Related
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Incorrect average lap power

sirfive
over 3 years ago

Dear all,

I just noticed that Garmin calculates the average lap numbers (including power) and saves the averages in the fit file.

However, the average power numbers are incorrect and this is not because of rounding.

In my workout yesterday I did 30/15, with the 30 seconds around 315 watts. For one lap Garmin shows an average lap of 301 watts, and TrainingPeaks also shows this number (at first).

However, when looking at the 1-second recordings, the average of the 30 recorded values is 313 watts !

When I change a 1-second recording in Trainingpeaks, change it back to the original value and 'apply changes', the averages (of ALL laps) are recalculated and they are correct again (in Trainingpeaks).

So I had a look in GarminConnect and indeed, the averages (here called intervals) show the wrong average power numbers.

Loading the fit file into GoldenCheetah gives me the right averages (GC calculates the averages itself).

Conclusions:

1) average power per lap calculated by Garmin and saved in the fit file by Garmin are wrong. There is no way to force Garmin Connect to recalculate.

2) TrainingPeaks 'believes' the averages in the Garmin fit file and shows them. However TP can be forced to recalculate these values.

3) Golden Cheetah calculates the averages itself.

Now this was obvious and easy to check for a short 30 second interval. I will have to take an interval of a totally different workout, much longer, will do that tonight.

Anyone noticed this before?

Probably this is something that can easily be repaired by Garmin ?

Top Replies

All Replies

  • 0 over 3 years ago

    Would you mind sharing the FIT activity file?

  • 0 over 3 years ago in reply to sirfive

    The lap values are actually correct in the FIT file. It would appear that lap record is not being processed correctly by some of the tools.

    The power record immediately  before the end of the lap marker should not be included in that lap if it has the same timestamp as the lap record. The lap start time indicates when it starts. Any record with the same timestamp should be included in that lap and not the previous one.

    So when you look at a block of data it is not the record between the lap markers, but the records whose time stamp falls within the lap start/end time. This is typically one record before the lap marker start and excludes the record before the lap marker end.

    Taking your first 30 second lap as example

    The lap starts at 

    985447114

     So it includes the record before the lap record

    The next lap start at 

    985448044

    So the last record before that lap marker goes into that lap and not the previous lap.

  • 0 over 3 years ago in reply to aweatherall

    Yes, the first few intervals do match, indeed.

    Here an overview of the differences per lap in the first part of my session:

  • 0 over 3 years ago in reply to sirfive

    The other thing to know is that the lap and session power averages are computed using the accumulated power data and not the individual power records.

    When a power meter sends the power information for the period it is reporting on it also sends a running total. This way if some records are missed the head unit can track the total work and decide how it wants to handle the missing records.

    For short gaps in the data the Edge will use the accumulated power data to back fill the missing records by averaging the gain in accumulated power over the missing period, instead of dumping it all in one record.

  • 0 over 3 years ago in reply to aweatherall

    Let me take a look at lap 6 as that appears to be one of the biggest difference. I will also check your data reception rate. 

  • 0 over 3 years ago in reply to aweatherall

    The data reception rate from your power meter is quite poor. Anything below 90% will start to impact the quality of the data.

    What power meter are you using?

    The issue could be due to other 2.4 GHz radio traffic near by. If you are using WiFi near by switch that to 5 GHz if you can. If you have to use 2.4 GHz select a channel that is further away from the frequency used by ANT.

    Your RX rate is 87.5%. This results in 20% of the power records having to be back filled due to drops in the data.

  • 0 over 3 years ago in reply to aweatherall

    I see it is a Power2Max unit. Is software build 0.13 the latest for this model?

  • 0 over 3 years ago in reply to aweatherall

    I see it is a Power2Max unit. Is software build 0.13 the latest for this model?

  • 0 over 3 years ago in reply to aweatherall

    So this gets a little complicated, but the difference is due to a couple of things going on. One is that there are radio outages that are causing some of the records to be back filled. The second is that the tools you are using to process the data are using the power values assigned to each record and are not using the accumulated power values. Using the accumulated power values is the better approach. Because of this the tools are subject to how the dropped messages are being backed filled.

    The accumulated power record at the start of that #6 30 second window is 199200. The end value is 208219. That is a difference of 9019 kj. Average that over 30 seconds gives 300.63 or 301 watts. This is what the Edge reports.

    The last 2 seconds on that lap and the 1st second of the next lap have the same accumulated power value because of an outage. 

    Using the 1 second power values that have been computed using the backfilled data because of the radio drops gives 9330 kj over 30 seconds or 311 as you get.

    For the next 15 second lap the accumulated power starts and ends with 208219, 210262 or 2042 kj over 15 seconds 136.2 watts.

    The sum of the individual power records gives 1720 over 15 seconds or 115.