Treadmill accuracy is awful

Thanks, but I've already done that and followed the recommendations. At this price, the accuracy should be excellent regardless of the type of exercise, even with significant pace variations (Z4/Z5 to Z1). I should point out that my FR935 was more accurate, ironically.

Couple questions for my knowledge:  What training load are you / device using that relies on pace vs just HR?  

To utilize likely the same sensor to calculate step & hang-time rates... to then corelate that to your outdoor paces recently (which are also effected by wind/hill/dirt/etc)... creates quite the mess of cascading error factors.  

Have you ever looked at the graph data for Cadence, Ground Contact Time, Stride Length and Vertical Ratio/Oscillation...?   They are fairly often riddled with inconsistencies. 

Now it would be nice if there was a way to select a time frame and help the treadmill calibration...  like select a nice 100m stretch from the graph... "use as good representative data for this pace"  ....which you could then do to help improve the calibration around a handful of important paces.   But that gets skewed pretty fast by the fact that we don't run the same on the Tread as we do outside - especially coupled with adjustments of the treadmill slope.   

Might be worth it to get a foot pod.

I have often run on treadmills in hotels, all of them were different brands and different levels of maintenance, so I had to calibrate after nearly every treadmill run. Got tired of that, dusted off an old foot pod, calibrated that outdoors, and problem solved.

My FR165 was working perfectly well for treadmill runs up until a week ago. Suddenly, the pace is all over the place. Even after calibrating it shows far more fluctuations than it did previously for similar runs.