Polar Power sensor

886014 said:

[font=Times New Roman]I[font=Verdana] have just (re)installed a Polar Power sensor and have been checking its performance on the trainer. At a constant speed I get a large decrease in indicated power as I go through the gears ie as cadence drops so does the indicated power. Ignoring grossly crossed chain combinations, at a constant 23kph the power varies between 115 and 175W depending on the gear; clearly totally useless.[/font][/font]

I’ve spent hours fiddling with the chain sensor (which interestingly didn’t make much difference, though raising it increased the reliability of the data) and it now sits just a few mm below the chain. All other sensors work well and my Ultegra 10 speed CN-6600 chain is entered as 280g and 1473mm on the Trek chain stay length of 417mm. Before I throw the whole thing in the bin has anyone got any further suggestions as to where I should look to get this reading something like correct? I don't mind if it's not accurate, but it should be consistent otherwise it's a waste of time.

You might want to do a search of comparisons of powermeters on the internet or read the power FAQs on the midweek club site or do a search of the wattage group on topica. This is a known issue for most people using the Polar powermeter on trainers, even when they are properly installed.

Thanks for that. I managed to cure the problem by replacing the chain would you believe. No other changes made at the time so convinced that was it. Got the idea from this article.biketechreview.compm review 2.htmOpen ↗

Now seems to be quite reliable.

There is also an issue with Polar Power on a trainer that you have to consider.

Polar bases its power calculations based on chain vibration. When you're on a trainer there will be a lot of vibrations from the bike-trainer setup that will cause interference with the power sensor.

I have read elsewhere that polar power is pretty much useless on a trainer because of this. It's better on the road.

_________
CycleFast

I haven't heard that one, and it certainly doesn't correlate with my own experience. On the road there are far more vibrations (think "dead" roads) than are ever going to be experienced on a trainer. The only remaining anomoly is that 2 of my cogs give slightly higher than normal power readings. As it happens these same cogs don't run as smoothly as the rest. I suspect those vibrations are what are causing the power errors.

I've trawled through a lot of info on Polar on the net and have concluded that there is a lot of BS out there. Unfortunately most of it seems to be in the variety of "I heard a friend of a friend had one of those and he said ...", you know the deal.

The bottom line is that the thing wasn't at all difficult to install, just tedious if you're anal like me and tape all the wires to the frame with clear tape rather than zip ties and try to make it all very neat and tidy. Getting the sensor as high as possible isn't too important but does seem to give more consistent readings. At some stage I'll machine up a plastic mount for it, until then the rubber they supply seems to be fine. Whether the readings it provides are accurate is debatable, and frankly I'm not too concerned. What I'm more interested in is whether it's consistent, and after solving the chain problem it seems to be, except at very low cadences.

Hope that helps.

886014 said:

I haven't heard that one, and it certainly doesn't correlate with my own experience. On the road there are far more vibrations (think "dead" roads) than are ever going to be experienced on a trainer.

I don't know from personal experience, but I've heard several people report the readings when using a trainer. I think you will have a totally different kind of vibration on the road and on the trainer. Maybe the vibration frequency is different (lower on the road than on the trainer).

It's mentioned several times on this forum as well. I think Ric has mentioned a few times as well.

_________
cyclefast

CycleFast said:

I don't know from personal experience, but I've heard several people report the readings when using a trainer. I think you will have a totally different kind of vibration on the road and on the trainer. Maybe the vibration frequency is different (lower on the road than on the trainer).

It's mentioned several times on this forum as well. I think Ric has mentioned a few times as well.

_________
cyclefast

I have Polar power sensor and have had dodgy readings on the trainer but most of the time it does work fine. On the road it is very consistant and I have checked the readings against calculations for hill climbs and it is pretty well spot on.

My own theory about the trainer (on carpet) is that it is too smooth with minimal vibration and when I'm pedalling smoothly the chain doesn't vibrate enough sometimes and the results wander.

