Login required to started new threads

Login required to post replies

Prev Next
Re: Quarq DZero Erratic Cadence and Power Readings [troy@quarq] [ In reply to ]
Quote | Reply
troy@quarq wrote:
Quarq’s accelerometer cadence was designed so that a cadence discrepancy in any single pedal stroke would be adjusted in the next pedal stroke. Therefore cumulative accuracy is unaffected and there is little or no impact on cumulative power values such as kilojoules, average power and normalized power. In objective terms, single pedal stroke accuracy in the worst conditions varied from magnet-based cadence by a maximum of 2%. And benchmark tests showed the accuracy of Quarq’s accelerometer cadence matched or surpassed other products that use accelerometers. We shared this information in the Quarq website’s FAQ for many years.
Not sure what this does to aero testing data, but it doesn't sound good.
Quote Reply
Re: Quarq DZero Erratic Cadence and Power Readings [troy@quarq] [ In reply to ]
Quote | Reply
Hi Troy,

Many thanks for the response.

Indeed, in my conversations with your support guys he mentioned the "adjustment" of cadence values from one pedal-stroke to the next and I agree that it certainly seems to work to ensure the stats over longer intervals are accurate; in my experience, even with laps as short as 10-seconds, the noise seems to average out pretty well. Are you saying that the bug causing these erratic values is something to do with this stroke-to-stroke compensation?

As for using the DZero for aero testing, I've done a bit of this myself and had excellent results from it, so I don't think that compensation data-processing or the power meter itself are a limitation there. As a side-note, I suspect that particular run was done when the unit was in one of its periods suffering from this problem, although I can't say for sure because the issue comes-and-goes and it's only really apparent under very controlled conditions (i.e. fairly steady-state turbo trainer rides). What I'm trying to say is that the data obtained there is likely a worst-case scenario, which is very encouraging indeed from that perspective. I very much want to do some more aero testing with it, but the British Winter/Spring weather (freezing temperatures, then rain, now wind!) has scuppered most of my available opportunities since.

I'm looking forward to the new firmware release and thanks for giving us a rough-timeframe. Is there any kind of mailing list that we can sign up to for notifications of firmware updates? And thanks again for the great support!
Quote Reply
Re: Quarq DZero Erratic Cadence and Power Readings [awenborn] [ In reply to ]
Quote | Reply
Hi awenborn,

The adjustment routine and display bug are related. We analyze cadence data many times inside one pedal stroke. We revised the algorithm that delivers cadence for the entire crank revolution -- what you see on the head unit -- so the number is steady. I must add that it is now more precise and remains real-time. It is not an average and not filtered or delayed. The enhancement to accelerometer cadence to improve accuracy was discovered and made as part of our investigation.

You can register your power meter here: https://www.sram.com/registration/quarq This mailing list is for firmware updates and important production information. At the bottom of the Quarq home page you can join our regular mailing list where we announce new products, share tips and tricks, and highlight sponsored teams and athletes among other things.
Quote Reply
Re: Quarq DZero Erratic Cadence and Power Readings [awenborn] [ In reply to ]
Quote | Reply
awenborn wrote:
Just thought I'd put this out there and see if you guys have any opinions or relevant experiences with regard to this.

I've been using a Quarq DZero power meter on my turbo trainer all winter and have noticed a weird phenomenon that arises when I do mechanical work on the bike. Each time, for the first couple of weeks (perhaps around 10 workouts or ~20 hours of riding) after refitting the Quarq to the bike it seems to give erratic cadence and power readings. For example, say I'm trying to hold a constant cadence of 90 rpm, my Garmin GSC-10 will say I'm holding it very steady in the 89-91 rpm range but the Quarq AxCad values will be jumping around all over the place in the 86-94 rpm range. What's more confusing is that after a couple of weeks, the Quarq AxCad measurements sort themselves out and become nice and steady like the GSC-10.

This process has happened to me twice now, once when I first installed the Quarq and a second time when I removed it to change chainrings and reinstalled; same process both times, with a couple of weeks of readings all over the place and then settles down to something acceptable. I trawled back through my TrainingPeaks library to find a nice visual representation for you guys; here's two identical steady-state workouts, two weeks apart before and after this apparent settling-in:

Note: the scales for both power and cadence are identical here and obviously the smoothing slider is set to zero for both.


