PM5 ANT+ FE-C for the RowErg

[HornetMaX]

When I select "Download original" in the Garmin website for a C2-upload, I get a FIT-file, which is a format that also available for download on the Concept 2 website. The upload specifications for Garmin indicate they prefer the FIT-file. So let's assume Concept 2 has to play by the same rules like everybody else and uploads workouts in FIT-format like the rest of us.

FIT-files must contain a timestamp for every message (i.e. datapoint containing a distance, velocity, etc.) it contains. When I download the fit-file from C2, I see a message for each recovery-end. So, there they are stroke based. But that is all there is.

OK but I don't see the point you're trying to make.

The OP logs on his watch and thinks the fact the PM5 doesn't send out anything during rest intervals is screwing things up on Garmin side, because his watch thinks he's putting out power even during rest intervals.

The only thing I'm saying is that without logging on the watch and simply allowing activity forwarding from c2 logbook to Garmin Connect I'm not sure I see the same problem: Garmin Connect has no data point during a rest interval and it knows it's a rest interval. So yeah, to me GC has all the info needed to compute whichever average or metric correctly. Do they do it ? Not sure.

Example, this workout https://log.concept2.com/profile/1528069/log/84289023

It is: 6'+9'+4rest + 6'+9'+4rest + 6'+9'+4rest + 6'+9' (no 4' rest on the last interval). So 60min work, 12min rest, 72min total.

C2 says my avg power is 181W and if I re-compute the average from each non-rest interval data (interval avg pwr and duration), I get the same number: (207*6+165*9+205*6+165*9+206*6+166*9+205*6+167*9) / 60 = 182
As you can see, I essentially stay at ~206W for 6min, then at ~166W for 9min (all repeated 4 times, with 4min rest in between).
So the average is simply (206*6 + 166*9)/15 = 182W. Cool.

On Garmin Connect the power graph looks OK (right values during each interval, no date during rest intervals).
However, the avg power reported by Garmin is 106W.
I have no idea how they compute that because, even assuming that they average including the rest intervals, that should leas to 182 * 60 /72 = 152W, way more than Garmin's 106W.
And if what the OP says also apply to activities forwarded by c2 logbook to GC, then Garmin would see my rest intervals at constant power (of, let's say, 166W), in which case the average would be (207*6+165*12+205*6+165*13+206*6+166*13+205*6+167*9) / 72 = 177W.

So garmin (in my case) underestimates the avg power (compared to the PM5).

I also see Avg Pace wrong: c2 says 2:04.6 overall avg, while Garmin says 2:29. But here the difference is indeed due to the fact that Garmin computes the avg pace also considering the rest intervals: if I do that too, I come up with their number.

They also have a separate section called "Workout Intervals" in which there's something called "other pace" (??) that actually says 2:05 (so aligned to c2 real average pace).

[JaapvanE]

The OP logs on his watch and thinks the fact the PM5 doesn't send out anything during rest intervals is screwing things up on Garmin side, because his watch thinks he's putting out power even during rest intervals.

On ANT+ a PM5 MUST send out a message every 400ms (that is the ANT+ spec). So the PM5 is sending the wrong data. As I explained earlier, with OpenRowngMonitor we implemented the exact same interface, and explicitly zero all relevant values during a rest to make sure a Garmin watch sees a stop.

The only thing I'm saying is that without logging on the watch and simply allowing activity forwarding from c2 logbook to Garmin Connect I'm not sure I see the same problem: Garmin Connect has no data point during a rest interval and it knows it's a rest interval. So yeah, to me GC has all the info needed to compute whichever average or metric correctly. Do they do it ? Not sure.

That is why I looked at the FIT-files, that I assume are exchanged behind the scenes. You can too, https://www.fitfileviewer.com/ is quite a good way to open a file.

In my case C2 clearly reports a rest period in their webinterface of the logbook. FIT-fies can contain rest periods (as the Garmin recorded fit-files show). When I manually upload the fit-file from C2 to Garmin, the rest periods are not handled properly. They lack a zero-ed marker of the start of the rest period (which C2 did recognize on their side), causes that for a five minute rest I get a gradual drop of pace and power from the end of the interval to the start of the next interval.

When I open the FIT-files in fitfileviewer, I see that C2 does mark the intervals. So it might be incorrectly marked or whatever.

It is: 6'+9'+4rest + 6'+9'+4rest + 6'+9'+4rest + 6'+9' (no 4' rest on the last interval). So 60min work, 12min rest, 72min total.

C2 says my avg power is 181W and if I re-compute the average from each non-rest interval data (interval avg pwr and duration), I get the same number: (207*6+165*9+205*6+165*9+206*6+166*9+205*6+167*9) / 60 = 182
As you can see, I essentially stay at ~206W for 6min, then at ~166W for 9min (all repeated 4 times, with 4min rest in between).
So the average is simply (206*6 + 166*9)/15 = 182W. Cool.

On Garmin Connect the power graph looks OK (right values during each interval, no date during rest intervals).
However, the avg power reported by Garmin is 106W.
I have no idea how they compute that because, even assuming that they average including the rest intervals, that should leas to 182 * 60 /72 = 152W, way more than Garmin's 106W.

A basic principle in this is that when C2 says a value is X, Garmin should not re-interpret that (something that Strava actually does do). You could open the file in fit fileviewer and see what power is reported there.

