Stomp Test is off -- what now?

frenchyge said:

did the stomp test described here.

Specifically how were you applying weight to the cranks. Human body? External weights? If the former, are you sure weight was applied fully onto the pedal?

frenchyge said:

I got suspicious when I noticed a large deviation between my PT Pro watts and my Kinetic Road Machine power curve, so I zeroed the torque and did the stomp test described here.

I repeated the test using 5 separate gear combinations, and got errors (from calculated torque) ranging from -6.1% to -7.7%. ...

Did you use a fixed weight or body weight?

FWIW, all my PT hubs are within 3% when torque tested with a fixed weight, body weight tests seem to have a lot more variability depending on how well I balance on the pedal and how else I stabilize myself. 2 of my SL road hubs are within 1.5% and my Pro MTB hub is high by roughly 2.8%.

I've started using a slightly different method for torque testing that's easy. I hang a digital scale from the seat rails in a position that allows me to attach the other end to a horizontal crank in the upstroke position. I use spectra climbing runners to connect the links to minimize stretch. I spin the wheel forward and then roll it backwards till the scale connection is tight and horizontal and the weight on the scale is in the 50 to 75 pound range. I then slide wooden dowel horizontally between a spoke and the seat stays to lock the wheel under load. Wait a few seconds for the system to settle and read weight and torque. It's easy to vary load and of course you can use various gears just like the other method. The results match a 50 pound weight plate very well and it's a lot more convenient.

Personally if your measurements are solid I'd say you should send the hub in for service.

-Dave

Steve_B said:

Specifically how were you applying weight to the cranks. Human body? External weights? If the former, are you sure weight was applied fully onto the pedal?

Body weight, carefully applied by supporting the bike in the trainer, clamping the rear brake with a hand clamp, putting the crank as close to level as the nearest rachet pawl would allow, standing upright with my right foot on left pedal and balancing using flat hands lightly touching vertical surfaces above my head.

I was being careful.... not Robert Chung careful.... but careful. 😉

frenchyge said:

Body weight, carefully applied by supporting the bike in the trainer, clamping the rear brake with a hand clamp, putting the crank as close to level as the nearest rachet pawl would allow, standing upright with my right foot on left pedal and balancing using flat hands lightly touching vertical surfaces above my head.

I was being careful.... not Robert Chung careful.... but careful. 😉

Not good enough. Borrow a couple of 25 lb plates and hang them from a pedal.

I tried using body weight the first time I checked mine too...I was convinced it read 5% low. Then I used the plates and found it was spot on.

The one nice thing about using separate weights is that you can VEERRRYYY slowly rotate the wheel by hand forward and back to find the peak reading.

That's too bad when you're noticing differences between your KK and Powertap... and the PT is the one that's off! I was in your shoes awhile ago when my workouts started getting really tough. I did the stomp test with body weight and it seemed to be 7% off. It may not be as accurate as these other methods, but accurate enough for me to know it was time to send her in. Saris only warranties it for a year, so you'll have to pay for the re-calibration. I think they said it was $150, but don't quote me. I must say, though, that Saris' customer service is without equal. When I got mine back in the mail, I opened it to find a new chest strap, wiring harness, computer, and they overhauled the rear hub for me too. Excellent people. Now I just wish I could recover from those 2 X 20 VO2 sessions....:p

daveryanwyoming said:

Did you use a fixed weight or body weight?

FWIW, all my PT hubs are within 3% when torque tested with a fixed weight, body weight tests seem to have a lot more variability depending on how well I balance on the pedal and how else I stabilize myself. 2 of my SL road hubs are within 1.5% and my Pro MTB hub is high by roughly 2.8%.

I've started using a slightly different method for torque testing that's easy. I hang a digital scale from the seat rails in a position that allows me to attach the other end to a horizontal crank in the upstroke position. I use spectra climbing runners to connect the links to minimize stretch. I spin the wheel forward and then roll it backwards till the scale connection is tight and horizontal and the weight on the scale is in the 50 to 75 pound range. I then slide wooden dowel horizontally between a spoke and the seat stays to lock the wheel under load. Wait a few seconds for the system to settle and read weight and torque. It's easy to vary load and of course you can use various gears just like the other method. The results match a 50 pound weight plate very well and it's a lot more convenient.

Personally if your measurements are solid I'd say you should send the hub in for service.

-Dave

That's pretty nifty - got a photo?

Tom Anhalt said:

Not good enough. Borrow a couple of 25 lb plates and hang them from a pedal.
I tried using body weight the first time I checked mine too...I was convinced it read 5% low. Then I used the plates and found it was spot on.

