A COMPARISON OF SPIDER POWER METERS

Back by popular demand, the general all-things Road forum!

Moderator: robbosmans

User avatar
kiritozhang
Posts: 101
Joined: Thu Dec 12, 2019 6:44 am
Contact:

by kiritozhang

Article published on cyclingcollege
https://cyclingcollege.com/index.php/20 ... er-meters/
By: kiritozhang  Translator: alanyu

by Weenie


Visit starbike.com Online Retailer for HighEnd cycling components
Great Prices ✓    Broad Selection ✓    Worldwide Delivery ✓

www.starbike.com



User avatar
kiritozhang
Posts: 101
Joined: Thu Dec 12, 2019 6:44 am
Contact:

by kiritozhang

I have used several spider power meters (PMs) after my test of Power2Max NGeco, so why not have a comparison of them?

Tested PMs list:

Power2Max NGeco (Shimano BCD 110 model with Rotor Aldhu cranks)

Quarq Dzero No.1 (Sram 8-bolt BCD 110 model with Red cranks)

Quarq Dzero No.2 (Sram 8-bolt BCD 110 model with Quarq cranks)

Quarq Dzero No.3 (Sram 8-bolt BCD 130 model with Quarq cranks)

Sigeyi AXO No.1 (Shimano BCD 110 model with Incolor Skypivot cranks)

Sigeyi AXO No.2 (Shimano BCD 110 model with Incolor Skypivot cranks)

Xcadey Xpower-S (Sram 3-bolt BCD 110 model with Xcadey cranks)

Pica-S No.0 (Sram 8-bolt BCD110 model with Quarq cranks)

Why not test SRM spider PM? There are two reasons: 1, SRM spider PM has been proved to be extremely stable and accurate, and it is used as a reference by many different brands. 2, the truth is: I don’t own one.
Last edited by kiritozhang on Fri Jul 10, 2020 3:43 am, edited 1 time in total.

User avatar
kiritozhang
Posts: 101
Joined: Thu Dec 12, 2019 6:44 am
Contact:

by kiritozhang

Player 1: Power2Max
Image
P2M NGeco is the entry-level product, which is a lite version of NG. It inherits the stability and accuracy, but some prime function such as L/R balance is limited and you need to pay for it.
Image
It uses a CR2450 battery and is claimed to work for 400 hours, which is the biggest difference between NGeco and NG.
Image
Since here I use the heavy Rotor Aldhu cranksets (Rotor products have never been light weight) and Shimano R8000 50-34T chainrings, the total weight reaches 803g. R8000 chainrings is around 150g and you can shave off 20-30g by using R9100 chainrings. (Translator’s note: my R8000 50-34T rings weigh 148g while R9100 50-34T rings weigh 125g.)
Player 2: Quarq
Image
Yes, it’s our old friend, Quarq. I have used Quarq PMs for several years, from the last generation Elsa to the current Dzero. The reasons why I chose Quarq are: 1, SRM was really expensive several years ago and the custom service in China is not good. It needs to be delivered back to Germany to change the battery, which means a lot of time. 2, I was and am keen on carbon cranksets, but there was no agent of Power2Max, FSA didn’t release Powerbox PMs and Easton didn’t release EC90SL cranksets either, so I have no choice other than Quarq at that time.
Image
One significant change from the old Elsa to the current Dzero is the bolt pattern. Dzero has the 8-bolt pattern, more stable, which is good for PM. The old 3-bolt pattern is an open patent, which means any brand can use it after application.
Image
Quarq insists on CR2032 battery and claims a 200 hours battery life, which of course depends on your battery quality and the environment, because the battery is self-discharging all the time when it is not used.
Image
The 170mm, 386 spindle Red Crankset combo with 50-34T red chainrings weighs 707g, which is 100g heavier than the non-PM version. The new Red AXS crankset is optimized a lot on the structure, resulting in a much better weight.
Player 3: Sigeyi AXO
Image
Here comes our new friend: the superlight carbon cranksets Skypivot from Incolor and Sigeyi AXO Spider PM combo. Incolor cooperates with two different Chinese PM brands and thus two basic choices: Sigeyi and Xcadey. You can also choose the SRM spindle version, which can be mounted with SRM spider PM as its name indicates.
Image
The spider bolt pattern is a new 4-bolt pattern. The spider is mounted onto the spindle independently, ensuring the flatness and the tolerance. The battery is the chargeable Li-ion battery and it uses the magnetic charging.

