Anyone test the Polar CS600 power unit ?

[Thom_y]

As I decide whether I will commit to a wireless PT SL powermeter/wheel, I want to explore some of the cheaper alternatives as well. Especially, when I read about some PT SL's crapping out early in their life.

I see Polar is highlighting their CS600 with Power unit on the global website (www.polar.fi) which was seen at Interbike last year. I guess it is not a whole lot different than previous Polar power meters, other than smaller/lighter. My reading suggested the original Polar was tough to set up, questionable in accuracy/reliability, and did not seem to work reliably on trainers etc. etc.

I am just wondering whether anyone has beta tested this new Polar computer and tested against PT SL or SRM powermeters. Someone asked this on wattage forum and not many responses.
The files seen on internet from Alan Cote (designer) comparing original Polar power to PT seem a bit too good to be true:
Polar vs. Powertap files

Thanks for your help.

 

[Thom_y]

Just heard back from Polar USA, seems the CS600 won't arrive until some time in the spring. The power unit delivery date is still unknown. I guess I better not hold my breath for the CS600. Well let's concentrate on the PT SL 2.4 for now.

 

[JTE83]

With a 2:40 min record time for 1 sec it's only good for after work night workouts. For a long distance weekend workout, I'm going ergomo pro. I finally want a fancier pm rather than my 4 Polars.

 

[RChung]

The files seen on internet from Alan Cote (designer) comparing original Polar power to PT seem a bit too good to be true:
Polar vs. Powertap files

Why?

 

[Tom Anhalt]

Why?

Yeah...why? It matches the data taken by Kraig Willet in his BTR comparison article.

BTW, I've been running a Polar vs. a PT Pro "head to head" for ~3 weeks now and have gathered a good number of hours of files. The "shape" of the power output graphs invariably are virtually identical (just like the data you refer to), with the average power level offset by the drivetrain efficiency. In fact, I believe I've been able to observe some of the drivetrain efficiency vs chain tension affects reported by Spicer from Johns Hopkins after analyzing some of the data on a climb I do a couple of times per week. Not so bad considering the stated %error of the 2 devices.

I'll be interested in seeing how the new CS600 handles the averaging of the data for the different recording rates before I fully condemn the short recording time in 1 sec mode. If the 5 sec. record is actually an "average" over the 5 secs, that will be a lot better than just a sample of "every 5th one second reading".

 

[Steve_B]

I'll be interested in seeing how the new CS600 handles the averaging of the data for the different recording rates before I fully condemn the short recording time in 1 sec mode. If the 5 sec. record is actually an "average" over the 5 secs, that will be a lot better than just a sample of "every 5th one second reading".

On the CS600, can you change recording rates in the middle of a ride? That might be one way of dealing with it. when you're doing sprints, set the recording rate to 1 sec, otherwise something less frequent.

 

[Thom_y]

On the CS600, can you change recording rates in the middle of a ride? That might be one way of dealing with it.

The manual for the CS600 can be downloaded from this site ... don't know if it says whether you can change the recording rate on the fly ... as well, I don't know if you go to 5sec recording whether it averages for the 5 seconds or just samples at 5 second intervals:

CS600 info

 

[Tom Anhalt]

On the CS600, can you change recording rates in the middle of a ride? That might be one way of dealing with it. when you're doing sprints, set the recording rate to 1 sec, otherwise something less frequent.

My guess would be no...since that would be part of the "setup".

My point about the 5 sec record being a "true" average would make the 1 sec recording not as necessary for most rides....including sprint workouts, since the most useful values aren't peak 1 sec values, but values over 5, 10, 15, or 20 secs.

The way the Polar operates now (display update every ~2secs, whatever is displayed saved at 5 sec interval), short, high "peaks" tend to get "clipped"...although not as badly as I had assumed compared to the PT. Part of it is making sure you keep the chain moving with the Polar after the sprint is over...as soon as the chain speed sensor detects zero speed, the power calculation goes to zero. If you keep the chain moving just slightly, the next reading in the display tends to be more of a gradually declining moving average as opposed to forced to zero...

Of course, the PT can tend to "over-report" peak values since it's calculation and recording rate is independent of crank position. This means that one particular sample may be skewed upwards by only having a portion of a crank cycle...with the next record, of course, being skewed downward for the same reason. Search on the google wattage list for "precession" or "aliasing" for a discussion of this "feature".

I REALLY hope Polar didn't "screw the pooch" in this area like with the memory...I won't hold my breath though.

