Anyone have any experience with the SmartRow handle yet?

[Tsnor]

...It's a family business with dedicated people, I had the chance to get to know a little.

Suggest you point them to Jon's post. viewtopic.php?f=10&t=209751#p595858 Let them decide if they want to incorporate the changes in their V2 version.

Also suggest handle based HR. At a guess more people would want HR then would want competitive data that is not normalized to C2's version of watts and splits.

At a certainty, if the handle (correctly) gives lower watts or calories than the rower's built in display then the rower's user will be mad at the handle. And if the handle were to get a few more inputs and display a fitness 'age' that was 15 years younger than actual age then people would love it. Stupid but true.

[JaapvanE]

Also suggest handle based HR. At a guess more people would want HR then would want competitive data that is not normalized to C2's version of watts and splits.

Depends where you are coming from. On a new C2 I'd agree, for older (pre-PM5) machines it would provide connectivity to apps and games, which many find interesting. Similar for unconnected machines, where data is BS to begin with and often not connected via Bluetooth.

At a certainty, if the handle (correctly) gives lower watts or calories than the rower's built in display then the rower's user will be mad at the handle. And if the handle were to get a few more inputs and display a fitness 'age' that was 15 years younger than actual age then people would love it. Stupid but true.

Seen it happen indeed.

[iain]

At a certainty, if the handle (correctly) gives lower watts or calories than the rower's built in display then the rower's user will be mad at the handle. And if the handle were to get a few more inputs and display a fitness 'age' that was 15 years younger than actual age then people would love it. Stupid but true.

Probably too true! But depends upon their target audience. The Watts I think would be an issue as many serious users may be interested in comparative performance or meeting thresholds for boat selection. C2's results are the standard. Clearly "pace" is arbitrary, I would hope that generally Watts are reasonably accurate given the meaning of ergometer!

As for cals, serious users generally treat this with the skepticism it deserves. I always hasdd a rye smile at the gym user I used to see regularly who did 100 cals of work at 3:00 pace!

[JaapvanE]

The Watts I think would be an issue as many serious users may be interested in comparative performance or meeting thresholds for boat selection. C2's results are the standard.

As far as I understood from various YouTube video's: most national teams use time trials for crew selection, and don't look at power.

Clearly "pace" is arbitrary, I would hope that generally Watts are reasonably accurate given the meaning of ergometer!

I can't speak for the C2 implementation, but I know the underlying physics and measurement issues extremely well. In my opinion, you are wrong here. There is a bit of a paradox here: there is a direct relation between pace and power displayed, making them essentially interchangeable. However, how you get to them, especially in the presence of noise, is a different matter.

First off: there is only one thing the PM5 really knows, and it is flywheel position and the time it took to get there. The rest is just maths and physics based on that. This is where you get your paradox.

Second of all: power is measured on the flywheel and thus completely ignores the rowers own weight which needs to move up and down a slide. Here strokerate and aggressiveness of the the drive come into play. These is considerable work (50+ Watts) which tends to behave totally different on the water. That part is problematic for both power and pace, so I'll ignore that part in the rest of my text.

In essence, the only theoretical difference between power and pace is the magic number 2.8. One can indeed debate whether the 2.8 is an arbitrary number that relates power to pace. However, it was determined by 2 Olympic rowers. And even RP3 acknowledges it is a pretty decent approximation based on their own research (they actually use of different ratio's for simulating different boats).

Life becomes a lot more interesting when we get to the practical side. Research by the University of Ulm shows that C2 uses the simplified power formula (i.e. average power across the stroke = drag * average angular velocity across the stroke), which completely ignores the netto acceleration/deceleration of the flywheel across the stroke (see here for my detailed analysis). Ulm's research confirms this as inconsistent strokes thus can have huge effects on power measurement. So to say that C2 measures power accurately, is extremely debatable to begin with.

With OpenRowingMonitor I actually experimented with using the complete formula next to the simplified formula. Despite ORM having a far more accurate stroke detection and a much more robust approach to flywheel speed measurement than a PM5, results were quite different and the complex formula results in more unstable power measurements (as is my rowing probably), and thus it becomes challenging to compare the two. So power is a really volatile measurement, prone to noise, even when done right. So there is a lot to say for the simplified formula, but it isn't accurate to begin with.

An additional issue is that C2's stroke boundaries are incorrectly defined (as in: easy and reliable to detect, but inaccurate/shifted), the effect of the simple formula is actually enlarged: small measurement errors, especially at the start of the drive, can create significant metrics errors. Although on average the across a stroke this stabilizes it a bit, it is a noisy measurement to begin with.

