is really accurate here. The Gen3, not so much in this area, even though avg. watts can be really close(50% of the time). But also remember the DFPM's takes avg. power per crank revolutions while I believe the Newton is picking up an immediate acceleration. So I would think this will cause a slight change or raise the NP while sprinting out of corners or accelerating from a coasting situations(crits and descents). I'm sure someone from Velocomp could let us know for sure, but DFPM's aren't perfect either and it's quite possible this could be an area where the Newton is more precise.

I looked through the file and I would say the winding 7.75 mile descent, avg. 5.2% that lasted 12'50" has the biggest discrepancy at 9 avg. watts.(NP13) for the newton, it read high. Just to let you know how we descended that pass, If I were to download it to Strava I would be about 6 seconds off from the KOM. Over 550 people have descended Glendora Mountain Rd. So we were pedaling hard and sprinting out of the corners. If we soft pedaled down and coasted, the numbers would be insignificant. But, yes, if there were an area where it could use work.....I'm just not sure if it will mean anything.
I don't do a lot of intervals going down mountains so it wont really affect training. But all the combined descents(major and minor) though-out the day is probably what adds the most to the overall wattage difference between the two(newton/DFPM) over the entire ride. I'm not going to try and fix this due to the fact, it's spot on with ascents, rollers, groups, and windy conditions. I'll give that one to the gods and just deduct a couple TSS points at the end of the day when doing group climbing rides, or probably not.

http://ibikeforum.com/viewtopic.php?f=4&t=2391&st=0&sk=t&sd=a#p14807

 

[Alex Simmons/RST]

Par for course for a bike mag tech article

Not sure what those images represent, but again, ride averages are one of the least robust ways of comparing and assessing power meter's performance.

 

[swuzzlebubble]

Not sure what those images represent, but again, ride averages are one of the least robust ways of comparing and assessing power meter's performance.

In the top image, compare the green power plots.

In each case the IF, NP & VI figures are worth comparing as they are effected by variability.

 

[CoachFergie]

In the top image, compare the green power plots.

In each case the IF, NP & VI figures are worth comparing as they are effected by variability.

They aren't.

 

[Alex Simmons/RST]

In the top image, compare the green power plots.

In each case the IF, NP & VI figures are worth comparing as they are effected by variability.

There are many way to arrive at the same /similar values. Again this is the least robust method of assessment.

But then the world seems to care less and less about accuracy of power data now days.

Here's an example test. Do some peak crank force-velocity testing and compare results. e.g. a series of full on accelerations from still or near still motion. plot the 10-second data points on a crank force v crank velocity plot.

 

[hiero2]

Not sure what those images represent, but again, ride averages are one of the least robust ways of comparing and assessing power meter's performance.

Agree - some labels on the images to tell us what we are seeing reported would be, ahem, useful.

 

[hiero2]

That's some pretty lame testing.

Personally, I wouldn't be so quick to throw the baby out with the bathwater. At least it IS a product performance test. And they did, at least, follow scientific methodology by repeating the tests and maintaining the large number of possible variables as constant as was practical. Maybe somebody could step up and run a better series of comparisons - or test under controlled conditions where one system should theoretically be at a disadvantage - like in a wind tunnel.

 

[CoachFergie]

Personally, I wouldn't be so quick to throw the baby out with the bathwater. At least it IS a product performance test. And they did, at least, follow scientific methodology by repeating the tests and maintaining the large number of possible variables as constant as was practical. Maybe somebody could step up and run a better series of comparisons - or test under controlled conditions where one system should theoretically be at a disadvantage - like in a wind tunnel.

Check many of the studies at the start of the thread that show how two or more different power meters were compared.

I have done comparisons of Power2Max, Quarg and SRM against a Powertap and also compared data from an Infocrank and Powertap for other people.

 

[hiero2]

and it's not even a power meter just a power estimator. The good news is that the competition is bound to continue to lower prices and drive the increase in feature development.

Hugh

The only developments I see are:
- more one sided only power meters
- a push to move more of the manufacturing process out to the consumer
both of which are techniques to provide lower cost offerings
- increasing reliance on third party recording devices

What I don't see are efforts to improve data quality, indeed I'd say that has declined on average over the last two years, matched by an increase in rhetoric over why that doesn't matter.

Aside from new power meter options becoming available, what useful new features have been introduced over the last two years? Please don't say power balance.

Alex,

Good point about not so useful features. Perhaps my use of the term "bound to" was a bit optimistic on my part. You have to agree that MSRP for all established brands except SRM have dropped significantly over the past year and I even see SRMS more heavily discounted than in the past.

