970 WHR accuracy - I'm impressed

can you please give an update after several sessions? 

garmin's WHRs are usually very accurate (and I rarely experience outliners such as the diff_965 here.

however, sometimes it is very wrong, not for seconds, but for long minutes it provides extremely off values.

let's see how it works on a long term and get back!

Great test! Thanks.
I also have a number of comparisons though the years, wrist and chest measures. Also swimming and cycling. More will come with 970. 

Do you have the graph showing actual values? This delta does not really show how the units are behaving... so just BPM per unit.

I don't really understand your baseline (the truth) here. It should be from a H10 connected to something?
Series:
- One 970 wrist
- One 965 with H10

Both the series show deviations.

I know these long off-periods and did not experience them with the 970 yet. In the plot, the 965 also shows some longer "drops" where the HR is off by more than 10-20 BPM for almost a minute or so.

As I wrote in my post, the baseline is the H10, from which the 965 recorded in parallel to the built-in WHR. So there are 3 data sets from which the plot was derived: 965 WHR, 970 WHR, H10 HR recorded with 965. The graphs show the differences of each WHR recording to the H10, it's just a simple subtraction of each WHR from the H10 rate.

The reason I did not show the original HR plots is that you can't see any real difference between H10 and the 970 WHR except for a few moments, hence I made the plot comparing only the deviations of the WHRs from the H10. Whether or not the H10 is correct is another question.

1339905 said:

I have to say: WELL DONE, GARMIN!

I'll be glad to be wrong, but let's wait for updates. The main problem with Garmin is firmware updates, after which problems occur.

TBH this is my biggest fear, too.

Thanks! Hence, as mentioned, I have made these comparisons through the years I am really interested.

Can you please share (plot?) the three data series. Just HR.

1. Reference, H10. Connected to a 965?

2. 965 wrist?

3. 970 wrist.

These units can only store one data series for HR. So you used two 965 I guess?

Okay, interesting! So some programming then.