SRM Zero Offset Drift

Aztec said:

I'm new to the SRM (Pro). In my first ride, I saw my watts creep up to an unrealistic level. Yes, it got warmer on the ride because I first reset it in my 65F garage, and then rode in the sun (maybe 70F).

The zero offset value rose approx 200. Toward the end of the hour ride, my watts were again up.

This can't be normal. Are all you SRM users going through this? If so, YUCK.

If you just (re)installed your crank, what you observed is normal, but it wasn't due to the temperature increase. Rather, the residual strain in the crank likely increased after you stomped on it a few times. For example, I just got my SRM Pro back from Colorado Springs after a battery change, and the initial zero offset when in the workstand was in the low 300 range. After riding for a short while, though, it went to ~440, which is where it has stayed. It is because of this behavior that SRM recommends setting the zero offset only after you have "bedded in" the crank by applying a high torque several times.

The good news is that your crank will likely quickly settle at a fairly constant value, such that you can correct the data from today's ride using the "Properties" option under the "File" menu in the SRMWin software.

acoggan said:

If you just (re)installed your crank, what you observed is normal, but it wasn't due to the temperature increase. Rather, the residual strain in the crank likely increased after you stomped on it a few times. For example, I just got my SRM Pro back from Colorado Springs after a battery change, and the initial zero offset when in the workstand was in the low 300 range. After riding for a short while, though, it went to ~440, which is where it has stayed. It is because of this behavior that SRM recommends setting the zero offset only after you have "bedded in" the crank by applying a high torque several times.

The good news is that your crank will likely quickly settle at a fairly constant value, such that you can correct the data from today's ride using the "Properties" option under the "File" menu in the SRMWin software.

Thanks for the info.

Thankfully, I'm not at all worried about a single day's data (I rode w/ my PT Pro still hooked up). And I am DELIGHTED to see that you ride an SRM. FWIW, after resetting the SRM once seeing preposterous watts, it was literally within 1w average for the next ~30 mins. It was then that it began to drift again.

Some other things here that might (?) be messing with my rapid offset change... SRM either forgot to include crank bolts or just doesn't supply them, so I used my Campy bolts. And I don't have a torque wrench for hex heads, so I torqued by feel. Finally, when I got home, I realized the sensor was lightly dragging against the white plastic cover at one point of the crank's revolution. Maybe none of this matters, maybe these are considered obscenely sloppy for an SRM owner. :-)

Quoted post said:

I am DELIGHTED to see that you ride an SRM.

I bought an SRM track crank (for my wife to use) in early '02, then switched from a PowerTap to an SRM on my road bike in the fall of '02 when I found a used Pro at a good price. I occasionally still use a PowerTap, but mostly use the SRM.

Quoted post said:

Some other things here that might (?) be messing with my rapid offset change... SRM either forgot to include crank bolts or just doesn't supply them, so I used my Campy bolts. And I don't have a torque wrench for hex heads, so I torqued by feel. Finally, when I got home, I realized the sensor was lightly dragging against the white plastic cover at one point of the crank's revolution. Maybe none of this matters, maybe these are considered obscenely sloppy for an SRM owner. :-)

Inadequate torque on the crank bolts might result in more variation in the zero offset than there should be, but I doubt it...square taper arms always squirm up the tapers as far as they can, so presumably the initial torque shouldn't really matter (much). Having the sensor rubbing on the disk also shouldn't make any significant difference. However, if it is too far away, then the zero offset will tend to be unstalbe...it should only vary by a couple of Hz when you turn the cranks slowly backwards.

I figured those wouldn't be causes. I just hope this stabilizes so I can get confident enough to yank the PT stuff off there!

I couldn't seem to get the Power V to go into offset mode while actually riding. I'm assuming I just haven't figured it out? That must be the case, as I can't imagine the engineers thinking you'd stop mid-race to check/adjust it. :-)

The SRM docs are a little thin on how the controller works. But after having had a Hac4 awhile ago, etc., I'm getting used to Germanglish.

acoggan said:

I bought an SRM track crank (for my wife to use) in early '02, then switched from a PowerTap to an SRM on my road bike in the fall of '02 when I found a used Pro at a good price. I occasionally still use a PowerTap, but mostly use the SRM.