Hugh

Competition and a strong market will drive price down. Glad you rethought the bit about better/more features on offer due to competition. Whether that happens will depend muchly on the demands of the buying side of the market. I.e. how fussy will the majority of consumers be? My crystal ball would indicate that we will continue to see a decrease in data quality in exchange for price leadership. Until there is a form-factor breakthrough - which will happen at some point.

 

[hiero2]

Check many of the studies at the start of the thread that show how two or more different power meters were compared.

I have done comparisons of Power2Max, Quarg and SRM against a Powertap and also compared data from an Infocrank and Powertap for other people.

Ok - since I made the comment - and you gave a reply - I went and looked - as you said - at the start of the thread. Up through page 4, if I'm not mistaken, there are some studies. They date back to 2005 and 6, which is a technological eon. I have no doubt as to the validity of the studies, but browsing them, it seems to me they VALIDATE the Rennrad product review. Also, I didn't see the Rennrad product lineup represented in those studies. Have you compared, or do you know of another study comparing SRM with iBike Newton, Garmin Vector, and Polar products?

Coach, I stand by what I said. This (the Rennrad paper) is a product review. I don't see how it deserves to be called lame. It seems quite valid afaik. It may not be as rigorous in technique as the studies, but, frankly, until we see something more rigorous that includes all the products in the review, then it remains the best we've got at the moment.

 

[FrankDay]

But then the world seems to care less and less about accuracy of power data now days.

Perhaps the world has figured out that accuracy isn't particularly important if one is interested in cycling improvement. Ignoring the fact that Knowing power per se has never been demonstrated to influence outcome if one did want to use the tool it seems reliability is more important than accuracy. Can I learn to correlate effort to the number I am seeing? Whether the number is 2 or 20 watts off is of little importance if it is always the same amount off.

 

[Alex Simmons/RST]

Perhaps the world has figured out that accuracy isn't particularly important if one is interested in cycling improvement. Ignoring the fact that Knowing power per se has never been demonstrated to influence outcome if one did want to use the tool it seems reliability is more important than accuracy. Can I learn to correlate effort to the number I am seeing? Whether the number is 2 or 20 watts off is of little importance if it is always the same amount off.

Stop trolling Frank.

For about the millionth time - the level of accuracy and precision required depends on what you intend to use the data for. The fact you persist in ignoring this and keep reciting a low-fi use of data as being the only use for performance improvement is, well, tiresome.

 

[Alex Simmons/RST]

Personally, I wouldn't be so quick to throw the baby out with the bathwater. At least it IS a product performance test. And they did, at least, follow scientific methodology by repeating the tests and maintaining the large number of possible variables as constant as was practical. Maybe somebody could step up and run a better series of comparisons - or test under controlled conditions where one system should theoretically be at a disadvantage - like in a wind tunnel.

Sure, and almost anyone can do similar sort of tests. But these sorts of tests are only testing low-fi, and don't address the performance of meters under various conditions of use.

e.g. simple tests on trainers are about as low fi as it gets. It's what I call the very minimum performance hurdle a power meter needs to jump over. But there are many other hurdles that are rarely attempted.

Like I said earlier, for example, do some peak force-velocity testing and report back with the results. Then you'll realise one application where the data isn't as good as some might have you believe.

 

[hiero2]

Sure, and almost anyone can do similar sort of tests. But these sorts of tests are only testing low-fi, and don't address the performance of meters under various conditions of use.

e.g. simple tests on trainers are about as low fi as it gets. It's what I call the very minimum performance hurdle a power meter needs to jump over. But there are many other hurdles that are rarely attempted.

Like I said earlier, for example, do some peak force-velocity testing and report back with the results. Then you'll realise one application where the data isn't as good as some might have you believe.

I think "low-fi" is a much better and more accurate descriptor than "lame". And, I think you are probably right - but the product review was not attempting to do a high fidelity test, were they? If you were to present your suggestions for test improvements to the testers, and if you convinced them that they were cost-effective to do, I imagine they might incorporate your suggestions. Which would improve their testing, right?

But regardless of that - if low-fi testing is all the testing you HAVE, why would you diss it?

 

[CoachFergie]

Ok - since I made the comment - and you gave a reply - I went and looked - as you said - at the start of the thread. Up through page 4, if I'm not mistaken, there are some studies. They date back to 2005 and 6, which is a technological eon. I have no doubt as to the validity of the studies, but browsing them, it seems to me they VALIDATE the Rennrad product review. Also, I didn't see the Rennrad product lineup represented in those studies. Have you compared, or do you know of another study comparing SRM with iBike Newton, Garmin Vector, and Polar products?

