HMs DIY #1-6

Hi. Firstly. Thanks Mcinner for providing the links to the NP calculation. I wasn't able to check myself until this evening

My understanding is that the NP calculation is based on a rolling 30s sample. No value is calculated until 30 seconds has elapsed. Thereafter; every second a sample is taken of the previous 30 seconds average power (0-30s, 1-31s, 2-32s & so on), which is combined with the result of all the previous samples for the selected time frame. Be it lap or entire activity, with the NP being calculated using that value

I know the edge has native datafields for this information. I actually thought it might be available in the CIQ SDK rather than you having to perform the calculation within your app. Your efforts are much appreciated.

For a NP calculation the device must be set to 1 second recording not SMART, however from memory if a powermeter is connected to the device, even if the device is set to use SMART recording it will switch to using 1 second recording for the activity

As for the cadence being a SDK bug. I did think so given I'd also seen it in the other app. It did actually work at some point in the distant past but have no idea, which SDK update introduced the bug. Thanks for looking into this too.

In EDGE 820 max cadence works correctly until software 9.10, and in Edge 830 from the beginning there is this error.

The 820 update following 9.10 was a big update over a couple of betas, which was officially rolled out at f/w 10.0, much like the similar 1030 update it added CIQ 3.0 support. It's possible that this is when the bug was introduced

No, this bug was reported 9 month ago in easuvorov DIY thread by Mali164

 https://forums.garmin.com/developer/connect-iq/f/showcase/5201/diy-data-field/55133#55133

I think 820 f/w 10.0 was released around 08/18 so some 16 months ago. Not sure when the bug started happening on my 1030. One if my layouts with the original diy was becoming increasingly unstable on my edge so I had to stop using it

Hi

But you still have the open question with "leftover" seconds.
In the next release there will be the normalized power available. It will only update every 30s, so there is no problem with "leftover" seconds as I only calculate the power with 30s junks. I dont think that the value will be any different after half an hour if I would also calculate in somehow the "actual" seconds which are remainders of MOD 30. 

I will also take a look into the gear calculation for mountainbikes.

regards

Erich

I‘m afraid that @hotdogpartysausage is right to calculate the 30sec-average every second after the first 30 secs. This would mean a huge amount of value-sets at the same time.

Description says: Calculate the rolling average with a window of 30 seconds: Start at 30 seconds, calculate the average power of the previous 30 seconds and to the for every second after that.

Maybe my English is too bad and I misunderstand the term „and to the for every second after that.“ Maybe it is a typo meaning „and do this for every second after that.“

I think in that case there would be no „leftover“.

And I‘m more than impressed about the figures you have to calculate: the fourth power of 200 Watts is 1,600,000,000! 

Yes, there is always a leftover. For example how would you calculate the np for a time of 100s ? You take the first 3 30s junks and then you have a leftover of 10s!

I found this post in another thread:

The answer is that there is no such concept as normalized power per record. 
Normalized power is only defined over an interval or lap. 
Also, because it employs a 30 second moving average, NP is not even defined 
(cannot be computed) for laps shorter than 30 seconds. 
Also, while it can be computed for any lap greater than 30 seconds, 
it is not a relevant concept over intervals shorter than 15-20 mins, 
so many software programs recognize this by refusing to compute it 
for intervals shorter than some cutoff like 15 or 20 mins.

I think that my approach to calculate it every 30s with one new value seems to be the correct one, and leftovers seconds are not taken into account.

I will release a new version today, so you can test the new value (np) on the weekend (-:

regards

Erich