New powermeter rumours

[maquisard]

Anyone tried the power estimate monitor from PowerTap based on heart rate? Obviously not a true power meter but looks interesting as a cheap way to get a kind of gauge

1. export ride data to spreadsheet
2. plot power versus heart rate

You get a big cloud. You can try smoothing power using different algorithms, even smoothing and delaying. You still get a cloud. HR isn't a good predictor of power.

- On a side note, one thing I would love to do is properly study the relationship between HRV ( heart rate variability ) and TSS over the course of a racing season. I have some HRV data for myself using ithlete on the iPhone but not enough to cover the couple of seasons for which I have TSS. In theory, HRV should be able to show increases in training stress but it would be interesting to do an analysis of just how much this holds true.

[kervelo]

A lot of research related to HR variability has already been done by Firstbeat.
http://www.firstbeat.fi/physiology/heart-beat-analysis" onclick="window.open(this.href);return false;

The Firstbeat algorithms are used in Suunto and Garmin devices.

[djconnel]

post-processing won't work because the data are not there to access- you need to be able to analyze (in milliseconds and in real time) the variability from one heart beat to the next- that is not the same as heart rate itself.

uraqt, I suspect they are using some algorithmic variant of that. There is a lot of work done on heart rate variability as it relates to autonomic function, and this is how they are deriving the power measurements.

I can also post-process speed and elevation data to estimate power. I suspect that does a lot better than heart rate.

Here's a good case: Kevin Metcalf riding the trainer. He's an excellent local racer, multiple master's national champions and even world championships I think, and a substantial fraction of his workouts are 2x20's on the trainer, Z3:
link to Strava

Interesting here is the combination of speed, power, and HR:
1. Power is the most constant of speed, power, HR. Kevin is good at these.
2. Speed is decreasing over the interval. Resistance changes as the fluid in the train heats up.
3. HR is steadily increasing over the interval. This is "cardiac drift"

In this case, speed is more closely correlated with power than HR, but even in the controlled environment of a trainer, the power meter is still not easily predicted by either of the other two.

[eric]

It's not heart rate they're using, it's the variablity in time between heart beats.

There's a ton of papers linked from the firstbeat site linked above.

I'm not saying that I think it's as good as a power meter, but if you're going to criticize it you should understand what you are criticizing.

[djconnel]

Okay -- sorry, I misunderstood. I have a hard time finding a description of the specific algorithm. However, HRV seems to depend on many factors, according to this paper. It's an interesting concept.

[kulivontot]

post-processing won't work because the data are not there to access- you need to be able to analyze (in milliseconds and in real time) the variability from one heart beat to the next- that is not the same as heart rate itself.

uraqt, I suspect they are using some algorithmic variant of that. There is a lot of work done on heart rate variability as it relates to autonomic function, and this is how they are deriving the power measurements.

How the hell does one analyze heart rate in milliseconds in real time when the time between heart beats is 333msec at 180BPM? I'm sorry, I'm not buying that you need 1kHz+ HRM sampling to do what the powercal is doing. I would argue that combining heart rate variation with GPS, elevation, and weather conditions in post processing you could get way more accurate than powercal can do on the fly. It's just that nobody's done it yet.

Ok, I'm starting to the argument here. 160bpm is not exactly 160 bpm, because there's variance around the 375msec point with a certain distribution, but average will be the same... I'll have to read these papers.

[maquisard]

HRV is not used* in any power estimation device to provide a real time estimation of power. It is used to provide an estimate of training stress,

* if you believe this statement to be incorrect please provide cited evidence.

[mentok]

I can also post-process speed and elevation data to estimate power. I suspect that does a lot better than heart rate.

Here's a good case: Kevin Metcalf riding the trainer. He's an excellent local racer, multiple master's national champions and even world championships I think, and a substantial fraction of his workouts are 2x20's on the trainer, Z3:
link to Strava

Interesting here is the combination of speed, power, and HR:
1. Power is the most constant of speed, power, HR. Kevin is good at these.
2. Speed is decreasing over the interval. Resistance changes as the fluid in the train heats up.
3. HR is steadily increasing over the interval. This is "cardiac drift"

In this case, speed is more closely correlated with power than HR, but even in the controlled environment of a trainer, the power meter is still not easily predicted by either of the other two.

why do you post such things? is it just to hurt me? those intervals look brutal and the guy only weighs 75kg. i have 15kg on the guy and i don't know that i could do that session...

on topic though, yes speed correlates to power far better than any other metric while on my rollers. I'm currently recovering from an injury and only using the back wheel on the rollers so power/speed/cadence relationships fall apart when my weight distribution changes, but in general it is extremely linear.

While my PT was broken last year i was stuck doing intervals fixed on my SRM'd track bike using my PCV to monitor/record. the PCV doesn't have garmin-like smoothing (3s, 10s) so the power display jumps around a bit. to overcome this, i would just dial in the speed that corresponded to the appropriate power and hunt it instead. this made it much easier to sustain my targets power targets without searching all over the place.

[jmilliron]

Welp, my Garmin Premium HR strap has been acting up for the last couple weeks. Replaced the battery and really cleaned the strap a few times. Still doing funny stuff mid ride, long after I've warmed up.

Went a head and paid a little extra for the PowerCal, as a replacement. Probably not even going to put the power up on my Garmin. I'll just be happy if the TSS is close. I've got a couple years of PowerTap and Quarq rides, so I should be able to tell.

[eric]

Can you track powercal data in addition to your power meter? That'd be interesting to see.

[jmilliron]

Don't think so. Would need a second head unit.

In between PMs at the moment. Quarq didn't fit on my frame and I wasn't happy with it anyway.

As long as the PowerCal gives me good HR numbers and a reasonable TSS, it will be worth the price.

[claus]

Can you track powercal data in addition to your power meter? That'd be interesting to see.

Are you asking for data from a "real" powermeter and a PowerCal for the same ride?

On one of my bikes I use a Polar powermeter and a Powercal to compare the data from them.

There are several other comparisons available, e.g., from Rodrigo and DC Rainmaker.

[djconnel]

Update on Garmin Vector firmware. All those Vector users probably want to upgrade .

http://www8.garmin.com/support/download ... sp?id=5713

This is clearly a subtle hint from Garmin.

[eric]

Its probably for the beta testers.

[jmilliron]

Yeah, saw one on today's ride:

Check out what I saw in person:



He of course couldn't give me any interesting info on them. Signed an NDA.