Inadequate torque on the crank bolts might result in more variation in the zero offset than there should be, but I doubt it...square taper arms always squirm up the tapers as far as they can, so presumably the initial torque shouldn't really matter (much). Having the sensor rubbing on the disk also shouldn't make any significant difference. However, if it is too far away, then the zero offset will tend to be unstalbe...it should only vary by a couple of Hz when you turn the cranks slowly backwards.

New issue... and follow up.

First, the offset drift has settled way down in just 50 total miles.

But now I have a different concern. While the power measure between the SRM and the PT is very similar at low wattages (e.g., 130-170w), it begins to get pretty wide by 200-240w. The SRM consistently shows 10-15w low in that range. And although I am way too tired to test it thoroughly at 300w, etc., ~30 seconds at that level showed ~20w difference (confirmed in my Cycling Peaks peak efforts).

I don't want to be too particular about the differences, but 10-15w is pretty big at your FTP! Can I mess with the slope to make it match my Powertaps (which are all consistent w/ each other and have similar spreads vs. my Computrainer)?

Aztec said:

New issue... and follow up.

First, the offset drift has settled way down in just 50 total miles.

That's good.

Aztec said:

But now I have a different concern. While the power measure between the SRM and the PT is very similar at low wattages (e.g., 130-170w), it begins to get pretty wide by 200-240w. The SRM consistently shows 10-15w low in that range. And although I am way too tired to test it thoroughly at 300w, etc., ~30 seconds at that level showed ~20w difference (confirmed in my Cycling Peaks peak efforts).

I don't want to be too particular about the differences, but 10-15w is pretty big at your FTP! Can I mess with the slope to make it match my Powertaps (which are all consistent w/ each other and have similar spreads vs. my Computrainer)?

Entering a lower slope will increase the calculated power. However, rather than depending on a cross-comparison against your PowerTaps to arrive at the new figure, I would recommend that you buy some weights and do a multi-point static calibration. That way you'll know that the data the SRM is giving you are truly correct, and can then evaluate it against your other devices to see where/when there are/are not any differences. For example, by using this approach I know that my SRM reads a constant 7-9 W greater than my PowerTap across the range of 50 to 350 W. (IOW, the efficiency of my drivetrain seems to go up with increasing power output, as you would expect it to.) Cross-calibrating the SRM against the PowerTap using just power data would obscure this offset, and the slope that you arrived at would vary depending on the power(s) at which you did the calibration.

The other reason that I recommend a static calibration is that I've found the dynamic approach (i.e., using power data collected while pedaling) to be a bit more variable. This might be due to the fact that we don't always pedal in precisely the same manner on every occasion/at every power. In any case, I figure that if I'm gonig to go to the trouble of calibrating something, I want data that are spot-on.

acoggan said:

I would recommend that you buy some weights and do a multi-point static calibration.

I see the next question forming...

srm.dePMCalibrate.htmOpen ↗

Scott

scotmart said:

I see the next question forming...

srm.dePMCalibrate.htmOpen ↗

Scott

:-)

You do realize, though, that those weights are extremely inaccurate. You'd have to weigh them on a reliable scale first.

acoggan said:

That's good.

Entering a lower slope will increase the calculated power. However, rather than depending on a cross-comparison against your PowerTaps to arrive at the new figure, I would recommend that you buy some weights and do a multi-point static calibration. That way you'll know that the data the SRM is giving you are truly correct, and can then evaluate it against your other devices to see where/when there are/are not any differences. For example, by using this approach I know that my SRM reads a constant 7-9 W greater than my PowerTap across the range of 50 to 350 W. (IOW, the efficiency of my drivetrain seems to go up with increasing power output, as you would expect it to.) Cross-calibrating the SRM against the PowerTap using just power data would obscure this offset, and the slope that you arrived at would vary depending on the power(s) at which you did the calibration.

The other reason that I recommend a static calibration is that I've found the dynamic approach (i.e., using power data collected while pedaling) to be a bit more variable. This might be due to the fact that we don't always pedal in precisely the same manner on every occasion/at every power. In any case, I figure that if I'm gonig to go to the trouble of calibrating something, I want data that are spot-on.

Thanks for the detailed thoughts. Very useful.

Now, if you had 2+ years of data in CP, would you still want to calibrate the *new* powermeter precisely, or would you simply want continuity with the old powermeter?

Aztec said:

Thanks for the detailed thoughts. Very useful.

Now, if you had 2+ years of data in CP, would you still want to calibrate the *new* powermeter precisely, or would you simply want continuity with the old powermeter?

