Power Test using Polar S-720 Road Test

I have seen many posts regarding the inaccuracy of Polar's Power Option. I have two of these units on two different bikes. I have been able to reproduce the erroneous readings that people have reported to occur on a trainer with cog to cog variation of up to 25-30%.
I therefore decided to take my main road bike on two different road tests to examine the accuracy and reproducibility on a 1.2 mile stretch of road with a steady incline of 0.4%. I repeated the test on 8 of the nine rear cogs, i.e. the 12...23. I am using an FSA compact carbon crank with a 50/34 chainring in combination with a 11..23 cog set. I was monitoring boith speed and cadence throughout the trial. Here are the results of the readings from the large chain ring:
Gear... watt @ 19mph...% from average
50_23... 226... -3.48
50_21... 230... -1.77
50_19... 237... 1.22
50_17... 234... 0.04
50_15... 232... -0.92
50_14... 235... 0.27
50_13... 230... -1.77
50_12... 241... 2.93
Average Power...234...
The errors are pretty tight and resemble a combination of riding variations and reading errors. Regardless, the error was within 3.5%. In fact, ignoring the 50_23 combination which shouldn't be used anyway and the 50_12, the error is less than 2%. The same bike gave me a 25-30% variability on the trainer.
So I am quite satisfied with the performance and reproducibility from the road test and can second the opinion posted by others before that this device works well on the road and poorly on a trainer. I should also note that I calculated the power that would be expected given my weight, bike weight, incline and so forth at 237 Watt for the above trial. My average came in at just 1% below that. Again, that would suggest that the unit also reports absolute values correctly, unlike one of the websites that suggested that the unit over-reports by 20%. I have repeated this test twice with nearly identical results. The second test was done at 20mph with a slight steady headwind requiring a ~285 Watt power output. The differences between the cogs was within 3.5%.
Hope this is useful for someone.
Harry

 

Wow, those look like promising results.

Did you manage to conduct the same test for the 34 tooth cog?
I would be out on my bike right now doing the same test if it weren't for the snow.

I was also curious if when tested on a trainer, the power meter ouput had a wave like patter to it? I have conduct 5 tests on both rollers and trainers, and they all produced similar results. There is a distinct patter as the cassette was traversed by the chain.

Brandon

 

Brandon:
I will be doing a test on the 34 this weekend weather permitting.
As to the pattern on the trainer, it indeed looks like a perfect sinewave running over the cassette with the two outer cogs (11 and 23) as well as the middle (15) reading correctly, the 17,19 and 21 reading too high, the 14, 13 and 12 reading too low. Hence the wavelenght is exactely 9 cogs or the full width of the cassette. As a scientist I find this very intriguing. In fact when plotted it looks like a perfect sinewave as if some vibration is traveling through the axel. I can't make sense of it yet. The pattern is the same on the 34 cog.
Interestingly, if you use Spinnervale tapes, you will be using almost only the middle and the two outer cogs which all read correctly.

 

Hey Harry,

Just wanted to check in and see how the test went with the 34 tooth. Also, along the same lines, I was wondering at what temperature your unit ceases to function. Mine stops working at around 0 to 5 degrees celcius. Is this strange, or have you observed similar results? On my winter rides the power unit and all output coming from it will stop. I am unsure if this is because of the circuitry (doubtfull since transistor and diodes should function to much lower temperatures), loss of battery Voltage(maybe, but I tested the battery output while under load at low temperatures and it seemed to be consistently 3 V, but haven't tested current output), or perhaps it is my wires which may be broken in the lines and shrink due to thermal contaction.


If you have any thoughts on the matter please let me know.

Brandon

 

I have not had the chance to do the 34 test. It has been raining all along. I am riding in temperatures around 0 centigrade and have not had any issues at all. I am surprsied that the unit would seize to function.
Harry

 

So the reason for my power meter stopping around the freezing mark, was because of thermal contraction. The metal cap which covers the battery holder, and completes the circuit for the entire unit, has some tolerance problems. THere is a little metal tab in the middle layer which the metal cap hooks onto. It seems that as the metal cap shrinks in the cold weather, it no longer makes contact witht the metal tab thus loosing power to the entire unit. The strange flashing of LED's (long fading bursts instead of flashes) is because the cap is disconnecting then reconnecting. My solution was to not latch the cap in, and just use a rubber band and electrical tape to hold it in place.