And if what the OP says also apply to activities forwarded by c2 logbook to GC, then Garmin would see my rest intervals at constant power (of, let's say, 166W), in which case the average would be (207*6+165*12+205*6+165*13+206*6+166*13+205*6+167*9) / 72 = 177W.

So garmin (in my case) underestimates the avg power (compared to the PM5).

I also see Avg Pace wrong: c2 says 2:04.6 overall avg, while Garmin says 2:29. But here the difference is indeed due to the fact that Garmin computes the avg pace also considering the rest intervals: if I do that too, I come up with their number.

They also have a separate section called "Workout Intervals" in which there's something called "other pace" (??) that actually says 2:05 (so aligned to c2 real average pace).

In all honesty, intervals are a complete mess across the board. How to tell that an interval really is a rest interval is pretty tough. And getting the recieving side to understand time and distance rowing isn't easy.

[HornetMaX]

A few findings:

1. Last session: if I download the .fit from c2 logbook and open it the viewer (thx btw, useful !) there's something fishy. After the 1st 6min interval (lap 0) instead of having a 9min interval as lap 1, the viewer shows a lap1 with almost no data (but 2 events, one stop and one start I think) and then a lap 2 of 9min. Same thing happens 4 times (as I repeat the 6min+9min intervals 4 times). So instead of having 11 intervals or laps, the viewer sees 15 laps.
2. Today I did a 10min skierg warmup and then 3x (20min +2min rest): both are in c2 logbook but only the skierg is in Garmin Connect. First time it happens to me in since I have the garmin (3 months). So I manually uploaded the .fit to Garmin Connect and it more or less worked (see next point). BTW it is flagged as "Uploaded from Concept2" (even if I manually uploaded).
3. However, I did notice something wrong in the intervals that Garmin Connect sees, in both my today's session and the last one:

   Code: Select all

   Last session on C2:
   Time 		Meters 	Pace 	Watts 	Cal/Hr 	S/M	HR
   1:00:00.0 	14,445m	2:04.6 	181 	922 	25 	157
   6:00.0 		1,511m 	1:59.1 	207 	1012 	27 	160
   9:00.0 		2,100m 	2:08.5 	165 	866 	25 	149
   r: 4:00 	16m 						
   6:00.0 		1,505m 	1:59.6 	205 	1004 	27 	163
   9:00.0 		2,102m 	2:08.4 	165 	868 	25 	154
   r: 4:00 	13m 						
   6:00.0 		1,508m 	1:59.3 	206 	1008 	27 	166
   9:00.0 		2,105m 	2:08.2 	166 	870 	26 	155
   r: 4:00 	17m 						
   6:00.0 		1,507m 	1:59.4 	205 	1006 	26 	166
   9:00.0 		2,108m 	2:08.0 	167 	873 	25 	157
   		r46m
   
   On Garmin Connect (kep only the first columns):
   Interv	StepType	Lap	Time	CumTime	Distance	AvgPace
   1	Other		1	6:00	6:00	1,511		1:59
   	Rest		2	0	6:00	11		--
   2	Other		3	9:00	15:00	2,100		2:09
   	Rest		4	4:00	19:00	16		2:05:05
   3	Other		5	6:00	25:00	1,505		2:00
   	Rest		6	0	25:00	9		--
   4	Other		7	9:00	34:00	2,102		2:08
   	Rest		8	4:00	38:00	13		2:33:33
   5	Other		9	6:00	44:00	1,508		1:59
   	Rest		10	0	44:00	9		--
   6	Other		11	9:00	53:00	2,105		2:08
   	Rest		12	4:00	57:00	13		2:33:33
   7	Other		13	6:00	1:03:00	1,507		1:59
   	Rest		14	0	1:03:00	2,102	--
   8	Other		15	9:00	1:12:00	2,108		2:08
   Summary	--	--	1:12:00	1:12:00	14,491		2:29
   

   If you look at the distance of each interval/lap you can see two weird things: there are rest interval of zero time, but they do have some distance. In interval 1 there's lap 16min 1511m (OK) and lap 2 zero minutes (OK, there was no rest after the 1st interval) but 11 meters.
   Each interval 1, 3 and 5 has the same behaviour: a 1st lap of 6min with proper distance followed by a lap of zero time but some (few) meters.
   Interval 2, 4 and 6 looks OK: a 9min lap with proper distance follower by a 4min rest lap with proper distance.
   However interval 7 is weird: its lap13 is OK (6min, 1507m) while the following lap 14 has zero time but 2102 meters. That's clearly wrong.
   
   And in today's session I have the same: the last empty (zero time) rest lap has a lot of meters that should not be there.

As you said, it's a happy mess

[HornetMaX]

Another example: 4x20min with 3min rest (so 7 intervals overall): logbook session here.

Logbook intervals on the left, Garmin Connect on the right.



In Garmin Connect rest intervals 1 and 3 (laps 2 and 6) have clearly wrong meters (rest 2/lap 4, is OK).
Active intervals are OK, total meters in Garmin Connect do include actual rest meters (19,092 = 19,038+54)

Fitfileviewer shows the same thing as Garmin Connect.

[HornetMaX]

And after emailing Fitfileviewer, the answer is:

"Looks like this is a bug in the .FIT file from Concept2. If you import the file in Garmin, you see the same."

I'll try to send this to concept2.

[HornetMaX]

And after a chat with concept2, the problem seems to be in ErgZone (ErgData works fine).
Now chatting with ErgZone