Foot pod calibration, best way?

The Stryd developers have investigated why the Stryd is less accurate on some threadmills, and they made an interesting discovery:

The speed of the belt oscillates with your running cadence. While you are in the air, the belt runs with one speed. While your feet have contact to the belt, the belt runs with another speed. Obviously, the speed of the belt while you are in the air doesn't affect your running effort. But it affects the measured belt distance and average speed during the run. So that would be one reason for not using a thread mill for calibration.

I picked a piece of road with marker poles at fixed distances. Then I started running with the GPS turned off and the foot pod set to measure both pace and distance. But this approach requires that you can trust the marker poles. The reliability of those poles will probably be different in different countries.

I had not even considered that, that's interesting.I always thought these things were stubbornly even and the runner has to adapt :-)
Thanks, that's also a new viewing angle.

I does not always run the same pace, sometimes slower sometimes faster for intervals, should Itake the "golden mean" for the measurement? That mean to calibrate it with a pace which is in avergae to my pace range?

From most things I have read, using the treadmill to calibrate if not a great solution. Besides the points mentioned previously, treadmills are not all that accurate from machine to machine. I tested this a while ago - foot pod calibration fixed at 100 - using same treadmill speed settings on two side by side machines and ran 3k according to the treadmills. Footpod distances where 2.92 and 3.01k. Not scientific obviously and cannot tell what caused what.

Calibration on a track is likely the most accurate but running outside in good GPS using autocalibration a number of times may give you a "good enough" factor depending on how accurate you really want it to be. Stryd (much more than a foot pod and much more expensive) may not need any calibration (manual suggest Autocalibration OFF and calibration factor left at 100). Another good way is to use the Fellrnr tool (link to a DCR article on this below) to assess a bunch of runs using TCX exports - takes a bit of work however.

https://www.dcrainmaker.com/2011/06/garmin-ant-footpod-calibration-tool.html
https://www.dcrainmaker.com/2011/01/garmin-ant-foot-pods-everything-you.html

I does not always run the same pace, sometimes slower sometimes faster for intervals, should Itake the "golden mean" for the measurement? That mean to calibrate it with a pace which is in avergae to my pace range?

Your calibration does not necessarily change with pace. The foot pod measures the distance of each step, using accelerometers. So even though you run with different step length at different paces, you can use the same calibration as long as the pod has the same error (in percent) in its measurement of step length at those paces.

What I have experienced with my Garmin foot pod (which caused me to ditch it and buy a Stryd) is that the error on the step length measurement tends to change abruptly at a certain pace threshold. But below and above this threshold the same two factors can be used with great accuracy over a wide range of paces, as long as you don't change your running style.

If you want to test this, once again you can use the marker poles at the road side. Or a roundtrack with known length. Run with GPS disabled and foot pod speed and distance enabled. Start running at your slowest running pace, and every time you pass a marker pole, press the Lap button on the watch and also increase your pace a bit. At the end, you will have a logged activity with 1 split for each marker pole. Knowing the exact time for each split and the distance between the poles, you can then calculate your actual pace for each split, which you can compare to the average pace for the split reported by the watch. Then it will be very obvious to see if you have a pace dependent error.

I did this test for both my Garmin foot pod and my Stryd foot pod. The pace range is not a very typical pace range, but I do a lot of low heart rate running, and those are the paces I use. The vertical axis in the diagram shows the calibration factor which would have been optimal for each individual split.

As you can see, for running at any speed between 5 and 7 km/h (walking speed!), a calibration factor of 123% would be the best choice throughout the whole range. For running at any speed between 8 and 10 km/h, a calibration factor of 100% would be the best choice. And for walking, I would need a third calibration factor:
[IMG2=JSON]{"alt":"Garmin pod","data-align":"none","data-size":"full","height":"421","title":"Garmin pod","width":"673","src":"http:\/\/i66.tinypic.com\/2h690zc.png"}[/IMG2]

For the Stryd, I can use a calibration factor of 97% for any activity. Perhaps a percent or two lower, if I was only moving at walking speeds, but I don't bother with that.
[IMG2=JSON]{"data-align":"none","data-size":"full","height":"421","width":"673","src":"http:\/\/i67.tinypic.com\/264gabb.png"}[/IMG2]

Hi,
many thanks for the tips.
I made my first run with it, left hand a Forerunner 235 and right hand Fenix 5x with the foot pod (pace and distance from pod), foot pod out of the box not calibrated, just to see if the pod works :-)

the distance is with pod 80 longer and the pace 5 seconds higher. I would have thought it was uncalibrated more different.

A question to the pro's. I know that the milestone could be calibrated byself. Is there empirical values if it is better to calibrate via Garmin or directly the pod?

I've gotten pretty reasonable results by just running outside with my footpod 4 or 5 times.

Hi,

additional info. I found it also in other posts.
Seems that the finx5x ignores the Distance "Always" settings.
I set both, Distance and Speed, to Always, but in board and tcx files still the GPS distance is saved.
How I found it (for me). With the foot pod calibration tool from fellrnr. This tool shows me the GPS distance and Pod distance. The GPS distance is showing in the statistic (and tcx files).
Is my experience right?