In this report we only review the AXO PM, while Skypivot still needs more time to ride and test.
Image
The 170mm, 386 spindle crankset combo with R8000 50-34T chainrings only weighs 556g, which is amazing.
Player 4: Xcadey Xpower-S
Image
Another Chinese brand released their spider PM crankset combo, which is made of alu and sells at CNY 2080 (USD 295) w/o chainring. One of its selling points is the adjustable crank length. They also have spider PM with other interfaces such as Rotor and Sram.
Image
It also uses magnetic charging and comes with a charger. The LED indicator will turn red if the battery is low.

Image
The crankset uses Sram 3-bolt pattern, which means you can switch the cranksets from other brands, such as the discontinued Sram Force 22 carbon crankset.
Image
The crank-length-adjusting function is reached by a pair of metal parts. You can change the direction of the metal parts or switch to another set to change the crank length.

I use the 170mm crankset, which has two pairs of metal parts. One pair is 172.5mm while another pair is either 170 or 175mm. A pair of metal parts weighs 18g.
Image
The crankset + FSA 50-34T combo weighs 854g, which is a big problem if you want a super light bike.

User avatar
kiritozhang
Posts: 101
Joined: Thu Dec 12, 2019 6:44 am
Contact:

by kiritozhang

Official claimed data
This paragraph is recommended to go to the original link to browse: https://cyclingcollege.com/index.php/20 ... er-meters/

P2M NGeco Quarq Dzero Sigeyi AXO Xcadey Xpower-S
Battery CR2450 easy to change CR2032 easy to change Li-ion Magnetic charging Li-ion Magnetic charging
Battery Life 400h 200h 300h 150h
Water Proof IP67 IPX7 IPX6 IPX6
Accuracy ±2% ±1.5% ±1.5% ±1.5%
Weight 158g 1258g 115g 101g
Models Rotor, Specialized, Campagnolo, TA, Easton, FSA, Praxis Works, Sram 3-bolt, Cannondale, Race Face, Shimano MTB Sram 8-bolt, Specialized, Sram 3-bolt (discontinued) Sram 3-bolt, Sram 8-bolt, Rotor, Easton, Cane Creek, Incolor, Specialized, Shimano MTB, Race Face Sram 3-bolt, Sram 8-bolt, Rotor, Praxis Works, Cane Creek, Incolor, Shimano MTB,
P2M has a best the battery life, while Quarq has the least models. Though the claimed IP level of P2M NGeco is the highest, the rubber seal ages a bit quickly, reducing its water proof. The claimed accuracy is either ±1.5% or ±2%, only a small difference, but what is the truth?

User avatar
kiritozhang
Posts: 101
Joined: Thu Dec 12, 2019 6:44 am
Contact:

by kiritozhang

PM test

My test doesn’t include the indoor trainer comparison, because usually you cannot find the design/structural defect indoor of a PM unless it is really bad designed. The indoor trainer environment is just, too stable.

The reference (baseline) in the test is a pair of SRM EXAKT PM pedals, which is complicated to mount but has an extremely high stability and accuracy after the correct install, so it is suitable to be used as a reference. Here we take P2M NGeco for example. There will be a couple of seconds difference due to the manual control of the GPS computers.

