PowerTap innacuracy?!?

[JohnMeyers]

I have never had reason to doubt my PowerTap, but I think today I feel like it may be misleading me. In recent weeks I've been doing 3x20's and 3x30's.

3x20 splits have been the following:
316
336
341

355
329
333

3x30's:
318
342
321

321
326
325

Today I did a 3x20 workout and got the following:
380
349
336

I also set a 5 minute w/kg PR during my first interval. TSB was -51.

Was I suffering? Definitely. Regardless, I'm having an extremely hard time beleiving the power that I put out. Also, average speed was slower than my other 355 watt interval... but it was much windier today... so it almost seems within the realm of possibility. When coasting the PT goes to zero. So there is nothing that I can do to adjust the PT, but is there anyway to test it?

I remember reading that the PT was the most consistently accurate powermeter and that the SRM was more prone to problems in this region... maybe I'm in the minority here?

Or should I be patting myself on the back for a job well done?

 

[acoggan]

I have never had reason to doubt my PowerTap, but I think today I feel like it may be misleading me. In recent weeks I've been doing 3x20's and 3x30's.

3x20 splits have been the following:
316
336
341

355
329
333

3x30's:
318
342
321

321
326
325

Today I did a 3x20 workout and got the following:
380
349
336

I also set a 5 minute w/kg PR during my first interval. TSB was -51.

Was I suffering? Definitely. Regardless, I'm having an extremely hard time beleiving the power that I put out. Also, average speed was slower than my other 355 watt interval... but it was much windier today... so it almost seems within the realm of possibility. When coasting the PT goes to zero. So there is nothing that I can do to adjust the PT, but is there anyway to test it?

I remember reading that the PT was the most consistently accurate powermeter and that the SRM was more prone to problems in this region... maybe I'm in the minority here?

Or should I be patting myself on the back for a job well done?

The fact that your PowerTap read zero when coasting doesn't necessarily mean that the zero offset was set correctly. I'd suggest examining the downloaded file from the ride to see what torque was being recorded when you were coasting.

If the above doesn't explain things, it's quite simple to perform a static calibration of the strain gages: just put your bike in the lowest gear, lock it in a trainer, hang some heavy weights from the pedal, and compared the measured torque to the value you'd expect based on the mass of the weights, crank length, and gear ratio.

 

[JohnMeyers]

Thanks Dr. Coggan. Attached is the ride file. Torque is zero when coasting...? The file may look a little strange--at the end I did a lot of standing pedalling followed by coasting.

I've just never looked down and so effortlessly put out ~400 watts for ~5 minutes.



NOTE: I just realized that Average power was 375. NP was 380.

 

[Alex Simmons]

Thanks Dr. Coggan. Attached is the ride file. Torque is zero when coasting...? The file may look a little strange--at the end I did a lot of standing pedalling followed by coasting.

I've just never looked down and so effortlessly put out ~400 watts for ~5 minutes.



NOTE: I just realized that Average power was 375. NP was 380.

Now John, you're supposed to do this yourself

Look at the power file and see if you have any sections of ride file with zero cadence but non-zero torque. That is a tell tale sign. AFAIK, the PT doesn't show negative torque zero offsets on the CPU (only positive), so that may be what's happening.

If the file does contain these non-zero torque values, then the file can be back corrected via some excel manipulation, which can be found by a search on wattage. I've done it before but not for a long time....

The procedure for checking PT calibration is listed on wattage forum under Pages.
http://groups-beta.google.com/group/wattage/
I'd post a direct link but Google is doing some maintenance and I can't seem to get onto the forum right now.

 

[otb4evr]

Thanks Dr. Coggan. Attached is the ride file. Torque is zero when coasting...? The file may look a little strange--at the end I did a lot of standing pedalling followed by coasting.

I've just never looked down and so effortlessly put out ~400 watts for ~5 minutes.



NOTE: I just realized that Average power was 375. NP was 380.

John,

The file looks good to me.

Considering that you have been doing SST and Threshold work, it makes sense that you are going to start seeing increases in your FTP.

Now that you have a 20 minute effort to use, throw down a killer 5 minute effort, a 1 minute effort and plug it into the Critical Power model and see what you come up with.