You do also get slight differences in different gear combinations as well and I'm pretty well convinced that this is due to the differing free to vibrate chain lengths. I read that Polar has some function built in to adjust for the different gears (and hence chain lengths) in their calculations but as the system doesn't allow for inputs of this kind and many different gear setups exist on bikes it can't possibly be right for all of them. My setup is pretty standard with 53/39, 12-25 and it seems to work fairly consistantly but Polar may not be so good with tripple chain rings or mtb set ups.

pod said:

I read that Polar has some function built in to adjust for the different gears (and hence chain lengths) in their calculations but as the system doesn't allow for inputs of this kind and many different gear setups exist on bikes it can't possibly be right for all of them.

As is described in the patent covering the Polar power sensor, "if the chain linear speed, and the rotational speed of the two sprockets are known, sprocket sizes can be determined without prior information about what sprockets are available. The rotational speeds of the sprockets are commonly measured on bicycles, with the speed of the front sprocket used to calculate cadence, and the speed of the rear sprocket (which is equivalent to the speed of the bicycle wheel) used to calculate bicycle speed."

This is the technical reason why the Polar power sensor includes integrated wheel speed and cadence sensors, in addition to the marketability of such an integrated product.

The patent is an interesting read for those that are so inclined:

"Method and apparatus for measuring power output and for measuring tension and vibrational frequency of a elongate flexible member"

Berend

squidwranglr said:

As is described in the patent covering the Polar power sensor, "if the chain linear speed, and the rotational speed of the two sprockets are known, sprocket sizes can be determined without prior information about what sprockets are available. The rotational speeds of the sprockets are commonly measured on bicycles, with the speed of the front sprocket used to calculate cadence, and the speed of the rear sprocket (which is equivalent to the speed of the bicycle wheel) used to calculate bicycle speed."

This is the technical reason why the Polar power sensor includes integrated wheel speed and cadence sensors, in addition to the marketability of such an integrated product.

The patent is an interesting read for those that are so inclined:

Berend

Interesting reading. I didn't wade through it all and yes the relative sproket sizes can be calculated using chain speed, and the relative speeds of each, however, there doesn't seem to be any mention of the differing chain angles (which of course will affect length) and the fact that you can get the same gear ratio with different combinations on different chainrings and quite different chain angles.

There was a study which I can't locate at the moment showing the accuracy of the Polar system and it showed a consistant pattern in the variations on each chainring ( for each rear cog) which to my mind fairly closely matched the change in chain length. They also say in the Patent application that the difference is minimal and could be ignored (or approximated by using slightly less than the nominal length) and I suspect that that is what they have done, despite pointing out how it could be calculated.

Whatever, for my purposes it is accurate enough and seems fairly consistant and I'm more than satisfied as I have a reasonable indication of power at a price I'm willing to pay. The other systems also have accuracy issues and are much more expensive.

886014 said:

I haven't heard that one, and it certainly doesn't correlate with my own experience. On the road there are far more vibrations (think "dead" roads) than are ever going to be experienced on a trainer. .

Not that kind of vibration. The road has a lot of wide band vibration, i.e. lots of different frequencies. This actually helps the polar lock on on the chain's frequency. Trainers have a dominant single frequency vibration that might be throwing off the sensing algorithm.

886014 said:

I've trawled through a lot of info on Polar on the net and have concluded that there is a lot of BS out there. Unfortunately most of it seems to be in the variety of "I heard a friend of a friend had one of those and he said ...", you know the deal.
.

I ran a polar for two years but could never get a level of precision I was happy with. During that time I ran several trials and tests which usually came up short. I never tried an A/B test with new chains but....how will you know when your new chain isn't good enough?

I am an electrical engineer and once upon a time I worked on a product which used vibration to measure timing belt tension and the principles of the polar power sensing system are good, they've just got to work out some details to get the thing more accurate and precise. Till then I'll stick with my PT.