 

[Alex Simmons]

Part of it is making sure you keep the chain moving with the Polar after the sprint is over...as soon as the chain speed sensor detects zero speed, the power calculation goes to zero. If you keep the chain moving just slightly, the next reading in the display tends to be more of a gradually declining moving average as opposed to forced to zero...

But if you ain't pedalling, then power is zero and that's what should be recorded, isn't it?

It's still a problem even if you do have 5 sec averaging since you may not start your effort exactly when the polar starts its next 5 sec recording "block". So the averaging may still miss a 5 sec max, since it might occur across two data points.

Of course, the PT can tend to "over-report" peak values since it's calculation and recording rate is independent of crank position.

Yeah - but in practice I see really consistent data when doing sprints efforts. Both peak power and shape of power curve v time. And so do my 5 trackie mates who all use PTs. Goes for standing starts as well as flying sprints (which have vastly different force/cadence relationships in the first 5 seconds).

 

[Tom Anhalt]

But if you ain't pedalling, then power is zero and that's what should be recorded, isn't it?

Well...not exactly, since the display value is more an average across ~2 secs. It's kind of like with a PT display set at a moving average. The problem is, that the "moving average" basically get's "shut off" when the chain speed signal goes to zero.

It's still a problem even if you do have 5 sec averaging since you may not start your effort exactly when the polar starts its next 5 sec recording "block". So the averaging may still miss a 5 sec max, since it might occur across two data points.

Yeah - but in practice I see really consistent data when doing sprints efforts. Both peak power and shape of power curve v time. And so do my 5 trackie mates who all use PTs. Goes for standing starts as well as flying sprints (which have vastly different force/cadence relationships in the first 5 seconds).

Of course...but I also assume that you don't use your 1.26s "peak" value as anything more than entertainment, right? Especially when cadences are low, like in a standing start?

 

[Alex Simmons]

Well...not exactly, since the display value is more an average across ~2 secs. It's kind of like with a PT display set at a moving average. The problem is, that the "moving average" basically get's "shut off" when the chain speed signal goes to zero.

I was thinking about what's recorded, not what's displayed. Recording a moving average would be just silly. Maybe I have my wires crossed?

Of course...but I also assume that you don't use your 1.26s "peak" value as anything more than entertainment, right? Especially when cadences are low, like in a standing start?

We still find them consistent. Enough to make decisions about when to pull the pin on a workout.

 

[Steve_B]

I was thinking about what's recorded, not what's displayed. Recording a moving average would be just silly. Maybe I have my wires crossed?
We still find them consistent. Enough to make decisions about when to pull the pin on a workout.

I should probably go read the Wattage archives before I comment but...

<geek mode on>
The Nyquist theorom would say that aliasing becomes more of an issue when the major underlying frequency components involved are > 2/1.26 (=1.58) Hertz. I'd need to do a Fourier transform on one of my sprint power waveforms (can do in Excel or Matlab) to be really sure but I would suspect that a sprint would involve some frequency components that might not get accurately reflected in the fastest PT sampling due to this.

Having said that, I find the recorded values quite consistent as well, assuming I'm not trying to do a sprint workout fatigued. Perhaps the PT is low-pass filtering my sprint workouts or but also perhaps my individual style of sprint doesn't have a lot of "high frequency" energy.
<geek mode off>

 

[RChung]

I should probably go read the Wattage archives before I comment but...

Go read the Wattage archives.

 

[Thom_y]

original Polar ...did not seem to work reliably on trainers

Some interesting observations just provided to a similar thread on the Wattage forum:

"... had the opportunity to take a look at the CS600 today at a bike show.
Evidently the batterycompartment has been shrunk down quite a bit. It now
contains something like two AA batterys it looked like.

It could be mounted either on the stem or on the handlebars.

I had a chance to ride it on a trainer and my strongest impression was that
the readings were very much more stable than I'm used to, with a Polar, on a trainer. I'm sure lots of it has to do with smoothing alghorithms.

But, and now comes the interesting part. It seems they have changed the frequencys around somehow as to make it less likely to be disturbed by radio *noise*' Exactly how this was accomplished
and how it would affect power readings from a trainer wasn't answered.
Unfortunately the people in the Polar booth knew less about it than I did

I did however get sufficiently interested in it and it sure looks like a
better buy than the old one. "

This sounds encouraging ... too bad it may be awhile before it hits North America

 

[J-V]

I'll be interested in seeing how the new CS600 handles the averaging of the data for the different recording rates before I fully condemn the short recording time in 1 sec mode. If the 5 sec. record is actually an "average" over the 5 secs, that will be a lot better than just a sample of "every 5th one second reading".

