How is Sense 2 heart rate accuracy while exercising?

[Rami.Khalawi]

Hi all,

To the owners of the new Fitbit sense 2, how is the heart rate accuracy mainly while exercising? Is there any improvement comparing to the previous model?

Thanks

Moderator Edit: Clarified subject

[mike-MKvip]

Hi,

I did 2 quick tests cycling, one 10km and one 25 km and in both instances the HR lots of times was way too high compared to my Polar H10, peaking at 10~15 beats plus on the first ride and up to 20 beats plus on the second. Fast rises (e.g. during sprints) lag siginificantly behind - but that is par for the course with optical HR sensors, they all have that problem.

cheers

[JerryHam]

I was hoping that it would do better. I had an original Sense, and what I would find is that it would start out way too high (for example showing 150 when my iFit band was showing a correct 80) and then after several minutes would usually "come to its senses" and exactly match the iFit band. Other times it would go "nuts" again later in the workout and be way too high again. I also found that at walking speeds or "very slow jogging" speeds the Sense 1 would be way too high. It would do fine with resting heartrate. 

I was hoping to see all this crap fixed with Sense 2 - that maybe it could sense the HR correctly. Nope. Same exact issues. Here's an example from very early in a workout on the treadmill. In this, the iFit band is paired to the treadmill and gives the correct HR and you can see that the Sense 2 is way off.

Later in the workout, it did end up correct but for the first about 5 minutes it was not even close to correct. Again exactly the same behavior as Sense 1.

This inaccuracy coupled with the inaccuracy when at a walking pace makes the calories burned be useless on both watches.

I actually took the graph of HR from my iFit band and the graph from FitBit and corrected their relative scale and they look like this:

In another workout, it went "nuts" again later. Definitely NOT to be trusted at all.

[t.parker]

I already posted some analysis for running with various intensity and walking. Today I ran half-marathon race (didn't get spot in London Marathon for this year, hopefully will get for next one :D):

Half Marathon race

This time GPS did quite good job, although the distance is underestimated. The green is the reference route, yellow is route tracked by Sense 2. Pretty good, no issues with connection.

The HR isn't perfect. It isn't even good but probably it's just ok. I've seen similar behaviour on original Sense and even original Sense sometimes registered HR almost perfectly aligned with PolarH10.

There are other analysis in another thread. I don't think there is a point to repost it as it will be probably gone same as similar analysis I've done for Sense in the past.

Yesterday's Parkrun 5k (almost 5k, alternative route) here .

Evening walk here .

Easy steady run here .

In my eyes, nothing changed. We received original Sense heart rate tracking in the new case.

[t.parker]

Here's another one. Today's 8km recovery run. I was keeping HR quite low (low running power). For this one, I have no map as GPS has been disabled due to power consumption testing (and the distance with no GPS was overestimated by 700m).

Orange - Fitbit Sense 2, Blue - PolarH10

Somewhere near the end of run the HR starts matching but until 25 minutes it's bad. It's a steady run, with no sudden changes in HR, flat paved surface and fixed cadence (190spm), not much external noise that could disturb readings:

This is the case when I'd expect the HR sensor to work flawlessly. It doesn't. What sits in Sense 2 is PurePulse2.0. Same issues, same behaviour, same inaccuracies.

[JerryHam]

What I see is that, at slow speeds - like a power walk or a slow jog - it is completely useless. At running speeds, it starts off whacky (way too high) but settles in nicely after that. Today I was really feeling my plantar fasciitis and almost didn't run at all. But sure enough, iFit phoned in the monthly "running" challenge with a slow jog instead of a run. And, feeling like I did, I did not override it and do a run. This supposed run started with a warmup walk, then a slow jog, then a cooldown walk. Perfect for showing how FitBit Sense 2 does in these scenarios! (Again, after a few minutes of craziness, it does a LOT better at running speeds as you can see in my first reply upthread).

Here, the numbers are so far off as to be meaningless. The average HR from Sense 2 is only one beat per minute below the MAX from the band.

[jfb9301]

Did 5K on my concept2 rower... About halfway though, I dropped out of cardio and peak to 70BPM.... shot back up as I sprinted to the finish.