What makes you think year-to-year reliability of your old powermeter is precise? In any case, while year-to-year comparisons are interesting, they aren't teribbly helpfull. Personally I'd want to know my current data is accurate (or as accurate as I could get).

Scott

P.S. And yes, I'm aware of the AIS study...

Aztec said:

Thanks for the detailed thoughts. Very useful.

Now, if you had 2+ years of data in CP, would you still want to calibrate the *new* powermeter precisely, or would you simply want continuity with the old powermeter?

I'm actually in the same boat that you are, in that I used a PowerTap exclusively from 1999 to 2002 and an SRM exclusively from 2003 to 2005. This year, I have been using both. I have opted to calibrate each powermeter precisely, and have accepted that this means some "discontinuity" in the data (although in fact I have been manually correcting this season's PowerTap data upward by 2.5% to account for the difference between the devices, such that it is really just the 1999-2002 vs. 2003-2006 comparison that I really have to worry about). In any case, however, this doesn't really pertain to whether you choose to do a static or a dynamic calibration of your SRM, since you could always use the statically-determined slope to arrive at a new slope that gives numbers comparable to your PowerTap(s).

PS. In part, my perspective on this issue is colored by attempting to calibrate my track SRM in a dynamic manner, by back-calculating the frequency from power and cadence (and assumed slope) when pedaling the track bike on the Velodyne at varying power outputs. I was able to get reasonable numbers this way, but they were more variable than when performing a static calibration.

acoggan said:

PS. In part, my perspective on this issue is colored by attempting to calibrate my track SRM in a dynamic manner, by back-calculating the frequency from power and cadence (and assumed slope) when pedaling the track bike on the Velodyne at varying power outputs. I was able to get reasonable numbers this way, but they were more variable than when performing a static calibration.

How were you able to calibrate the track bike on the Velodyne without being able to "coast down"?

JustCurious said:

How were you able to calibrate the track bike on the Velodyne without being able to "coast down"?

I pedaled it up to speed (>25 mph) while wearing tennis shoes, then just stepped off the pedals when the Velodyne beeped to indicate that I should start coasting. At least in theory, when you do it this way the coast-down parameters should include frictional losses in the drive train (albeit not under load) as well as rolling resistance, thus making the SRM and Velodyne data (more) directly comparable.

Pooling the data from three such experiments (on different days), I came up with an R^2 between the frequency "spit out" by the SRM crank and the torque being dynamically applied as I pedaled for 3 min at each of 50, 100, 150, 200, 250, 300, and 350 W (as measured/set by the Velodyne) of 0.9974. The slope, though, was only 14.1 Hz/Nm, vs. 14.7 Hz/Nm as determined statically. More disconcertingly, the intercept of this relationship always turned out to be higher than the zero offset of the SRM. I was never able to really figure out why this might be true, and I was never entirely comfortable with the idea of using the Velodyne to calibrate the SRM (vs. vice-versa). So, when all was said and done I went back to doing what I've always done, which is to statically calibrate my powermeters, then cross-check/use them to verify the accuracy of the Velodyne.

acoggan said:

That's good.

Entering a lower slope will increase the calculated power. However, rather than depending on a cross-comparison against your PowerTaps to arrive at the new figure, I would recommend that you buy some weights and do a multi-point static calibration. That way you'll know that the data the SRM is giving you are truly correct, and can then evaluate it against your other devices to see where/when there are/are not any differences. For example, by using this approach I know that my SRM reads a constant 7-9 W greater than my PowerTap across the range of 50 to 350 W. (IOW, the efficiency of my drivetrain seems to go up with increasing power output, as you would expect it to.) Cross-calibrating the SRM against the PowerTap using just power data would obscure this offset, and the slope that you arrived at would vary depending on the power(s) at which you did the calibration.

The other reason that I recommend a static calibration is that I've found the dynamic approach (i.e., using power data collected while pedaling) to be a bit more variable. This might be due to the fact that we don't always pedal in precisely the same manner on every occasion/at every power. In any case, I figure that if I'm gonig to go to the trouble of calibrating something, I want data that are spot-on.

I read through the SRM calibration method posted above. Seems simple enough so long as I can find a way to keep the bike stable in the air and find a scale that accurate. How did you test across a range of power levels?

Aztec said:

I read through the SRM calibration method posted above. Seems simple enough so long as I can find a way to keep the bike stable in the air and find a scale that accurate.

