How many times to calibrate watch?

[TxWatch]

I ran for another 60 minutes with my phone and 60 minutes without my phone to see if the accuracy improved. As expected, the distance with the phone in an armband was within 5/100s for a seven mile run. With that small of a difference it could easily be the Garmin or the phone that is off slightly, so I am not concerned.

My second run without the phone was "slightly" better than my previous phone-less runs. This leads me to believe the Watch "may" be learning my stride length. However, if true, it is a very slow process and not something worth advertising in all of their Watch literature as a valuable feature.

Before Calibration:
1. 7 mile run: Off by a factor of 1.22
2. 7 mile run: Off by a factor of 1.26

After Resetting Calibration Data:
4. 7 mile / 60 min Calibration Run: Off by 1/10 of a mile (Garmin FR 201)
5. 7 mile run: Off by a factor of 1.22 (Same as first run before calibration)
6. 7 mile / 60 min Calibration Run: Off by 5/100 of a mile (Garmin FR 201)
7. 7 mile run: Off by a factor of 1.18 (only slightly better)

I added a BT headset for music on the last run, which I had to adjust a few times, so my watch arm may have been static for a couple of minutes in the early going. (More on that in another post)

With that said, I believe there is a problem with how they are calibrating the watch. I am not sure how some of the review website's who tested the calibration were able to get to the numbers they claimed. It is possible it was working better in Watch OS 1, but I know Watch OS 2 is far from accurate after 120 minutes and 14 miles of consistent arm movement calibration data.

Thanks,

TxWatch

 

[garrym13]

This is slightly OT, but when I run I always carry my phone and run the Nike running app and the Apple outdoor run workout simultaneously. The distance between the two are always spot on. I get why here due to GPS. When I run indoors, on a treadmill, the distance is generally off by a half mile or more using the same apps but with indoor run. Thoughts?

 

[exxxviii]

I think there are two things at play. First, I think the watch calibration generally comes up with a longer stride length than reality. And like you, it measures accurate distance when running with the phone, but it just seems to calibrate longer stride length than it should. Second, I do not think Apple thought to adjust the calibration for indoor treadmill running where people generally run with a more efficient gate (shorter stride length and faster cadence). So the net is a compounding effect that will result in much greater distance than reality.

 

[maflynn]

In my humble opinion, Apple's calibration does not work.

That's my feeling as well, for short runs i'll use the apple watch. If longer runs (in excess of 5 miles), I'll wear my FitBit Surge which has a GPS in it.

I know the milage of my routes, so knowing the exact mileage is generally not a major issue with me, which is why I don't mind the less accurate apple watch on the shorter runs. For the longer ones having accurate splits and pace however is more important.

 

[BarracksSi]

Is everyone getting a clear GPS signal?

 

[exxxviii]

Is everyone getting a clear GPS signal?

Crystal. Most posts have the same key data points. External devices or apps are accurate (my Garmin and Runkeeper on the iPhone, in my case). The watch is accurate when connected to the phone. These all suggest that GPS is not the problem. It is the watch calibration that is failing.

 

[TxWatch]

Is everyone getting a clear GPS signal?

Yes, the GPS signal on the phone is very clear. When I run with my phone using an armband, my distance for the total 7 miles is within 1/10 of my Garmin. This is a small enough of a difference that it does not concern me or factor into the calibration equation.

After another 7 mile run today with the watch only, my distance was off by a factor of 1.21. This leads me to retract my earlier statement that the Watch "might" be learning my stride length. Whatever they changed in Watch OS 2, the calibration feature is no longer working.

TxWatch

 

[BarracksSi]

Went for a run earlier tonight with my Garmin 410, the Nike+ app running on the phone, and the Watch in airplane mode to guarantee that it won't piggyback off the Phone's GPS.

The Garmin and Nike+ agreed at 5.56 miles, almost exactly, and the Watch thought I ran about 6.7 miles.

The calorie counts were pretty close among all three, at least good enough for me to not worry about. I'm also trying to dismiss the big overestimate by the Watch by telling myself that the only occasion where distance truly matters is in a race, and I'm not going to stop running short of the finish line because my watch said so.

But, dang, it's a big difference. If I plan to jog much more in the coming weeks, I'll reset the calibration data and try it with the phone some more.

 

[amherstbias]

Mine is always off but a lot. I really just want after the run is complete to force touch and be able to enter the correct distance. Is that too hard!

