Heart Rate Extremely Inaccurate, Microsoft Says Band is Entry-Level Device

[realwarder]

I have the smallest one which goes reasonably tight and have so say it's not great at monitoring HR. Before exercise I tighten it. Wearing face down. I'll probably try face up and see if that makes any difference.. but I've noticed that the heart icon often goes non-solid.

I did at one point wonder if sunlight or less sunlight made any difference too - e.g. when my sweater sleeve it on it or off it. Things to try.

Still, I love the little thing. Sleep info it interesting and notifications alone make it great to have.

 

[DroidUser42]

I'm hoping we'll see more comparison graphs before then.

I suspect the only graph that matters is yours. Either it works for you or it doesn't. All that other graphs tell you is if your misery has company and if it's a big enough problem for MS to fix it.

I hope to do a comparison in the next few hours.

 

[DroidUser42]

Ok, here's my result. I was wearing the band the way I normally do (display out, on wrist bone). This places the sensor right on a visible blood vein. I put it in "run" mode with the GPS off. My interest is in the readings you get as a "workout coach", not the graph you get later. So what I did was walk around the mall with my Band and Polar for 25 minutes taking a photo about every 30 seconds. For the last 5 minutes I just sat. Note that that while resting, it tracks very nicely. I also have data as to if the Band HR was locked or not, but I'm not good at figuring out how to graph that in a way that's understandable. Let's just say it was unlocked for 21 of the samples. And not all of the wild differences was when it was unlocked. That big slide from 143 down to 106 was locked. (But was unlocked on either side of it.) So simply ignoring unlocked readings doesn't work.

I did include the graph I got in MS Health. Notice that it's smoothed out. The graph is probably adequate for recording the results, but the display readings have some fairly wild swings.

 

[Upstate Dunadan]

Dupe post. Deleted.

Posted via the Windows Central App for Android

 

[Parasky]

Compared to my Garmin again on a 6 mile run this morning. I noticed about a mile in that the HR on the Band was consistently about 10 to 15 beats too high. This is the first time that has happened. After a mile of this, I stopped and readjusted the Band to fit lower on my arm, i.e. closer to my wrist bone and right below it. After that, the HR was the same as the Garmin for the rest of the run. So, at least in my case, it does make a difference on where the Band sits on your arm.

On a side note, I would love to be able to just wear the Band on my runs. It would be really nice to not have to wear a heart rate strap. However, even if the HR on the Band is perfect, I don't think I would solely wear it for most runs because it doesn't give average pace for the run, ect...Also, the display is hard to read in direct sunlight.

 

[ChumsFuture]

I mentioned my theory a few pages ago. The display is lagging behind and cant keep up when there's HR changes. I stair climb and saw the display lag anywhere between 10-30 seconds. The info it sends to the app seems to he correct though. This is a non issue that a future update should fix.

 

[jwpear]

Maybe we should compare lot numbers as well?

Maybe so. It is perplexing that some folks are seeing good results and others are not.

 

[jwpear]

I mentioned my theory a few pages ago. The display is lagging behind and cant keep up when there's HR changes. I stair climb and saw the display lag anywhere between 10-30 seconds. The info it sends to the app seems to he correct though. This is a non issue that a future update should fix.

Interesting. Possible.

 

[DroidUser42]

I mentioned my theory a few pages ago. The display is lagging behind and cant keep up when there's HR changes. I stair climb and saw the display lag anywhere between 10-30 seconds. The info it sends to the app seems to he correct though. This is a non issue that a future update should fix.

It looks to me that the info sent to the app is smoothed out.

It's been claimed that the HR updates once a second in run/exercise mode, but that would be a good trick when your talking about something that's running less than 3Hz. You'd have to take period measurements and calculate the BPM from that. But any error in detecting the same "point" in the cycle would cause a fairly large error in the reading. (An error that would cancel itself out when averaging a few readings.)

I can't help but think that arm motion complicates the blood flow and may lead to errors in the instantaneous reading. It may need to take longer samples and not update quite so quickly. Fortunately, that would be a software fix.

My biggest fear at this point is the fix requires a hardware change.

 

[realwarder]

My biggest fear at this point is the fix requires a hardware change.

