Powertap Torque Test - Rough Initial Test

I have wondered all season why my data has looked a bit off. Since I am fairly new to using a power meter I slowly learning the finer details. Unfortunately if the following test is correct my power meter is about 22% in error and so is about 4 months of my collected data in WKO. One of the biggest lessons I have learned is to "zero" the cpu before each ride and then to do a periodic torque test similar to the one below to see if the power meter is still in calibration. Mind you that all scales or means of measurements need to be checked on occasion to verify the accuracy. Otherwise it is a rather useless tool of measurement.

The following was just a rough experiment as a preliminary test. I will follow up with a more refined test at my gym using a little better scale and a measured olympic 45 pound plate.

I filled a bucket with tile grout with a weight of 53.0 pounds

Spin the wheel to wake up the CPU

Using the mode button set to the watts line on top

Using the select button hold it until you see the "watts" text flash

At "0" hit the select button to zero the cpu

With the bike stand lifted up and stick to hold the wheel from spinning

With a level I checked the crank arm for level position
Press the select button until the "watts" begins to flash

Hang the measured weight on the pedals / crank arm

Torque is at 130 on the 34 ring and 17 cog

Switch to another gear and repeat the process

Torque is at 179 on the 34 ring and 22 cog

Measured Weight: 53.0 lbs

Crank Length: 170

Crank Ring: 34

Measured torque  =  (weight in lbs)  ×  (crank length in mm)  ×  (1 in/25.4 mm)  ×  (cog teeth/chainring teeth).

Measured Torque for 17 Cog = 177.36

CPU Displayed Torque for 17 Cog = 130

26% diffference

Measured Torque for 22 Cog = 229.53

CPU Displayed Torque for 22 Cog = 179

22% difference

_______________________________
UPDATE June 11, 2010:

After sending in my wheel to Saris last Friday I received it exactly a week later so that is a pretty good turnaround for service. Especially since most of that time the wheel was in transit.

I tried the same rough test noted above and here are the results.
Of course this is a very rough test, but something I just wanted to do as a quick check.

Measured Torque for 22 Cog = 229.53

CPU Displayed Torque for 22 Cog = 224

2% difference

_______________________________
UPDATE April 28, 2011:
Spreadsheet Tool