Garmin, It’s Time to Get Your Act Together — Your Heart Rate Algorithm Is a Mess

I've been noticing this since day one and commented on other threads in here,

i do 20 sec max sprints and out of 10 the FR970 paired (BLE) with the HRM600 misses from 1 to 2 of them which makes me wonder how accurate the heart rate sensor is keeping in mind Garmin's long tradition in using us customers as beta testers to hurry up watch releases

somtimes I have no HR data at all which makes me ask a friend if I am still alive or not :( 

I've had the Epix 2 and FR265. The Epix was great, until firmware 16 - just like you said. Once that firmware hit, the heart rate data seemed to be completely made up. It was especially bad on the C2 RowErg, but it wasn't even reliable on a normal Z2 or Z3 run anymore. I got fed up with it after a few months, sold it, and waited for a while for Garmin to hopefully sort it out. I bought the 265, and it had issues from the start, accuracy was questionable and it would take 30 to 60 seconds to respond to a rise in heart rate, which is totally unacceptable and makes it useless for anything other than gradually increasing steady state activities. 

I am planning on trying out a FR970 or Venu 4, hoping that the 5th gen heart rate sensor actually responds to increases in heart rate in a reasonable amount of time and shows a heart rate within 5-ish BPM of reality. I'm not looking for chest strap levels of accuracy here, a 5 or 10 second lag is fine with me, but upwards of a full minute is useless while trying to stay in a heart rate zone. Unfortunately, after reading posts like these, it's becoming clear that Garmin is not interested in fixing these issues. I sold my Epix 2 years ago when the issues started, and it appears that nothing has changed. 

I would love to have Garmin metrics and sport profiles again, but not at the cost of having all my metrics messed up because of inaccurate algorithms/sensors. For everyone that says "just use a chest strap": no, I won't. Other watches (Apple, Google, etc) work perfectly for my use case, but they do not provide the training metrics I want. I'm not going to add an additional peripheral because Garmin can't get it together and make accurate devices. 

I'll still try out a watch with a 5th gen sensor, and in all likelihood if posts like this one are accurate, be thankful that REI has a generous return policy. 

I completely agree. I only mentioned one critical point, but there are many more. It makes no sense to release a product or go into production if it is not ready. 

Bugs can always pop up, but when you see the changelog with a huge list of fixes, something is clearly wrong. Perhaps they should review their processes and implement a development cycle with the relevant checks and revisions...

I agree. When walking around town I can look at my watch and often find it reading ~130bpm. If I stop for a moment and click on the heart rate widget itself, it'll lower to mid 70s. Likewise, wearing it well above the wrist and snug enough.

I thought the Instinct 3 Solar had many errors and bugs because it was an 'inexpensive' watch.
But when I checked the forum for the expensive 970 watch, I saw that customers were basically doing beta testing even after paying a lot of money. What on earth is Garmin doing?

I totally agree. For running I wear a HRM600, and that is fine. But when I do a simple walk I only wear my FR970. At the start of the walks my heartrate goes berserk. For about 5 minutes it increases to a HR off 120-130, and then drops to around 90 where it stays the rest of the walk. My Apple Ultra2 shows a HR of around 90 all the time.

So there is something "off" with the HR registration

I had some wacky readings from mine also. I calmly called Garmin tech support, explained what my issued were, they looked at my data, they had me send them some diagnostic data using Garmin Express, and they reviewed. A day later they determined my watch probably had some hardware issued and they ended up replacing my watch.

Granted, I also had some random crashes but still. I have always had a great experience when dealing with the costumer support folks.

I've used Garmins since Forerunner 101. I have a Forerunner 970 and don't have any of these problems. I think you need to understand how optical HR works and try to troubleshoot it. Things like hairy arms, dark skin and watch tightness can affect accuracy. I wear my watch snug up against my arm with no gap and shave the hair on my arms under the sensor. HR data outside of an activity are not captured or stored at the same level of granularity as when you are within an activity - outside of an activity they are stored as averages over a period of time (2 minutes). So if you do some short high heartrate activity that lasts for 30 seconds, 90 seconds into the HR capture cycle, the peak can be greatly reduced by the averaging. I find using a chest strap (I use Polar H10 and Garmin HRM-Dual) will give you better results in an activity because they are measuring the actual electrical signals from your heart and not the secondary effect of your pulse. You have to make sure the sensor pads are damp before you put on the strap - otherwise it will not be accurate until you sweat enough to get them damp.

Absolutely, I'm aware of the limitations of optical sensors—things like how light affects them, not wearing it snugly enough, tattoos, or skin tone—but that’s not the issue here. I'm not complaining about the watch during exercise; I'm talking about the watch when it's 'at rest,' when, as you rightly say, it's operating at a different measurement level.

The problem isn't the sensor itself. For about a year and a half now, there was some kind of change where it detects spikes (and not just for a few seconds, but several minutes, sometimes up to 15) showing an unreal heart rate, with a ridiculous difference of 40 or 50 beats per minute more. Sure, a brief sprint or a short burst of effort might fail, that’s normal—it’s not the same as a chest strap heart rate monitor. But I mentioned a stress test where it failed to detect the effort for 10 minutes; that's quite something. In fact, a couple of weeks ago, I tried the 42mm Epix Pro on my left wrist and the FR 970 on my right, running on a treadmill for two and a half minutes without starting an activity, just to see how well it worked. To no one's surprise, it didn't go over 85 bpm; in fact, as time went on, it actually dropped to 70.

This issue I'm describing doesn't happen with Apple. It’s not necessarily that their sensor is better or worse, but their algorithm doesn't seem to flag anomalies or absurd data, etc. And this hasn't always been the case (with Garmin).

Look, I genuinely like Garmin for its watches and features, and I'd prefer to stick with it over getting an Apple Watch... but if I'm paying a premium price, I also expect it to work, and for them to address their weaknesses, which in this case are their readings and algorithms when you’re not tracking an activity.

On another note, with the Epix, I had issues with corrupted activities and reboots during intervals. The response was always the same: 'reset the watch.' Honestly, they have pretty subpar and hard-to-maintain software, and there’s not much they can do about it. I was hoping they would start from scratch with this generation and use a more mature, professional development process, because releasing a bug-ridden product is pretty bad (errors can always happen—we’re talking about software—but not to this extreme).

I'm glad you're not having problems. My wife hasn't had any issues with her Fenix 7S or Forerunner 970 either, so it probably depends a lot on the individual.

Thanks for your reply!