It's easier to just lock the bike into a trainer, especially if you have one that clamps the front fork (e.g., your CompuTrainer). You can then just grab the rear wheel and rotate it backwards to lift the weights into the air (hint: this will be easier if you put it in the big ring...although I'd aim for a straight-ish chainline, since 1) that can have a small effect and 2) presumably that's where you ride most of the time anyway). Also, this way there is no need to use a level, as you can just rotate the wheel back and forth to find the spot at which the frequency is maximized. Finally, life is a lot easier if you 1) mark the reed switch under the white plastic cover with a marker or a piece of tape so you can find it easily, and 2) have a strong magnet available with which to "wake up" the crank periodically. (Of course, Murphy's Law says that the reed switch is going to end up behind the chainstay where you can't easily get to it, thus requiring you to remove the weights to rotate the crank far enough to activate it, then put them back on.)

Aztec said:

How did you test across a range of power levels?

After calibrating the powermeter and the Velodyne, I then ride for ~3 min at various powers, tossing the first ~30 s worth of data (since it takes a bit for the Velodyne to "settle", plus I will sneak in my butt breaks during that time) and using the average over the last ~2.5 min for comparison. You probably don't have to go quite so long to get a representative average, but when it comes to numbers I tend to be a bit compulsive. 😄

Aztec said:

I read through the SRM calibration method posted above. Seems simple enough so long as I can find a way to keep the bike stable in the air and find a scale that accurate. How did you test across a range of power levels?

I put my bike in a rear-wheel trainer, and the front wheel up on a paintcan. I figure it makes sense to calibrate it with the cranks in the position where maximum torque occurs (my set up makes the cranks level at about the 4 o'clock position). Use a velcro strap on the brake lever to lock the rear wheel.

For scales, any comercial business that deals in weights will have a certififed scale. I used a MailBox Etc near my house.

Scott

acoggan said:

It's easier to just lock the bike into a trainer, especially if you have one that clamps the front fork (e.g., your CompuTrainer). You can then just grab the rear wheel and rotate it backwards to lift the weights into the air (hint: this will be easier if you put it in the big ring...although I'd aim for a straight-ish chainline, since 1) that can have a small effect and 2) presumably that's where you ride most of the time anyway). Also, this way there is no need to use a level, as you can just rotate the wheel back and forth to find the spot at which the frequency is maximized. Finally, life is a lot easier if you 1) mark the reed switch under the white plastic cover with a marker or a piece of tape so you can find it easily, and 2) have a strong magnet available with which to "wake up" the crank periodically. (Of course, Murphy's Law says that the reed switch is going to end up behind the chainstay where you can't easily get to it, thus requiring you to remove the weights to rotate the crank far enough to activate it, then put them back on.)

After calibrating the powermeter and the Velodyne, I then ride for ~3 min at various powers, tossing the first ~30 s worth of data (since it takes a bit for the Velodyne to "settle", plus I will sneak in my butt breaks during that time) and using the average over the last ~2.5 min for comparison. You probably don't have to go quite so long to get a representative average, but when it comes to numbers I tend to be a bit compulsive. 😄

Ah, I see now.

Hmm. How do you actually hang the weights if you are using a Computrainer? Either those weights are very small or the CT is up on blocks. And I assume SRM's use of a "pedal blank" is superfluous and that I can just hook something over my pedals.

I don't see why the SRM document is so concerned about starting w/ a level bike and level crank arm. It's all about finding the highest frequency, no?

scotmart said:

I put my bike in a rear-wheel trainer, and the front wheel up on a paintcan. I figure it makes sense to calibrate it with the cranks in the position where maximum torque occurs (my set up makes the cranks level at about the 4 o'clock position). Use a velcro strap on the brake lever to lock the rear wheel.

For scales, any comercial business that deals in weights will have a certififed scale. I used a MailBox Etc near my house.

Scott

Thanks for the ideas. The UPS store weighed my old weights.

SRM sent me their calibration spreadsheet and instructions. Interestingly, they only suggest you measure with the crank at 3 o'clock/level. No looking for the max frequency as the crank rotates upward.

I'm anxious to get at this project!

Hi Andy, what do you mean by a 'multipoint calibration'? I test at 3 o'clock, crank forward obviously, on both sides, with wieghts of 5, 10, 15, 20, 25 30 and 35kg all certified masses. Compared to the PT on my trainer the SRM then reads 11w higher at 200w through to about 15w higher at 450w. When I checked the PT it gave readings 1% low, the Ergomo....well thats work in progress!

Gavin