The blue line is SRM EXAKT while the yellow line is P2M NGeco in the figures below.
Image
Image
Control variates method is used in the test, and we choose 200W as our baseline, which is around FTP of most riders who has a PM. (Translator’s note: I doubt that 200W is a bit higher than reality in China) In this test I first rode on big chainring and medium cog, and then small chainring and small cog, both around 90 cadence and 200W average output, to compare the different PMs. This test aims at the structural defect which causes the error due to different chainrings.

The power curve of two PMs matches quite well with each other.
Image
Big chainring AP: SRM 209W, P2M 200W

Small chainring AP: SRM 204W, P2M 198W


SRM EXAKT reported a higher AP than P2M NGeco under both conditions, which is quite ideal, since SRM has official claimed that their spider PM is around 5% difference compared with PM pedals.)
Image
Low cadence test aims at the accuracy under high torque. Some riders are used to low cadence (grinders), which is not recommended, as your muscles get tired quicker and thus easier to hurt yourself.

The power curve still matches with each other well and SRM reported a bit higher than P2M.
AP: SRM 205W, P2M 202W.
Image
Low power test aims at the accuracy under the low torque, which reflects the situation like the slow weekend riding. The curves match well and the average power is almost the same around 90W. (89.96W vs. 90.29W)
Image
One-minute high output test, which is usual in a crit and the power can change rapidly. There is small difference between two curves but in average is similar. (Translator’s note: if we align the zero at the end, there is a delay of the P2M report, compared with SRM, which is a property of the spider PM vs. pedal PM)

AP: SRM 377W, P2M 380W.

The difference is very small, but the peak power of SRM is 653W while that of P2M is 612W.

Image
There is some deviation in the curves during the sprint test. (Translator’s note: again, the delay of the spider PM)

AP: SRM 429W, P2M 420W

Peak: SRM 826W, P2M 795W

Image
Cobble test aims at the filtering algorithm. Though the data on the GPS computer changes per second while ANT+ transfers the data once each second and BT transfers four times each second, the gauge in the PM can read and report hundreds of times in each second and these data is flitted before transferred to the GPS computer. While you ride on the cobble, the gauge reading is jumping now and then, the data received by GPS computer depend a lot on the algorithm. (Translator’s note: it also depends on the sensitivity of the cadence sensor, if your cadence is not that stable on cobble) The curves again are similar to each other, except one missing peak of P2M.

AP: SRM 108W, P2M 104W

User avatar
kiritozhang
Posts: 101
Joined: Thu Dec 12, 2019 6:44 am
Contact:

by kiritozhang

Image
Summary
Quarq, as a big brand under Sram now, can be considered as stable in quality, though there is some deviation between different individuals. The difference to SRM EXAKT of each individual is small, which can also be influenced by different chainrings used here. Moreover, Quarq officially lowered their retail price in 2020, which is one big competitiveness, though the new model can only match SRAM DUB cranksets.

Power2Max NGeco has a high value to its price, and it is very stable under different conditions, not-too-much more expensive than Chinese PMs, which is also a good choice for those who don’t trust Chinese PMs. It has a lot more models than Quarq.

Sigeyi is the elder PM brand in China and it has amazing results in the test. Two individual spider PMs are both highly consistent with SRM EXAKT in different conditions, even better than some PMs from international big brands. However, some riders have reported that there is some fluctuation or unexpected peak power, of which reason is unknown. (Translator’s note: I have met this in the earlier firmware but it seems to be resolved after 3.2) When the big brands decide to lower their price, only high quality can make Sigeyi take hold in the market.

Xcadey, as another Chinese brand, its PMs’ performance is not as good. The first tested PM failed due to the problem between the spindle and the crank while the second one is broken in the electronic hardware during the test. Only the third one can pass through all the test. In most situation it performs well actually, while there is a big deviation during the sprint. (Translator’s note: Xcadey claimed that they had solved the sprint problem by FM update)
Image
Pica, a perfect negative example. It deviates from SRM and other brands significantly. Although they have gone bankrupt, there are still some products on the second hand market, which should be aware of.