But I've seen comments from people that weightlifting plates can vary by a few percent in their weight, too and have to be independently weighed for accuracy. 😕 I don't have access to much in the way of quality scales, weights, etc. to do a really thorough job or it. By using something as heavy as myself, I'd hoped that any errors in absolute weight would have smaller relative effect.

Tom Anhalt said:

The one nice thing about using separate weights is that you can VEERRRYYY slowly rotate the wheel by hand forward and back to find the peak reading.

I think that's a really good approach, but any small angle error from the true horizontal just effectively shortens the crank length, right? Seems like it'd take an error of 5.7 degrees to make a .5% difference in the reading.

Bikeridindude said:

That's too bad when you're noticing differences between your KK and Powertap... and the PT is the one that's off!

Heh, I didn't notice the difference while riding with both. I wasn't until I had to send my hub in for an overhaul that I decided to do a couple sessions using the KK speed chart to set my trainer target.

The first thing I noticed was that the PT computer makes a *terrible* cycling computer when used with a separate wheel magnet. I don't know if it's because of the sampling frequency or what, but the speed resolution is in steps of about .7 mph (ie, speed is 20.2 mph or 20.9 or 21.7 mph, with nothing in between). Then, when I downloaded my workout file and applied the KK power formula I thought, hmmmm.... that didn't feel like a new FTP test. 🙄

Tom Anhalt said:

The one nice thing about using separate weights is that you can VEERRRYYY slowly rotate the wheel by hand forward and back to find the peak reading.

Right. If you stand on the pedal and move your weight around, you will see huge variations in the indicated torque. If you try to use your weight, it's hard to get 100% of it through the pedal.

Free weight plates might be labelled inaccurately, but you can weigh them at the post office. Every town in the US has a post office.

frenchyge said:

I think that's a really good approach, but any small angle error from the true horizontal just effectively shortens the crank length, right? Seems like it'd take an error of 5.7 degrees to make a .5% difference in the reading.

That's still better than -6.0 %, no? Use a bubble level to verify that your floor and crank arm are level (to the extent that you can) and see what you get. I think that you will get closer than 5 degrees with a bubble level.

Steve_B said:

That's still better than -6.0 %, no? Use a bubble level to verify that your floor and crank arm are level (to the extent that you can) and see what you get. I think that you will get closer than 5 degrees with a bubble level.

Yeah, that's what I was saying -- that I was thinking I could get closer than 5 degrees to horizontal and that my potential error would be less than .5% of the true peak torque as a result. I did check the crankarm with a bubble level, and raised it to the next ratchet pawl if it was below level at all to allow for the flex that came with me putting my weight on one pedal.

Maybe I'll try some more measurements before sending it in for calibration.

No real need for a level - eyeball it and then slowly roll the wheel back and forth. You'll see the torque peak and fall off. Record the peak value.

I got a 35lb weight and took it to the UPS store to verify - it happened to be 35.00 lbs exactly. Girl at the store thought I was a bit daft.

Alex Simmons said:

That's pretty nifty - got a photo?

I'll take one this evening and post it.

daveryanwyoming said:

I'll take one this evening and post it.

That approach does sound a lot more manageable than borrowing some weightlifting plates and taking them somewhere to be weighed. I'm visualizing that the dowel holds everything in place and steady for the reading, but this method doesn't give you the ability to rock back and forth over the peak does it? You're using eyeball or bubble level to find the right crank position then?

frenchyge said:

But I've seen comments from people that weightlifting plates can vary by a few percent in their weight, too and have to be independently weighed for accuracy. 😕 I don't have access to much in the way of quality scales, weights, etc. to do a really thorough job or it. By using something as heavy as myself, I'd hoped that any errors in absolute weight would have smaller relative effect.

Hmmm...if that's the case, how do you know your body weight to any accuracy?

The plates are just a suggestion. Another variable weight source you could use is a bucket filled with a measured amount of water. Just need to figure out how to elevate the bike so you can hang the handle from the pedal spindle.

Here's a link to a BTR thread on this subject that includes some pics from a torque test.

http://biketechreview.com/phpBB2/viewtopic.php?t=1951

frenchyge said:

That approach does sound a lot more manageable than borrowing some weightlifting plates and taking them somewhere to be weighed. I'm visualizing that the dowel holds everything in place and steady for the reading, but this method doesn't give you the ability to rock back and forth over the peak does it? You're using eyeball or bubble level to find the right crank position then?