I've got some bad nes for you, Tom. Just got this from Polar Finland:

"Dear J\V,

Thank you for your email, Tony has forwarded it to the Polar Global
Customer Support.

1. With 1s sampling rate and all features on you can record 2h 40min.

2. The calculation in the new 2.4GHz Power Output sensor is the same, i.e. the samples are taken every 1, 5, 15, 60s exactly on that point, not averaging or random value between these points.


Best Regards,
Global Helpdesk/Riitta
Polar Electro Oy
Finland"

Looks like we'll now be stuck between a rock and a hard place. They could have fixed either issue (i.e. adding memory or fixing the 5-second issue) and solved the problem, but apparently weren't quite 'on-the-ball' enough to get it figured out prior to production.

Looks like I may be looking at other options after all...

-J\V

 

[J-V]

And my response to them:

Thank you for the response, I appreciate it.

Unfortunately, since Polar choose not to address the ultimate problem, I can assure you that this means that there is a large part of the potential market that will remain untapped until Polar either:

1) Increases the storage/memory capacity of the CS600

or

2) Makes the 5-second recording a more accurate portrayal of reality by averaging values, rather than merely picking them at 5-second intervals, which in itself is a rather 'random' way of approaching it.

I'm quite dismayed by this lack of forethought on Polar's part; flash memory of many sorts is so incredibly cheap these days, hobbling a brand new product with insufficient storage capacity is almost unbelievable. Turning off one of the recorded values in order to increase the recording time effectively negates one of the Polar unit's primary advantages over the competition, that is, that you can capture ALL facets of a ride's data (i.e. speed, distance, cadence, HR, watts, and altitude).

Cyclists using this product are pretty serious, and a 2:40 ride is considered a 'short training ride' by many; the 5-second recording scheme having an 80% chance of missing peak values makes it invaluable for the evaluation of neuromuscular power, so sprinters are effectively 'left out'. I acknowledge that for evaluation of FTP (i.e. longer intervals) that the 5-second issue doesn't have a large impact, but that's besides the point.

I was very excited by this product and I'm a dedicated Polar user (have been for 10+ years), but I cannot spend $700 US on a product that incorporates an element of planned obsolescence which makes it partially unusable for its intended use.

Respectfully,

J\V

I'd suggest anyone who cares send an e-mail to:

[email protected]

and

[email protected]

and let them know your thoughts on this new product.

 

[JTE83]

Thanks for this email and your reply to Polar!

I just got an ergomo pro, and only have 1 ride into it but by looking at my power numbers I'd say I had better watt numbers from my Polar units! With proper setup and using my supplied K factor BB calibration of 194 I was getting high readings for my ergomo watts! Now I have to fudge my K factor to get proper watt readings. Polars calibrations are just the chain measurents and that's so straightforward and once you've got those you get proper watt readings on the Polar.

With 2:40 record time I'd use the new Polar only for after work night workouts. Yeah, maybe I'll email Polar too and give them my thoughts about more memory.

 

[Tom Anhalt]

I've got some bad nes for you, Tom. Just got this from Polar Finland:




Looks like we'll now be stuck between a rock and a hard place. They could have fixed either issue (i.e. adding memory or fixing the 5-second issue) and solved the problem, but apparently weren't quite 'on-the-ball' enough to get it figured out prior to production.

Looks like I may be looking at other options after all...

-J\V

Wow...what a bunch of dumb-a#%es (pardon my French, er...Finnish). If there ever was a product that had the potential to be a "market leader" that was continually hamstrung by poor product development decisions, this would be it.

Just yesterday I had a teammate ask me if he should see about getting a CS600 when they become available...I'm not sure exactly what to tell him...

I guess I'll be holding on to the PT Pro I recently acquired a bit longer...

 

[JTE83]

Hey, would all of you Polar guys consider the iBike Pro? I think it has 10 hrs of 1 sec recording. I just don't know the latest news about this powermeter.

1 ride into my ergomo pro and I'm kinda disappointed. But I forgot to bring my anemometer to measure the wind speed during my workout. Too early to say more about my ergomo.

 

[J-V]

Hey, would all of you Polar guys consider the iBike Pro? I think it has 10 hrs of 1 sec recording. I just don't know the latest news about this powermeter.

1 ride into my ergomo pro and I'm kinda disappointed. But I forgot to bring my anemometer to measure the wind speed during my workout. Too early to say more about my ergomo.

However much potential there is for the iBike to be accurate, it'll never be more than potential, since power is not measured. So, I guess that'd be a "no" for me.

I was willing to settle for the Polar's idiosyncracies to a point, but not this.