Vector 3 accuracy thread

Thanks we take look at the data. There definitely appears to be something going on with some models of trainers when in erg mode.

Just got my pair of Vector 3 and did a 3-way comparison between Vector 3, Stages Gen 2 and Kickr Snap 17.
Setup: Kickr Snap: Tyre pressure 110PSI. 2 1/4 turns resistance. Adv. spindown performed a week ago. Warmed up @100 watts for 10 min. Performed normal spindown: 13.5 sec. Vector 3 and Stages were also zero offset. Vector 3 recorded on Fenix 5, Stages on Wahoo Bolt, Kickr Snap on Zwift. Zwift controlling the Snap in a Erg mode workout, with some freeride at the end.

Here is the result: https://analyze.dcrainmaker.com/#/public/a16c1d49-0c2c-4ab5-5b40-e6b93f52868d

I have no reason to believe either Stages or Snap is accurate, but I'm surprised to see Vector 3 reading lower than both Stages and Snap. I would at least expect Vector 3 to read higher than Snap due to drivetrain losses.

Vector 3 confirms I have an average 50/50 L/R balance.

I'm also observing that the Vector 3 and Snap are matching more and more closely towards the end of the workout when Erg mode was turned off.

What do you think?

Interestingly looking at the Left/Right power chart for much of the time one of the Vectors seemed to be ~10W different to the other series but it switched around. From 17:55 to 37:55 the left seemed low, from 42:55 to completion the right seemed low.

I've been riding with V3 since late november on my Kettler Ergoracer S. Although there is a difference in absolute power (for whatever reason and which is not the point here) I always end up seeing the same difference in power.

What I mean is that when I do a classic ride, that would be auto 235W (this results appr. 215W in average), the V3 always displays 8-10W less. I'm writing this because it's quite frankly everytime I ride a classic steady ride. I'm seeing almost the same when doing hard intervals; once again the compared result is always within 8-10W of Kettler.

I'm putting this up, just to add, that I do get a picture of the V3's are very consistent - regardless of they're not showing exactly the same wattage as Kettler. The error is always the same. I don't have the exact numbers listed. This is just coming from me looking at the numbers while riding + some post analysis comparison.

I think this is a really good sign - consistency must be above all other parameters. Whether I will see the same consistency once starting to ride outside, has yet to be answered.

One thing to note; I use auto-zero-offset, so I never do it manually.

I posted this in another thread but realize that this belongs here..... So in erg mode on my kickr2 ( 207 watts at a cadence of 90), I was performing 1 min time trials to see what was causing my vector3 to avg consistently at 199-200 watts. I turned off cycling dynamics and for the next trial, the vectors avg 206 watts for the 1 min trial!! I turned cycling dynamics back on for the next trial and the avg went back down to 199ish. The strange thing is that when I turned cycling dynamics back off the avg remained at 199ish. Seems like I could not duplicate the 206 trial. And then I had to go to work. Completely dismayed I am at the fluctuating avg as well as the generally underreporting of power. Cycling dynamics has my left right balance generally at 50/50 so my stages and kickr2 are in line.

Completely dismayed I am at the fluctuating avg as well as the generally underreporting of power.

It has been reported (see this article as an example) that the Kickr can have test to test variability, and that the accuracy of Kickr units generally can be in the 3% range. As such, and based on the limited number of data sets, I would not be concerned that either your Kickr, or the Vector 3 set, is not working correctly within their corresponding limits of accuracy and repeatability. The fact that the difference between Vector 3 and your Kickr seemed to be correlated to changing the cycling dynamics setting, in two out of four trials, may just be a coincidence due to variability in the erg mode tracking on the Kickr.

If you have doubts about the way your Vector pedals measure force then you should do the hanging-weight test. It will independently establish whether or not your pedals are OK.

I've recently been using the Garmin Vector 3S's and am having problems with them reading high. They seem to be accurate at lower power levels (around 200w or lower) but read ~20 watts higher at higher output (becomes more noticeable above 250w). I calibrate the powermeter at the prompt with the correct crank arm length before every use.

I have done the hanging weight test and it seemed to be accurate (18.71 by calculation, 18.73 in practice). This is what I hypothesized before doing it considering it seems to be accurate at low power.

I have used 3 different power meters before these (Powertap, and gen 1 and 2 of Stages) and they all read very similar, and my power curve has been on a steady, and realistic, incline over the past 4 years of riding with power, all while using 3 different poweremeters. I also use a power trainer, and the Vectors are consistently 20 watts higher using that, too. I have multiple attempts up climbs with relatively the same time and power, and now I'm doing them slower at higher power. For example, I have exact times of 9:20 up a climb with Stages at 293, and Vector 3S at 316. Obviously, there are variables at play, but I don't think 23 watts worth..

Since the two weeks I've owned these pedals I have completely re-written my power curve with numbers I have never come close to hitting, and it is the off season and I don't trust I am in shape enough to re-write my power curve on a single ride.

My 520 edge is up to date on its firmware and so are the Vectors. I'm really hoping I can get some advice on this as it has become extremely irritating.

Just an interesting observation:

I ran another session with a Tacx Vortex and the dual Vector 3. Zwift was controlling the Vortex in ERG mode using power information from the Vector 3 (via an ANT+) stick and controlling the Vortex via FE-C - or so I thought. I used my BOLT to capture power data from the Vortex (connected to the trainer via Bluetooth) and was wondering why ERG-mode felt really choppy. Turned out the BOLT was trying to set up a long steady Z2 effort (read from TrainingPeaks) while Zwift was trying to set up the chosen workout. I was in between pedaling for live to reach the respective power numbers.

I recognized the faulty setup and ended the spurious workout on the BOLT around minute 58. The resulting picture is quite interesting

Two things to observe:

(a) the Vortex takes a very, very long time to warm up (maybe)
(b) after ERG mode sorted itself out things are rather well aligned

I am wondering whether this points to some underlying problem with the ERG mode power display on cheaper trainers like the Vortex?

The original data sets can be found under: https://analyze.dcrainmaker.com/#/pu...7-e6a01a736d86

Did this update patch fixed anything related to accuracy?

No. The specifics of the referenced "irregularity" was that sometimes the balance could be calculated when it should not have been. These spurious calculation results would not make it to the head unit in most cases (because a subsequent proper calculation would overwrite the result). So the end effect to the recorded data is not likely to be visible in most cases.

The calculation that drives the total torque that is sent to the head unit, and the power values directly transmitted by Vector, are not affected by this change.

No. The specifics of the referenced "irregularity" was that sometimes the balance could be calculated when it should not have been. These spurious calculation results would not make it to the head unit in most cases (because a subsequent proper calculation would overwrite the result). So the end effect to the recorded data is not likely to be visible in most cases.

The calculation that drives the total torque that is sent to the head unit, and the power values directly transmitted by Vector, are not affected by this change.

This is interesting. I was looking at GPLama's results, and I can see often when there is an accuracy issue, the left pedal power is much higher than the right (or vice versa).

In an earlier post you talked about a clever scheme to deal with ANT+ losses, where accumulated power was transmitted, and if the Torque/cadence values goes missing, then the accumulated power could be used to determine the average power to ensure continuous power logging.

Is there a separate left and right accumulated power, or is it just one value for both? I can forsee some scenarios where signal dropouts combined with old balance calculations might explain GPlama's results.

It would be interesting if GPLama logged ANT+ messages in not only on the head unit, but via a USB dongle really close to the pedals.