Jim

 

[JohnMeyers]

Now John, you're supposed to do this yourself

Look at the power file and see if you have any sections of ride file with zero cadence but non-zero torque. That is a tell tale sign. AFAIK, the PT doesn't show negative torque zero offsets on the CPU (only positive), so that may be what's happening.

If the file does contain these non-zero torque values, then the file can be back corrected via some excel manipulation, which can be found by a search on wattage. I've done it before but not for a long time....

The procedure for checking PT calibration is listed on wattage forum under Pages.
http://groups-beta.google.com/group/wattage/
I'd post a direct link but Google is doing some maintenance and I can't seem to get onto the forum right now.

I did look it over and it looked fine to me... I just wanted to make sure everyone elses opinion matched mine.

Now, could there be any other reason for the disparity? Just wanting to be clear.

That's just a pretty tremendous jump in ability in such a short time.

Thanks all for the help!

 

[JohnMeyers]

otb4evr...

Who do you race for? I feel like I should know you.

 

[Alex Simmons]

I did look it over and it looked fine to me... I just wanted to make sure everyone elses opinion matched mine.

Now, could there be any other reason for the disparity? Just wanting to be clear.

That's just a pretty tremendous jump in ability in such a short time.

Thanks all for the help!

I hear ya!

I had same for a trial pursuit effort a couple of months back. A 50W improvement which I had trouble believing. That's harder to do these checks on as you can't coast on a fixed gear bike.

BTW - I did look at your file and an extract in excel in PT's .csv format and the torque zero looks fine there, no obvious problems.

But like I said, when it's negative, it reports as zero.

I suppose the real truth lies in repeatability....

 

[JohnMeyers]

I hear ya!

I had same for a trial pursuit effort a couple of months back. A 50W improvement which I had trouble believing. That's harder to do these checks on as you can't coast on a fixed gear bike.

BTW - I did look at your file and an extract in excel in PT's .csv format and the torque zero looks fine there, no obvious problems.

But like I said, when it's negative, it reports as zero.

I suppose the real truth lies in repeatability....

Correct me if im wrong... but if it was negative, wouldnt it lower my average power? It clearly didn't do that!

 

[Alex Simmons]

Correct me if im wrong... but if it was negative, wouldnt it lower my average power? It clearly didn't do that!

Intuitively yes, but I don't fully understand how the CPU actually calculates when the zero offset is negative, IOW whether or not it treats negatives as such. When there is a positive offset, the outcome is as expected, i.e. higher power readings.

 

[JohnMeyers]

A quick test revealed my PT offset is at 521... which is pretty well within range.

If this is in range, is it possible for the PT to misread?

 

[lobster1]

The fact that your PowerTap read zero when coasting doesn't necessarily mean that the zero offset was set correctly. I'd suggest examining the downloaded file from the ride to see what torque was being recorded when you were coasting.

If the above doesn't explain things, it's quite simple to perform a static calibration of the strain gages: just put your bike in the lowest gear, lock it in a trainer, hang some heavy weights from the pedal, and compared the measured torque to the value you'd expect based on the mass of the weights, crank length, and gear ratio.

I would suggest a static calibration, Powertaps are not supposed to be more than 2.5% out but I was also getting some very high readings. I got a freind to calibrate it and it was 7% out. That was a bad day

 

[JohnMeyers]

With some help, I had my rear brake locked up and my bike held vertical against a wall while I teetered on one foot with all my weight on my right pedal (Crank at 3 o'clock). The highest number I saw was 820 in-lbs, though I was frequently seeing numbers as low as 780-790 in-lbs. It was tough to balance with 100% of my weight on the 1" diameter Speedplay pedals. Anyways, that number seems pretty close. The math:

25/39 x 180mm x 179.2 lbs x 1in./25.4mm = 814.05 in-lbs

(I was in the 39/25. Used 180mm cranks, and yes, am a fatass at 179.2 lbs fully clothed. Lastly, the conversion from mm to inches)

Calculate error percentage: (820-814)/814 * 100 = .73%

Thus, my guess is that my PT is pretty close to accurate (around ~1% accuracy).

 

[gza]

I've been banging on for ages for people to check their PT's but blind faith in the product seems to rule the day - there seem to be more and more cases of people finding out of spec product recently.



The only way to know for sure if your PM is reading correctly is to directly calibrate it using known weights. It seems like a hassle but I calibrate mine with weights from 5-35kg in 5kg incrediments in about 20mins - well worth it for peace of mind!



