heart rate values totally wrong after update 18.22

I am meanwhile extremely dissatisfied. My heart rate is too low at the start of almost every run. The correct heart rate is often displayed after 5-10 minutes. In the example picture, it was necessary to stop the activity ( resume later), take off the watch and then resume the activity. Sometimes, however, even this does not help.
The VO2max calculation and load also no longer work correctly due to the error.
The Forerrunner 955 (solar) is a very expensive device and I expect that Garmin will solve this in the short term. What do I want with a sports watch if the values are not correct? I've been very happy with the heart rate in the past, but if that doesn't change, Garmin should take the watch back!
Garmin, please solve this.

Garmin-SierraWhen can we expect a solution? 

Thank you very much Garmin-Sierra.
Here I leave the activity screenshots from other threads just in case. Activities of different days with almost the same OHR behavior and the same erroneous graph and very similar to another user. Both have the change of HR close to the minute 8 of the activity.

Thanks!

Garmin FR955 Solar FW v19.18

Hi Garmin-Sierra,
I also have examples.
Usually for sports (running, biking, indoor rowing) I use external HRM (Dual or Pro) and those are fine. With walking and hiking also until now I had no big issues with the optical HR sensor.

But sometimes I use the bike to go to office and I don't wear external HRM then but still push the pedals quite hard, so my HR goes up.
And I could see this Bug too with suddenly drastical change of HR (before= too low, after=correct), with those biking activities and optical HR - see here these two examples:

Hello Garmin-Sierra, I have a 3 week's old FR955 Solar and are running the latest software. I am always using chest belt HR when running and it works fine.

But often after workouts when going back to OHR again, heart rate is crazy high (have been up to 200bpm and my max HR is 190bpm. This could be just standing still or out on a walk but often after a workout. If I turn off my watch and then turn it back on again, HR is back to normal (~80bpm).

Except from this, I'm super happy with the watch.

Hi, The strange thing is that I ran another test yesterday with only OHR and the graph was normal. There was no change between the previous ones and this one (same outfit, almost the same place, same warm up, almost the same ambient temperature, etc). Thanks!

Normally I'm running with a chest strap, but out of curiosity I also tried it and I can reproduce this behaviour.

First I thought it is because my hairy arm and I removed my hair where the watch is placed, but still the same on the bare skin.

If you need some more information feel free to reach out to me any time.

it is not a convenient way to wear the watch but, if you place upside down to the bottom of the wrist, where a doctor measure the hearth beat on the wrist, than you will se pretty correct HR along the whole run....

I tried this out today and it is not as bad to wear the watch like this. When you have a lot of starters somewhere you do not bump the watch against other people or things.

But the result in my case was not better. Too low HR the first minutes and then catching up.

Even if I know some out here are really annoyed of the bad performance of the sensor, it is ok in my point of view. If I want to have good results I use a chest strap. If estimations are engouh I use the optical sensor.

You also use a measuring tape and not a smartphone app if you want to know the size of your desk table.

Since Garmin does not tell very much about any algorithms or the sensor it is hard to assess for all external people what happenes there and which parameters are tuned at the updates.

But software will never fix what the hardware is not able to do ... 

Nevertheless as I do not know where the real challenges are or if they are already solved with new hardware revisions you have to deal anyway with different "arms", "sknis (colors)", "hair", gender, ambient conditions, ...

Maybe there is already some self learning feature in the algorithms, maybe it would be useful to add some. Or even an idea to ask the user in the connect app if he thinks the result was "good"