I never changed pace all that much from 2:12 splits (500M) so I'd say this heart rate was wrong.

Cooling down, I noted that I had not tightened my band before starting, or moved it up above my wrist. The HR monitors FB uses are fiddily and can benefit from being above the normal wrist, and tighter than normal. It's been like that since my first Charge.

Check out https://help.fitbit.com/articles/en_US/Help_article/1565.htm#:~:text=For%20a%20more%20accurate%20hea....

I normally wear the tracker about 1/2 finger up, and it does OK, but 1 to 2 fingers up for best performance.

Try this, it might even help. It certainly couldn't hurt.

[JerryHam]

I've tried that with both the original Sense and the Sense 2. It doesn't really make much difference as long as the watch is fairly tight, it works about the same. Today I did one 6k run in zone 2 and zone 3 (treadmill) and it was nearly perfect. Then, without changing anything, I did a power walk (still on the same treadmill) and the Sense was 47% off (HR was 81 and Sense 2 said 119). As I said before - it does not do well at all on walks or slow jogs. Anything fast jog and higher and it does pretty darn well.

Again on the actual run it was pretty much right on the whole time. Just on the power walk that it did this.

[t.parker]

@JerryHam with my experience from Sense, I wouldn't try to find any rule or pattern. Sense from time to time could be perfect on runs with various intensities and I believe Sense 2 can be, too. Yet, I still have to have such a run for which Sense 2 gives me accurate HR. I have a recovery week now so doing only easy runs (but more than a jog) with a pace between 5:00/km to 5:30/km and a fixed cadence of 190spm. This is enough to elevate my HR into an easy/low-aerobic zone and keep it there. Today 8km run:

Orange - Sense 2, Blue - Fenix 7 + PolarH10

There is a dip in the blue line near the 9th and 10th minutes. I've been waiting a while to cross the road. Sense 2 on the other hand shows an increase in HR. This is the same story as the original Sense (and I bet this is the same sensor and software). I'm also quite sure that once in a while I will have a run that Sense 2 will nail matching the PolarH10 data (like Sense did several times).

Sense 2 - avgHR: 145bpm, Max: 171bpm (that never happened)

PolarH10 - avgHR: 143bpm, Max: 157bpm

The average HR comes pretty close so such errors shouldn't matter for people who are mostly interested in the calorie count. On the other hand, this is useless for those who want to exercise within a particular range.

[Dopovo]

I'm shocked about these terrible results.

When Sense was initially released many users and notably reliable Youtubers and professional reviewers such DC Rainmaker and The Quantified Scientist immediately raised concerns about the unreliable HR sensor.

Fitbit ignored the problem but then was forced to change opinion "promising" a firmware update to solve it, update that never came. Due to flow of discussions in the forum about the same issue Fitbit then changed approach and went into denial mode even deleting factual posts of comparison between Sense (or Versa 3) and other trackers (including the very reliable Polar H10).

The attitude shown by Fitbit was despicable and made everyone thinking that the issue was the hardware sensor itself so impossible to solve it. Fitbit should have retire the smartwatch from the market as HR is the core function of its health features.

And then Sense 2 came. People were asking just one thing: keep the watch as it is and fix the issues, notable the HR sensor.

Now it's clear that Fitbit decided to strip down the watch from all its smart components (we can discuss about it in a separate topic) to concentrate to the health aspect. Ok but..... the HR sensor is a bad or even worse than the previous one. Fitbit what are you doing????

On another note have you noticed that no reliable Youtube reviewers have so far posted a comprehensive review of the new Sense 2? Maybe they are struggling with the HR sensor as the community do and asked Fitbit for some clarification or, possibly, they don't care any more about Fitbit.

[t.parker]

@Dopovo I wouldn't say it's worse than Sense. I honestly believe it's exactly the same. I'm planning to finally do some cycling tomorrow to see how it works on the bike (as I mentioned in the past, I have never had correct readings using Sense while cycling outdoors) and visit the gym in the evening. I have had one more recovery run, 8km at a low steady pace (base run) this lunchtime:

Orange - Fitbit Sense 2, Blue - Garmin Fenix 7 + PolarH10

There is clearly a bit when suddenly HR aligns with the chest strap after the 27th minute of the run (despite two strange spikes) and that's the accuracy I would count for. Yet, the remaining part of the run is just a total mess. And again Fitbit gets away with horrible HR when it comes to calculating an average value which is 141bpm while the PolarH10 indicates 140bpm (that's the problem with averages, inaccurate data may lead to the correct average). Knowing the story of the original Sense, I don't count that HR inaccuracy will ever be fixed.