A key issue is that flywheel speed is the first derivative of flywheel position, and that small measurement errors in position/time reporting tend to be enlarged. So small measurement errors in time/position lead to bigger errors in flywheel speed. As the power calculation depends on flywheel speed measurement, it tends to enlarge noise. When you look at the raw force curve data communicated via the BLE interface you see that happening: that data contains quite some noise and erratic behaviour. When you look at apps like EXR (which doesn't smooth force curves), it is quite noisy. It shows that the first and especially the second derivative are unstable, even for rowers with decades of OTW competition experience.

Looking at linear speed, life becomes a lot easier. The simplified formula provides a direct route to linear distance travelled: Linear Distance = (drag/2.8)^1/3 * Rotational Distance.

This formula is completely independent from stroke detection (although the drag is updated per stroke, it tends to be pretty stable). In essence, it uses a direct linear relation between rotational distance and linear distance. In my book, this is actually a pretty good approximation as the flywheel tends to have the same drag behaviour as a boat. As a PM5 updates distance in-stroke, I suspect they use the same approach.

For speed across the stroke you still need stroke detection. Small time errors tend to drown out in the length of the measurement. The only noise is stroke detection being unstable, and there the PM5's approach is sufficiently robust enough to handle that. In OpenRowingMonitor we use this approach as well, and our experience shows that you can completely mess up stroke detection and still get decent distance and pace estimates. So we calculate power based on pace measured, not the other way around.

So claiming that power is the most reliable measurement in the presence of noise. Can't agree with that.

The SmartRow actually approaches this from the other side: they really measure the force and acceleration of the handle. First integral gives you speed, second integral gives you position. As force x distance travelled is power, there this might hold more true that their power calculation is most robust against noise. Measurement noise tends to create displacement errors, having quite a significant impact of power/distance calculations, making this part extremely challenging. But it would be interesting to compare the output of their force curve with the PM5's.

[iain]

The Watts I think would be an issue as many serious users may be interested in comparative performance or meeting thresholds for boat selection. C2's results are the standard.

As far as I understood from various YouTube video's: most national teams use time trials for crew selection, and don't look at power.

Sorry I was not clearer, I was referring to the overall "time"/distance for the TTs that depend upon these. For this it is the relative not the absolute values that matter. Also Power / a calculation based on power are equivalent for this purpose.

In addition, there are far more people going for College selections than National! (My definition of "serious" is obviously lower than yours!).

Clearly "pace" is arbitrary, I would hope that generally Watts are reasonably accurate given the meaning of ergometer!

I can't speak for the C2 implementation, but I know the underlying physics and measurement issues extremely well. In my opinion, you are wrong here. There is a bit of a paradox here: there is a direct relation between pace and power displayed, making them essentially interchangeable. However, how you get to them, especially in the presence of noise, is a different matter...

I was not aware of the validation of the 2.8 factor, but even so it will change as boats improve and most of us would have a much bigger discrepancy from stroke efficiency OTW than OTE and so introducing a further difference.

Thanks for sharing your knowledge. I know I am overly simplifying, but I take from your analysis that essentially for longer non-extreme purposes, the C2 is pretty good at measuring an "average" power although this ignores the impact of the cube of average speed NOT being proportional to the average of the cube of speed (ie the deviation from a straightline of the cube law). So essentially C2 gives a pretty accurate relative measurement of average pace between different TTs.

[JaapvanE]

I was not aware of the validation of the 2.8 factor, but even so it will change as boats improve and most of us would have a much bigger discrepancy from stroke efficiency OTW than OTE and so introducing a further difference.

That is actually a pretty interesting question. The 2.8 was set decades ago, but would it change measurably due to newer boats/oars?

Thanks for sharing your knowledge. I know I am overly simplifying, but I take from your analysis that essentially for longer non-extreme purposes, the C2 is pretty good at measuring an "average" power although this ignores the impact of the cube of average speed NOT being proportional to the average of the cube of speed (ie the deviation from a straightline of the cube law). So essentially C2 gives a pretty accurate relative measurement of average pace between different TTs.

Indeed inconsistency across strokes is punished by the monitor: the energy stored in the flywheel isn't accounted for. And indeed, high peaks in the strokes are flattened while these require huge power. There are lots of these optimizations and assumptions in a rowing monitor but results are mostly repeatable. As a C2 is designed as a training device, one can actually debate if some of this behaviour (like punishing stroke inconsistency and exploding at the catch) is actually a good thing.