SRM's seems to get more stable with time, but you sure can't trust the factory calibration - if you are into powermeter data at some point you will fall foul of an calibration drift over the years if you don't periodically check them - I test mine every 3 months now.

 

[gza]

With some help, I had my rear brake locked up and my bike held vertical against a wall while I teetered on one foot with all my weight on my right pedal (Crank at 3 o'clock). The highest number I saw was 820 in-lbs, though I was frequently seeing numbers as low as 780-790 in-lbs. It was tough to balance with 100% of my weight on the 1" diameter Speedplay pedals. Anyways, that number seems pretty close. The math:

25/39 x 180mm x 179.2 lbs x 1in./25.4mm = 814.05 in-lbs

(I was in the 39/25. Used 180mm cranks, and yes, am a fatass at 179.2 lbs fully clothed. Lastly, the conversion from mm to inches)

Calculate error percentage: (820-814)/814 * 100 = .73%

Thus, my guess is that my PT is pretty close to accurate (around ~1% accuracy).

780-820 is a 5% difference - there's no substitute for a known weight that you can hang from the pedal with no movement.

 

[giannip]

I've been banging on for ages for people to check their PT's but blind faith in the product seems to rule the day - there seem to be more and more cases of people finding out of spec product recently.



The only way to know for sure if your PM is reading correctly is to directly calibrate it using known weights. It seems like a hassle but I calibrate mine with weights from 5-35kg in 5kg incrediments in about 20mins - well worth it for peace of mind!



SRM's seems to get more stable with time, but you sure can't trust the factory calibration - if you are into powermeter data at some point you will fall foul of an calibration drift over the years if you don't periodically check them - I test mine every 3 months now.

would you share the way you do it ?

Gettin PT in the next few days so would like to start off on the right track!

 

[Shawneebiker]

I've been banging on for ages for people to check their PT's but blind faith in the product seems to rule the day - there seem to be more and more cases of people finding out of spec product recently.



The only way to know for sure if your PM is reading correctly is to directly calibrate it using known weights. It seems like a hassle but I calibrate mine with weights from 5-35kg in 5kg incrediments in about 20mins - well worth it for peace of mind!

With a PT what can be done to correct it? Other than send it back.

 

[JohnMeyers]

Yea, I'm aware that my method is not as accurate as I'd like it to be. The goal was to get an idea if it was within range... and it was. When I had all my weight on the pedal it was hovering in the 810-820 range, but keeping balanced in that awkward position was tough.

Regardless, I'm on vacation right now and don't have all the things required to do the testing accurately. A friend of mine (who's getting his PhD in aerospace engineering) has some calibrated weights that we can use.

Further, it seems to me that you don't even need different weights... just pop it in a different gear to see how torque values change.

Anyways, I'll post my results.

 

[Alex Simmons]

Yea, I'm aware that my method is not as accurate as I'd like it to be. The goal was to get an idea if it was within range... and it was. When I had all my weight on the pedal it was hovering in the 810-820 range, but keeping balanced in that awkward position was tough.

Regardless, I'm on vacation right now and don't have all the things required to do the testing accurately. A friend of mine (who's getting his PhD in aerospace engineering) has some calibrated weights that we can use.

Further, it seems to me that you don't even need different weights... just pop it in a different gear to see how torque values change.

Anyways, I'll post my results.

FWIW - I did the same stomp test a month or so back (campag cleats obviously a little easier to stand on) and mine was bang on where it should have been.

As for the Polar - well I have three friends using them, two are operating fine and quite accurately (using an SRM for checking) once all the set up issues were resolved but another, well it just generates a load of data that bears no resemblance to what the rider is producing. I feel for the guy - he's done the tests I've asked so I can help him determine FTP / set some training levels but the data is useless. Been through all the set up (weighing chain, careful positioning of pick up etc etc etc all the stuff Tom talks about) but with this one user it is simply "not fool proof enough" for them (not the shapest knife in the drawer, who knows what settings they inadvertently change or bits they bump). We are gunna need to migrate them to a PT methinks.

 

[gza]

With a PT what can be done to correct it? Other than send it back.

Nothing - its just for piece of mind. Like you say if you find its out of spec you have to return it - having no data for a few weeks is less of a problem than many months of incorrect data in my opinion!