You can rock back and forth for maximum torque readings, but then it's unlikely a dowel will be at just the right place to lock off the rear wheel. The length of the straps, chains, whatever you use to connect the scale to the cranks determines how the spokes line up relative to horizontal cranks. You could just hold it steady and take the readings, but you'd either have to crane your neck to read both instruments from back there or get a helper.

Horizontal is a good starting point and I considered a bubble level on the cranks as others have suggested but 90 degrees between the straps to the scale and the crank is really what you're after with this method. A peak torque reading is still the goal and you'll get that when the force(in this case the tension through the scale) is at a right angle to the crank.

I also scratched a bit of trig to see how big an effect an angle error makes on either method to justify my eyeballing approach. It takes +/- 8 degree error in angle relative to horizontal cranks or perpendicular force application to create a 1% error in effective crank length. And the created torque will always be lower than desired regardless of which direction the angle is in error. IOW, either method is fairly tolerant to small angle errors and cranks will effectively shorten with angle error not lengthen. A +/- 5 degree error in force application results in an error of ~ 0.004, well below the accuracy of power meters.

I'll try to post some pics.
-Dave

Tom Anhalt said:

Hmmm...if that's the case, how do you know your body weight to any accuracy?

That's a good question, and the answer is that I don't. My thinking is that I can estimate my body weight within a couple pounds, which represents a pretty small relative error because of the large weight involved. I could see that inaccurate weight/force would be a potentially big source of error, but I'm not really seeing that crank angle or the inability to rock over the peak is going to introduce a noticeable effect (beyond the 1-3% accuracy that I would expect from a PT). I wonder if the post office would let me climb on their scale? 😄

My questioning of weightlifting plates was just based on comments I've seen that they can even be off by a couple pounds, which would be a much larger error from a 25lb plate. Calibrated plates like the BTR pics would be ideal if I could find a set.

Tom Anhalt said:

Here's a link to a BTR thread on this subject that includes some pics from a torque test.

biketechreview.comviewtopic.phpOpen ↗

Interesting discussion, thanks. Looks like common sense prevailed in the end. 🙂 I could see a huge source of error from someone standing over their bike on one foot, with the opposite hand holding the brake steady. That's why I clamped the brake and stepped off a chair to place right foot on left pedal to be sure that I was able to stay upright and centered over the pedal. The displayed torque was definitely stable when I was balanced steadily with no bouncing.

What about the comments from the one crazy guy about the torque setting (offset, right?) in the PT computer itself? It looks like that just adjusts the zero value, and as long as the unloaded torque shows zero there shouldn't be anything to change there, right?

Thanks everyone for the comments and help. 🙂

frenchyge said:

That's a good question, and the answer is that I don't. My thinking is that I can estimate my body weight within a couple pounds, which represents a pretty small relative error because of the large weight involved. I could see that inaccurate weight/force would be a potentially big source of error, but I'm not really seeing that crank angle or the inability to rock over the peak is going to introduce a noticeable effect (beyond the 1-3% accuracy that I would expect from a PT). I wonder if the post office would let me climb on their scale? 😄

My questioning of weightlifting plates was just based on comments I've seen that they can even be off by a couple pounds, which would be a much larger error from a 25lb plate. Calibrated plates like the BTR pics would be ideal if I could find a set.

Interesting discussion, thanks. Looks like common sense prevailed in the end. 🙂 I could see a huge source of error from someone standing over their bike on one foot, with the opposite hand holding the brake steady. That's why I clamped the brake and stepped off a chair to place right foot on left pedal to be sure that I was able to stay upright and centered over the pedal. The displayed torque was definitely stable when I was balanced steadily with no bouncing.

What about the comments from the one crazy guy about the torque setting (offset, right?) in the PT computer itself? It looks like that just adjusts the zero value, and as long as the unloaded torque shows zero there shouldn't be anything to change there, right?

Thanks everyone for the comments and help. 🙂

that 'raw' torque setting should be around 512 ???? or so - definitely within a certain range or it indicates there's definitely trouble.

I had an m/b hub in which that 'raw' torque value decreased to 300 or so. Saris told me to send in the hub. Symptoms first were that the power was reading WAY WAY low.

Just a total [censored] of a thought but are you ensuring you zero your PT before doing any of those stomp/weight tests. err .. stranger things have happened . . . ... ... . .

frenchyge said:

My questioning of weightlifting plates was just based on comments I've seen that they can even be off by a couple pounds, which would be a much larger error from a 25lb plate. Calibrated plates like the BTR pics would be ideal if I could find a set.

They're readily available but can be so expensive (I've seen some that are $600) that it would be far cheaper to buy a digital scale and accurately weigh weightlifting plates.