I was a curious yesterday at how sensitive the HR sensor was and so unstrapped the watch and moved it around my arm from wrist to elbow.

Providing I held it against the skin for a few seconds amazingly it got a lock just about anywhere, so I think the hardware sensor is pretty sensitive, which likely means it's more firmware issues that can be adjusted. There's likely some DSP code looking at the raw sensor readings doing some pattern matching they can adjust.

 

[DroidUser42]

Another theory worth playing with is perhaps we just have it too tight. With too much pressure, the blood may not flow properly. Generally we want to make sure it's snug so it's making good contact, but that might just be the wrong thing to do.

 

[Parasky]

From what I've read about other optical heart rate sensors, the biggest enemy to accuracy is outside light. Any light will mess with the light from the sensor. I think that's why wearing the band tighter might help, it prevents any light from coming in.

 

[DroidUser42]

From what I've read about other optical heart rate sensors, the biggest enemy to accuracy is outside light. Any light will mess with the light from the sensor. I think that's why wearing the band tighter might help, it prevents any light from coming in.

True. But I ran a test. I loosened my band and pushed on the display just enough to create a slight gap. If the gap is too big it will say "where did you go?" (I think the electrical contact tips it off.) But if I have just a narrow gap - just enough to see the light, but still have one side of the electrical contact touch the skin - it works. It locks.

Note that if you take your finger and press in on the skin for for a bit, when you pick it up, you'll see that spot is pale - the blood has been pressed out. Now that may be extreme, but too much pressure is a problem too. It hinders blood flow and makes for a weak signal.

So for those having problems, this may be worth trying - loosen the band until it's just tight enough to touch the skin in all conditions. If you wear it with display inside, that's going to be so there's no gap when you rest your hand (with normal pressure) on the table. For the folks who wear it display out, it may be even looser than that.

I'm going to give it a try and see what happens.

As for the sampling frequently - I was able to catch it on the "bobble" going between 79 and 80 at about one a second. Maybe even faster. So I still have concerns about it's fast sampling.

 

[Stodds1967]

I used a simple Polar Heart Rate Strap. I can't emphatically say the strap and Mio Alpha were identical, but I can say they were close enough that I ditched the strap and never brought it back.

 

[Stodds1967]

I went for a 6.2 mile run on Saturday. This time I wore my Mio Alpha on my right wrist and my Microsoft Band on my left wrist. I couldn't take pictures comparing the 2 but I can emphatically say that the readings were never more than a few beats different. I was "on the fence" about the band before this run, but after the run I'm "sold out". There's nothing like this on the market for this price, and it does an exceptional job!

I will also emphatically say...getting an accurate and trustworthy heart rate reading has always been tricky. There are times with my Mio Alpha when it just doesn't read, or it is way off. I've found that adjusting the tightness or location of the Mio Alpha always brings it back to reality. I suspect many of the issues people are experiencing on this forum with the band HR accuracy are related to where the band is being worn (I'd recommend wearing a little higher on the wrist and very snug during a run).

 

[DroidUser42]

I suspect many of the issues people are experiencing on this forum with the band HR accuracy are related to where the band is being worn (I'd recommend wearing a little higher on the wrist and very snug during a run).

I hope that's it. The trick is finding the right spot. When you say "higher", do you mean toward the hand or toward the elbow? If it's toward the elbow, I'll give it a try, but the Band has a rather limited adjustment range.

I tried my mall-walk experiment again with a looser fit. I can't say it improved. I also tried positioning it display in. I still have wild readings. I see two factors - movement seems to confuse it. It seems to settle in if I stop. It also seems to jump around a lot. For example:

In 4 secs it went from 108 to 121
In 3 secs it went from 106 to 114
In 2 secs it went from 121 to 111
In 5 secs it went from 140 to 124

Keep in mind, I'm not running sprints. Just a brisk walk, with pause to take photos. This is just what I was able to document with photos. I may need to try video next time, as I think it's even quicker than that.

 

[agentmikeyd99]

Just return it and get a Polar if u want accurate hr monitoring. Band does other phone stuff well, not hr

 

[DroidUser42]

I'd recommend wearing a little higher on the wrist and very snug during a run.

