Some months after the official launch of P1 pedals, there are lots of reviews talking about the main advantage of this powermeter: it is the only real PLUG & PLAY powermeter in the market. There is no need of installation torque measurement, drive train adjustement, or special tools need.
You need a Hex 8 mm key tool and less than one minute to install them, what makes this system unbeatable in case of owning two (or more) bikes and several wheels.
Some of these reviews talk very briefly about the accuracy of the system giving some (few) data comparing several powermeters.
As user of Powertap Hub Powermeter since 2008, I’d like to know exactly which this difference was as I have a vast database obtained during years of races, training, intervals and analysis that for me is much more valueable than any FTP test result. So I asked for them and for the first time in my life, a Cycling industry company borrow me a product to test it for free.
In July, Tamalpais, borrow me one of the very first pairs of pedals in Spain. It was only for 48 hours, but was enough to get the first data and impressions.
The first test consisted on some 4 minutes intervals in open velodrome with my time trial bike, and these were the results.
16w and 6% difference. That was a little surprising for me. I initially thought that there would be a difference of 5-6w due to friction in the drivetrain.
As seen in the graph, the difference is more or less regular, and not linear with the applied power.
So, I contacted Powertap USA, checked with the zero offset values, and explained the results.
They told me it was more or less the expected results and I got the pedals again in November for some weeks to continue testing them.
First of all, I decided to repeat the test type in the velodrome to check if may be there was an error in July, as at that time, there was no Garmin firmware available to set the crack length, although the default crank length (172,5 mm) was the one of my time trial bike crank. Fortunately, the firmware for 920xt was released the same week I receieved the pedals and I set this parameter in a 910xt for G3 and a 920xt for P1.
It is very important to remark that this parameter can only be set if the powermeter is linked to the Garmin.
Crank length set in Garmin 920xt procedure:
Step 1: link the powermeter
Step 2: connection double check
Step 3: Set crank length
Step 4: Zero Offset
Crank length set in Garmin 910xt
Step 1: link the powermeter
Step 3: Set crank length
Step 3: Zero offset
Step 5: Laps & laps…
This time I used the P1 pedals (and also the G3 hub) to compare the aerodynamic of several HED wheels, as I explained some weeks ago HED front wheels aero test:
The difference was the same than in July. 6% and non linear relation with applied power.
I took advantage of the opportunity of having some top of the range rear wheels borrowed by some friends, and I tested these wheels with exactly the same protocol used the day before (Top of the range front wheels aero test)
Once again. I got the same results. 6% difference, so I decided to go to another bike, another rear wheel, another crank length and another kind of road. I rode to Arlaban, a 8 km climb, with a constant 4% slope. These are the results.
Surprisingly, there was a nice variety in the results, that was more noticeable in the 20 laps I rode in the velodrome. 12%, much higher than the laps with the time trial bike….
In this case, it seems that there is a linearity with applied power, but not with the uphill interval. I was a little bit confused.. But I continued with the tests.
I got back to the time trial bike, and I did a 1h30 workout, flat and constant pace, without intervals. The result was a 5,3% difference, very near to the 6% of the first trials in the velodrome.
It can be noticed that, apparently, in the first minutes the P1 is clearer higher than in the last part of the workout.
Maybe something related to temperature management?
First part (20 min): 233/217=7,3%.
Second part (80 min) 238/228 : (4,3%)
But all my thoughts and theories changed when I did the comparison in the turbo trainer. And these where the results:
Virtually the same!! If the difference is lower than the accuracy of the system (1,5%), it can be said that the result is equal
These results changed my mind, and I started to think that the power loss was not only in the drive train but it should be somewhere else: Wheel deflection? Spoke deformation? Frame deformation?
It is impossible to separate this effects in the road (a complete strain gauges system would be required with dedicated actuators to introduce controlled load), but I least I wanted to test if what I thought that would introduce more flexural deformation in the bike (pedaling out-of the saddle), would affect the results.
So in my following workout, I did some (short) out- of the saddle efforts at different power levels.
Well, not very and probably less than expected but it seems that there is a difference.
Other test that was very interesting for me was the difference in one of my favourite workout: 4×10 min at 85-90% FTP. Although it was a difficult day due to intense fog and some problems with multiple sensors detection (I was using Garmin 510 instead of Joule) I got the same result as used before with the time trial bike: 5-6%.
And again I repeated the 4x10w, but this time I went to my favorite road, the Vitoria triathlon circuit, a perfect road to do this kind of test.
This time, the difference was a little higher: 7,7%. This means that instead of the previous 6%, that would be 250*1,06= 265. I got 267, 7w… well, 2,7 watts,… I think I could live with that J These are the graph comparison of each interval.
My last test consisted on applying more tension in my both rear wheels (in order to reduce deflection and power loss) and to compare.
So, the final conclusion is that I have a constant 6% difference between my G3 hub and the P1 pedals in my time trial bike. More or less, is similar to for instance, the values Starykowicz has seen with his pedals (http://www.slowtwitch.com/Interview/Starykowicz_in_full_5526.htm)
In my road bike I have more difference, probably due to some power loss in the frame or in the rear wheel, that are more noticeable in the velodrome where some lateral forces could affect the whole structure more than I expected.
In any case, the measure is consistent, maybe more responsive than the hub, and although I would have to update my power files database, it is highly recommendable in case of having two bikes ore several wheels.