 

[Julien]

Went for a run earlier tonight with my Garmin 410, the Nike+ app running on the phone, and the Watch in airplane mode to guarantee that it won't piggyback off the Phone's GPS.....

Why don't you want the Watch to have access to GPS calibration while you run?

 

[BarracksSi]

Why don't you want the Watch to have access to GPS calibration while you run?

For just this run, it was part of my experiment to see how well the Watch was matching step size with true distance.

 

[TxWatch]

The calorie counts were pretty close among all three, at least good enough for me to not worry about.

But, dang, it's a big difference. If I plan to jog much more in the coming weeks, I'll reset the calibration data and try it with the phone some more.

After upgrading my Watch to OS 2.1, I ran a couple of times without calibrating to see how close it was to OS 2.0.1. Unfortunately, I found it the new OS version is even further off for my stride length. After two initial runs, I calibrated the Watch using the method outlined on numerous web sites.

To calibrate, I reset the calibration data and carried my iPhone in an Armband using the Outdoor Run for two 1 hour runs which were 7 miles and 11,000+ steps each. With that in mind, I feel 120 minutes, 14 miles and 22,000 steps should be plenty of data to calibrate the Watch to an approximate stride length for any person. Since my Watch calculated distance was almost the same before and after my "calibration runs", I do not think they are doing anything to calculate or store a persons stride length in OS 2. (Or, their method is so flawed, it would take me weeks of running outdoors with my phone to get close enough to be accurate)

It is possible calibration was working in OS 1, but I did not own a Watch at the time, so I cannot comment.

Watch OS 2.0.1:
Run distance: 7 miles (Garmin FR 201 and iPhone GPS)
Watch by itself: 8.75 miles - Off by a factor of 1.25 (average of multiple runs)

Watch OS 2.1:
Run distance: 7 miles (Garmin FR 201 and iPhone GPS)
Watch by itself: 9.10 miles - Off by a factor of 1.30 (average of multiple runs)

Since your numbers were off by a factor of 1.20, you must be taller than me and closer to the average stride length they are using for their calculations. (I am 5' 5" which is shorter than the "average" person)

I also noticed the active and total calorie count was very different when I run with my phone vs running without my phone. In other words, if my Watch thought I was running further in the same amount of time, it was recording a higher active calorie count by almost the same factor. (~1.25 and 1.30) For now, I would not trust the calorie count, either.

TxWatch

 

[Rhinoevans]

Until the watch includes GPS for Activity tracking its going to be best estimate. My first Garmin cycling computer was basic with no chest strap for heart rate and no wheel sensor to verify speed or show cadence. It took its best guess but was wrong on my activity level but good with distance. Current computer has all needed sensors and does a nice job.

My watch has the same problem at the moment. Its calibration is close but not exactly right. One thing it has been though is consistent. Right/wrong it has been showing the same basic results each run so I have been able to see my progress and track changes in my ability to cover my route in more or less time. I feel like the distance is off compared to what my phone with GPS would say and has said but in the end I don't need exact results. I would like very much to have GPS included but since its not I enjoy other features. I like that it tracks movement, has the ability to store music and play to my bluetooth headphones and gives me a way to adjust pace. If I wanted more accuracy I would go with a dedicated device like a Garmin but for a device that does many other things too its fine for me that its not perfect. Wish it was and its not off the hook for being imperfect because I know they can do better but I knew its limits before I decided to keep it and use it.

Well said. I totally agree, mine is off compared to my Surge, but it is consistently off. I really enjoy all the other features and can deal with a slightly off distance.

 

[TxWatch]

Well said. I totally agree, mine is off compared to my Surge, but it is consistently off. I really enjoy all the other features and can deal with a slightly off distance.

I agree that my overall experience with the Watch is positive and I plan to keep it. Like you, I will deal with the distance differences until they get around to fixing it in this version or the next.

My watch has the same problem at the moment. Its calibration is close but not exactly right. One thing it has been though is consistent. Right/wrong it has been showing the same basic results each run so I have been able to see my progress and track changes in my ability to cover my route in more or less time.

You are also correct. Even though it is off by a large margin for me, it is always off by the same consistent amount. I guess I will have to ignore the Engineer in me saying "this ain't right" and just use my route times to measure my progress.

TxWatch