I didn't have my Polar for compairson, and I still got some 10 BPM swings. But my impression is that it's settled down. I expanded the band as far as I could and pushed it as close to the elbow as I could - didn't do more than put it next to my watch tan. I'll have to run more tests. If that's to be the Band's new home, I'm going to have to trade it in for a large.

 

[DroidUser42]

I think I'm on to something. In this run I expanded the Band to max and "tilted" it so that the green light didn't fall on my vein. There's a few "off" by as much as 8. But it certainly looks a lot better. I think I can accomplish the same thing by switching to the large band. The size difference places the sensor in a little different spot when tightened to the same size.

The "start up" error is a little troubling. I've seen that a few times. But it just may be the band needs a bit to "come up to speed". Some of it can probably be attributed to the normal mode that only comes on 1 minute out of every 10. But it still continues a bit after starting in run mode.

 

[jwpear]

I had better results with my Band during my last workout. I wanted to test my hypothesis of hydration impacting Band HR accuracy. I follow low carb, high fat nutrition. I have to work to stay hydrated since the body does not retain fluid like it does on the standard American diet. The day before, I made sure to consume more pure water. I had roughly 90 ounces, which still lower than what nutritionist Maria Emmerich (http://mariabodymindhealth.com) recommends for someone of my height (6'2") and weight (300) following LCHF. But it was more pure water than what I would normally drink. The only other drink I had that day was 12-16 ounces of unsweet tea during lunch.

I wore the Band and my Cardiosport watch on the same wrist. The Cardiosport was, of course, linked to the chest strap. Wore Band between bone and hand--my arm is too big to wear it on the other side of the bone--toward elbow. I've tried there, but it is too tight. I watched both while alternating between steady cadence and sprints on the bike and also while doing heavy weight lifting. The strap reacts very quickly to changes in HR. The Band has some delay--5-10 seconds--in reacting to sudden changes. But the Band tracked more closely with the strap this time and the delay seemed mostly in the 5 second range. Often, by the time the slow Lumia Camera app on my Lumia Icon had started up, the Band had nearly caught up with the strap.

Here are the pics:

It's pretty close after a period of steady warm up on the bike. During this time, I maintain a steady cadence and slowing add resistance.


After first sprint on the bike. During the sprint, I peddle as fast as I can for a full minute with a full turn of resistance knob before starting. You can see it is taking it a little time to catch up. It probably took me 5-8 seconds to pull the phone out of my pocket and start the Lumia camera app.


At a roughly a minute after the last sprint, the Band is tracking pretty closely to the strap.


Doing pretty well during recovery period.


A little off for a period during recovery for some reason, but this is still within an acceptable range for me.


After 2nd bike sprint, it's trying to catch up.


A minute and a half into a period of recovery, it's looking good.


After 3rd bike sprint, it's trying to catch up again.


And then at a little over a minute in recovery, it is spot on with the strap.


After completing a period of steady cadence on the bike, it's off a little.


I'm done with the bike at this point and moving on to weights.

I've set up the bench press to begin alternating between it and curls. Strap is showing a significant drop in my HR since the bike. Band still says my HR is up.


A few seconds later and the Band catches up.


After a few reps of bench and curls it looks pretty good. I think this was one of those occasions where it took me 8-10 seconds to get the phone out, start the app, and try to focus on my wrist. I definitely noticed a lag in the Band catching up with the strap. But it did catch up instead of staying way off like it has in the past.


Another checkpoint and it looks good. I don't remember what I was doing prior to this. I move through quite a bit of upper and lower body weights during my workout. All of them are heavy weights.


Wanted to post one last pic, but the forum has a 15 pic limit. The last one showed the Band my HR 20 bpm above the strap. This was 30 or so seconds after a heavy set of weights.

So overall, it did a better job of tracking my HR compared to the strap during this workout. I've had workouts where it was very slow to catch up and very slow to show recovery. While it wasn't as quick to show a change as my strap, it tracked more closely this time and the way off readings were fewer and farther between. I couldn't quite catch the delay in the Band catching up with my camera app as it would be close by the time I pulled the phone out, started the app, and snapped the pic. I'm estimating this was 8-10 seconds. If it worked this way all of the time, I'd by happy and feel comfortable leaving my strap at home.

I'll experiment more with the effects of hydration on it's accuracy and report if I continue to see the same trend or have more days where it is way off, despite the hydration.