Coach, I stand by what I said. This (the Rennrad paper) is a product review. I don't see how it deserves to be called lame. It seems quite valid afaik. It may not be as rigorous in technique as the studies, but, frankly, until we see something more rigorous that includes all the products in the review, then it remains the best we've got at the moment.

I haven't even read the review so it's not me calling it lame. I questioned the use of average powers from a ride and the suggestion that normalised power, intensity factor etc were measures of the variability of power.

 

[Alex Simmons/RST]

I think "low-fi" is a much better and more accurate descriptor than "lame". And, I think you are probably right - but the product review was not attempting to do a high fidelity test, were they? If you were to present your suggestions for test improvements to the testers, and if you convinced them that they were cost-effective to do, I imagine they might incorporate your suggestions. Which would improve their testing, right?

But regardless of that - if low-fi testing is all the testing you HAVE, why would you diss it?

Because passing a low-fi test is pretty easy and not particularly instructive about a power meter's performance. The main issue is people passing off such tests as being more instructive than they really are. That's the bit that's lame.

A peak force-velocity test is free and takes less than 10 seconds. And it's very instructive about how well a power meter operates.

That's not the only test of course, but is simple, quick and cheap to do.

 

[CoachFergie]

The validity and reliability of the data is very important. What is to say if a PM is 20 watts at threshold that it is 20 watts out at other levels. What if you have several power meters and are looking to compare across the data?

The validity and reliability of data is important for tracking training. Studies show that functional overreaching is better than non-functional overreaching and overtraining so using the best data metrics to spot when performance is decreasing is essential.

For constant efforts determination of race pace is crucial. Whether a 200m TT or a 180km Ironman leg pacing is vital. Start too hard and it all ends badly.

Where a power meter really earns it keep is by measuring the variability of performance. Whether it is due to the course, the competition, weather or tactics used if one knows what to expect a coach can ensure a rider is properly prepared.

As the rider improves the rate of learning decreases so the precision of measurement is important to ensure gains made are real. Unless they use a Powercrank when the learning curve slopes the other way. When an athlete is fighting for marginal gains the validity and reliability of data is important.

Do they need this level of precision at the start? I don't demand juniors get a power meter as they have more important things to learn but when they do start measuring power it should be done right. So no left side PMs in Team Ferg.

 

[swuzzlebubble]

Well guys my point is that iBike Newton measures power via accelerometer, inclinometer, other sensors and algorithms in the same sense that DFPM measure it via their sensors and apply algorithms to calculate a power number. The only issue is accuracy and I guess user-influence such as calibrations.

Systems that "estimate" power are those eg that convert speed to watts using a 'power curve' for a wind trainer etc, or estimate from eg heart rate.

In regards to accuracy I would be very surprised if many of these one-sided or other similarly priced units would be more accurate than the Newton. It therefore deserves to at least be considered in comparisons considering other advantages such as simplicity of switching between bikes and additional features.

All systems I have seen seem to claim accuracy in the 1-2% range (including Newton, afaik) and yet the left-right balance difference for a rider can well exceed that to start with.

Re the screen shots from earlier, this the the power plot of G3

This is the same ride via Newton

 

[Alex Simmons/RST]

All systems I have seen seem to claim accuracy in the 1-2% range (including Newton, afaik) and yet the left-right balance difference for a rider can well exceed that to start with.

No argument there.

 

[Alex Simmons/RST]

Re the screen shots from earlier, this the the power plot of G3

This is the same ride via Newton

Hard to tell but the peak powers looks quite different to me. A PT probably isn't ideal for that type of comparison though due to precession effects.

But what happens with Newton data if road surface changes, or you change position on the bike, or change clothing, or swap wheels over? Simple normal everyday things.

This is what one tester found:

What if you alter your ride position? Well, I did another ride yesterday with my hands on the handlebar drops despite having the Newton PowerStroke set up for riding on the hoods. Over that ride, not surprisingly, there was a big discrepancy between the figures – nearly 11%. In other words, you have to do what you say you're going to do in order to get accurate results.

 

[swuzzlebubble]

^ Answer in your quoted bit.

For my usage I have one profile set for commute bike with backpack etc
The other for roadie with standard kit
Two others for my son's road and track bikes.
Does the job for us.

Hey, I'd be interested in a DFPM if these was one in my price range (new)
I've heard of some crowd here in Melbourne developing something that sounds interesting. I'll see if I can find out more.

 

[Alex Simmons/RST]

^ Answer in your quoted bit.

So if you take on/off your rain/wind jacket, or decide to change hand position from tops to hoods to drops, then the data won't be anywhere near accurate. OK.