February 16th:




March 2nd:




Furthermore, in between these two workouts, I found another one where the erratic cadence values seemingly fix themselves mid-workout; here the cadence was all over the place for the first 1/3rd of the workout as it had been for the last couple of weeks, I start doing some seated-to-standing transitions (the series of dips in cadence below) and after this the erratic cadence has fixed itself and has been great ever since:

February 24th:




The cadence itself doesn't annoy me so much, more the fact that the Quarq is using this rubbish cadence data to calculate the power and as you can see in the two workouts above, it has a big effect on the variability of that too.

It looks to me like there's some kind of mechanical settling-in going on here and the eccentric stress of the out-of-the-saddle efforts have fixed whatever was wrong and helped the Quarq to read more consistent values. Last time I checked all of the torques on the chainrings, bottom-bracket etc and nothing was amiss so I'm at a loss as to what could be happening, does anyone have any idea? Possibly the chainrings settling in and being forced up tightly against the chainring bolts or something? I've noticed a few complaints about Quarq's AxCad online and initially thought it just sucked compared to a magnet-based system, but then once it settled in it was almost identical to the GSC-10, so I'm thinking it might also be an issue for these people too. Has anyone seen this before with a Quarq or other crank-based power meter?

I'm planning on a full strip-down service of my bike at the weekend to get it ready for the outdoors again and fully expect to see the issue resurface for another few weeks. Just wondering if there's anything that I can do to avoid it returning?

Details:
Quarq DZero AL 110 BCD
SRAM Team GXP BSA bottom bracket (no frame adapters etc)
Shimano 105 5750 rings (50-34)

KMC X10-EL chain

N.B. regarding firmware, I was running v2.0 for the first workout here and upgraded to v3.0 that supposedly addresses an issue with erratic cadence values shortly thereafter and it didn't help at all; it was a couple of workouts later (Mont Gosford above) that the issue fixed itself mid-workout as shown.

This is a very simple problem, one that had me condemn Stages and somewhat more "accurate" PM as this, in one word: ACCELEROMETER. What this means is that it uses angular velocity (and the internal accelerometer ability to keep up e.g. smartphone not turning with rotation?) to calculate and measure RPM and then offer you POWER. BAD BAD BAD BAD BAD BAD BAD BAD BAD. Challenged DC Rainmaker and guessed that he lives in hilly places where he could not reasonably determine that items such as Stages were 15-20% off, 10-40x less accurate than magnet sensors, all relative to the ANGULAR power or rotation speed, which may vary.

GC-10 is an ELECTROMAGNETIC based, light speed measurement speed and accuracy: each time the magnet crosses the circuit, it generates an impulse. The frequency of these pulses give your RPM, and, on flats, I can attest to it being 99-100% accurate. Whereas GC-10 would flatline me at 95, 97, 100, 120 for minutes or hours. It would only change with my physical effort. Not Stages: it would report, STROKE AFTER STROKE 90- 100-110-90-116 etc. Same for power. 270-220-240-300- etc Up and down angular, down stroke vs pull etc.

Stages also hijacked RPM reporting, thus I could never pair and use GC-10 etc. Threw Stages (well, reimbursed after three units and 20 bad reviews online)...

So this Quarq is MORE precise, as it has a more refined angular mechanism, but, in your case, 10, 20x more inaccurate than your GC-10 which could be within 0.5 RPM, your unit oscillates by 10 (so x20 inaccuracy). Reasons could be accelerometer sensitivity, some liquid or gas got in, heat, who knows? In the end you get Accelerometer based RPMestimate, whereas GC-10 gives you light speed accuracy.

Had Quarq kept the RPM separate from the accelerometer, might have been worthy Instead, it becomes a garbage experience that forces you to look down, more than you should, and be unsafe. RPM is a precise light speed thing to size up with a magnet, and the ONLY inaccuracy should be your physical stroke. With the Accelerometer, you add a multifold amplification of errors (relative to the quality of the accelerometer liquid etc), and it goes HIGH during the downstroke, and LOW during the pull.

Hence why I think Powertap is the best tech out there, or something inside the bracket. unsure which. But whenever I hear accelerometer, I hear Innacuracy.

I also had your charts, ups and downs, went back to GC-10, HR and VO-2, and am still waiting for a non-accelerometer based PM, or one that does not shove up our throats inaccurate RPM estimates...
Quote Reply

Prev Next