Sense not tracking run and pace properly

[KerrieD23]

My Sense is not properly tracking my run and pace. While using the gps and run exercise app, it’s showing my pace at 59/60 mph while running at least 4.2mph on the treadmill. I’ve had this sense since fall and it’s always tracked correctly, until this week when I’m trying to focus on goals/pace for the new year 🤦🏼‍:female_sign: I’ve done a hard reset and make sure it’s charged to over 80%. Help! It’s making my daily workout so frustrating!  

[JohnnyRow]

First thing, GPS makes no sense on treadmill.  Your position is not changing.  I'm not even reading past that.

@KerrieD23 @LizzyFitbit @StefanieFitbit @EllenFitbit 

Edited: I guess some interpreted the part I just struck through as harsh. My meaning was that it didn't seem to me to make sense to deal with other parts of the question as long as you were trying to use GPS to get pace while in a stationary position.  Once that primary initial error was corrected, then I would have been glad to come back and see how you fared after correcting that. 

Perhaps I sometimes err on the side of not always giving long courteous enough responses, in the hope of getting to more of the user questions, which do overwhelm our limited ability to get to respond to.

[LizzyFitbit]

Welcome to the forums, @KerrieD23. Hi there, @JohnnyRow.

@KerrieD23 Thanks for the details and steps tried on your own. I'm sorry for this inconvenience. As you may know, when using the Exercise app, we recommend selecting the option that best fits your activity description so your watch can track your details correctly.

While I understand you'd like to track your pace, my best advice is selecting Treadmill as this will ensure the accuracy of your data. To learn more about the Exercise app, see this help article.

[jdgRunner]

I did not realize that there is a Treadmill option.  I find that on a treadmill, running, say, 6mph (10km/hr) it get about 80% of the right distance with my Sense.  When I had a Charge 2 back before the pandemic so I ran more on the treadmill (I'm in the upper Midwest and it gets cold) the Charge 2 did a pretty good job of getting the speed/distance right.

In general, on a carefully measured linear 3-mile bike path, I will get anywhere between a slight underestimate of distance (shows 3.90 over 4.00 kilometers) to substantial underestimate (3.3 - 3.6 km verses a 'true' 4.00).  On the track, I get much better of a read and even once had an over estimate (10,190 meters verses 10,000 meters), but usually on the track it is a little bit under.  You can see it in the position trace because it gives the position every kilometer (which I use since it is a more convenient distance metric for a track and field guy).  All of the marks should be at the half-laps for odd kilometers and at the start/finish line for the even ones.  Whether it is under- or over-estimating distance it is consistent.  The track might give the GPS a cleaner signal because there is nothing getting between the watch and the signal whereas on the path there are trees and several spots going under streets. 

I would be interested in Guy commenting here, he is, I think, an ultra marathoner and posts a lot on HR and many other issues.

[t.parker]

@jdgRunner the treadmills are very tricky machines. I don't know any watch that is able to measure the distance correctly (or match the treadmill). Most of the time the distance you see on the treadmill isn't correct either. When you step on the belt then for a very short time you slow it down. It's possible to stop the belt for a fraction of a second, for example, a heavy person or one who's running with higher GCT (ground contact time). The belt changes speed but not the counter. The counter (display) isn't connected to the belt. It just runs at a fixed pace assuming the belt does the same. The belt doesn't. It looks even more interesting when you look at the data. Let's say the average GCT is 237ms (from one of my treadmill workouts, this is time spent touching the surface within the whole second, it's nearly 1/5 second and that is pretty good GCT). It means that in the 30min run (1800 seconds), one spends 426 seconds in contact with the belt. Almost a quarter of the time is affected by GCT which modifies the real speed of the belt. Count the moment of impact, then incremental release until the foot stops touching the treadmill and this is the time when the treadmill goes back to the calibrated speed. Depending on the selected speed, the runner may in fact also add to the speed of the belt by pushing it during the contact (usually when one wants to run faster but the treadmill pace is too low). There is only one type of treadmill I mostly trust and this is a non-motorized curved treadmill in which counters are linked with the belt and distance and pace depend on the actual motion of the belt. So if the treadmill can't calculate its own distance and pace accurately then how the watch could do it? The watch has no connection to the treadmill and is worn on the wrist. It's all just an estimate in that case, sometimes will come better, sometimes worse, and will be different from one machine to another, different shoes, paces, incline, cadence - anything will affect the pace and distance that the watch is showing you.

There are ways to get accurate readings from the treadmill - using footpads or a device that can be attached to the belt of any treadmill and "read" the belt (if you are interested, the device is called RUNN sensor, lately very popular due to Zwift app since they introduced virtual running).

On a treadmill, I focus on effort (HR, RPE, power) and time rather than distance and pace. This does not convert to outdoor running and the metrics should be digested differently. If you know your tempo or endurance effort outdoors then set the treadmill to match the effort rather than pace.

Let's continue with the next thing 🙂 "In general, on a carefully measured linear 3-mile bike path". Define carefully measured. For me, it means that it's measured with a suitable accurate device (for example, when I calibrated my footpod I ran with the measuring wheel, on the bike I use a speed/cadence sensor rather than GPS for distance etc.). Running on the track is probably the most accurate (and this is also how I calibrate my pod whenever I run on the track). GPS in general isn't the best way of measuring running distance. It isn't as accurate as most people think. There is a reason why a majority of running watches use a hybrid method - inertial and GPS distance measuring methods. Basically, the watch uses all available sensors. Each sensor can introduce an error but it's very unlikely that all of the will provide wrong data at any given time. Under that assumption, sensors correct each other. If the pedometer counted N steps but GPS drifted away then this information says "something is wrong with GPS". Also, using inertial methods allows for providing 3D speed and distance (including elevation). Magnetometer makes sure we know where north is, gyroscope and accelerometer provide their own specific data (and the accelerometer suffers from almost instant drift which is corrected with other sensors). This is a lot more complicated than one may think. See this:

This "knot" is me being stationary. Not moving at all. GPS tends to be more accurate the faster you move. Once you slow down or stop it will provide the quality of a drunk fly. Fitbit somehow doesn't seem to be able to solve it. With two other devices, on the same activity:

Accuracy is okay-ish and the "drunk fly effect" doesn't appear. The reason is that other sensors of two other devices realized I was not moving and cancelled out a bad GPS signal.

So the conclusion here is that you can't just rely on GPS. GPS isn't that accurate especially for runners when you want to see your pace in real-time. The method is a lot more complicated. Now, that's my educated guess. Fitbit probably tried to reinvent the wheel and do something similar to what competitors do but seems to be failing algorithmically and (maybe) due to a lack of sensors. There is no magnetometer (compass), and no gyroscope. The less sensor, the less correctional data in the system.