[t.parker]

No surprise here. I didn't have much time to cycle but went for a short ride just to pick up my sunglasses. Mere 2km ride.

Orange - Sense 2, Blue - PolarH10

As I expected, the Bike exercise would deliver an utter HR mess. So far I see so many similarities to the original Sense that tells me we are dealing with the very same sensor. If this hasn't been fixed for the original Sense then I don't have much hope that it would get fixed for Sense 2. The PolarH10 aligns perfectly with power and cadence:

It's easy to see that the HR goes up when the power is generated from pedalling (and there is a non-zero cadence) and HR goes down when the bicycle is coasting (no cadence, no power). This cannot be figured out from Sense 2 HR readings (I didn't bother to process data to overlay it as it's pretty much visible with the naked eye).

[t.parker]

I tested Sense 2 during Indoor Rowing (Rowing Machine on Sense 2) and Weights:

Orange - Fitbit Sense 2, Blue - PolarH10

Both tests show very inaccurate HR on Sense 2. In the Weights, when looking at the blue graph, it's possible to count sets (each large spike). This isn't the case in readings taken on Sense 2.

[JerryHam]

FitBit should probably put "heart rate monitoring is for entertainment purposes only" on the box except that would probably call into question whether their aFib detection stuff is legit or not...

[jfb9301]

I don't know how to post the graphs like t.parker but....

I got a Polar Verity to monitor HR on my concept2 rower this week. Yes I know it uses an optical sensor like the Sense 2. My first workout my graph comparison looked a lot like t.parker's. it took a bit longer for the HR on the sense to come up to what the Polar monitor was logging.

I got to thinking about all the tips that Fitbit puts out and other things known to mess with optical HR.

Proper placement - 2 fingers up

No tattoos - nope, don't have any that would interfere

Skin color - not that anyone can change that it is what it is, but I'm Caucasian, so that's unlikely to be causing issues

Body hair- I'm not hairy, but could it be? I got out an electric clipper and cut what little I had. Surprise, it worked. I'm not sure it is as accurate as the polar monitor, but it's a LOT more responsive. Maybe we can test further.

[Bearbrook]

I'm seeing lots of big changes when running. Goes up when I slow down and walk.  

I had the orginnal sense and in the first few weeks or months early 2021 it wastage same.  Maybe an update could help at least make me feel like it's not jumping around so much. 

Also it's dropping the heart rate a lot.  A small move seems to knock it off the tracking. 

[sherriblalack]

I've found that when I exercise, the Sense 2 isn't able to track my heart rate at all.  Most of the time when I look at the Sense 2, it shows blank for the heart rate.  I had similar problems with the original Sense but they seemed to get better with updates.  This is very disappointing for a next-gen product.

[JerryHam]

Yes, it is very disappointing for sure. I've also seen it just show blank sometimes, although more frequently it just shows a completely bogus number. For example in this image, the 88 in the background is my actual HR during a walk (from the iFit Band connected to the treadmill), while the 157 is shown on the SenseLess 2.

It makes the whole "Zone Minutes" thing meaningless since this shows peak zone (which gives 2 zone minutes per minute of time) when I am actually at "below zones" and should not be getting any zone minutes. And of course your Zone Minutes is what drives Premium's "readiness score" - which is of course just bunk since it is being fed bad data.

[POWRSURG]

I feel like I'm the outlier here. I have found it to be pretty accurate when I go on runs (questionable on walks). It's when I've done weightlifting (or weights -- IDK the difference between the two) that I am somehow too low. I have sweat dripping down my sides and yet my heart rate was below the zone and I got no zone minutes.

[jfb9301]

Cardio training is reasonably accurate, after the first minute or two.

The more dynamic the exercise the worse it gets. As proven by t.parker's graphs.

For HIIT or weights, the accuracy is best described as a dumpster fire