User avatar
kiritozhang
Posts: 101
Joined: Thu Dec 12, 2019 6:44 am
Contact:

by kiritozhang

Off Topic
1, There is a design defect in Xcadey crankset and they are recalling their products as of press time (June, 9, 2020)
Image
The locking structure of the crankset is confusing and stupid: there is an inclined surface and a split ring at the end of the spindle, and the crank cap works like the expander bolt to lock the crank by expanding the spindle from inside.
Image
However, the diameter of the cap is smaller than the inner diameter of the crank, so there is no enough lateral pressure on the crank. Thus, it needs to be pressed manually and if you don’t have put enough force on it, there will be a gap and thus it will be creaking all the time. Moreover, this structure is prone to be loosen while you ride on it, and finally may dismount itself.

2, Incolor have recalled their earlier products, whose threaded metal parts for pedals may be broken. I will post the review of their crankset after a while. (Translator’s note: my Skypivot is broken due to the spacer and thread part failure. Though it’s free to have a replacement after the end of June.)
Image

TobinHatesYou
Posts: 13050
Joined: Mon Jul 24, 2017 12:02 pm

by TobinHatesYou

SRAM/Quarq still make a GXP crankset for the new DUB DZero spiders.

RyanH
Moderator
Posts: 3222
Joined: Tue Jan 17, 2012 4:01 pm
Location: Los Angeles, CA
Contact:

by RyanH

Very interesting. The three Quarqs you listed, are they different units or the same and you just swapped rings?

My last Dzero came from the factory about 5% high compared to my SRM (verified calibration). I think they ship PMs with boosted slope values or their QC process/factory calibration is crap.

AnkitS
Posts: 1456
Joined: Wed Jan 09, 2019 5:03 am
Location: Santa Cruz, CA

by AnkitS

TobinHatesYou wrote:
Thu Jul 09, 2020 5:23 am
SRAM/Quarq still make a GXP crankset for the new DUB DZero spiders.
Almost impossible to get these.

User avatar
kiritozhang
Posts: 101
Joined: Thu Dec 12, 2019 6:44 am
Contact:

by kiritozhang

RyanH wrote:
Thu Jul 09, 2020 5:27 am
Very interesting. The three Quarqs you listed, are they different units or the same and you just swapped rings?

My last Dzero came from the factory about 5% high compared to my SRM (verified calibration). I think they ship PMs with boosted slope values or their QC process/factory calibration is crap.
different

kwakekeham
Posts: 94
Joined: Fri Jul 22, 2016 8:03 pm
Contact:

by kwakekeham

I normally don't reply to these anymore but someone sent it to me and I figured one more hurrah. Someones starting into the world of testing and seems they need some help fixing their process to do testing. Also to prevent even more bad info circulating.

Firstly you're comparing multiple spiders to an ASSumed correct Exakt which only reads in one axis. At 1 degree errror of installation sin(89) degrees = 99.98% force from what you want, but cos(89) = 1.7% force from the WRONG axis. So by the time you notice 1% off (which is insanely easy with the "push down" app approach) at about 82 degrees you are no imbuing up to cos(82) = upwards of 13.9% percent from the radial force to be read as torque! GASP -- thats huge . That puts Exakt worst than a Dura-Ace 9100 with a simple bendy bridge and no extra corrective elements (4iiii's, Shimano only corrective). Why is this important, torque might always average out forward BUT radial doesn't. This radial problem is the basic and background of every patent I have on powermeter tech. All 27 for 7 different patent families in on average 4 countries each. A lot are still applications, some granted.

Now, lets get another thing straight. Accuracy is marketing wank claims. There is no unified method of testing. As from the fact some advertise accuracy of 1.5% (meaning 98.5% error) and can't accurately communicate what they mean or how they test, lets question something. Now how do you calculate error between two meters. 99% of publications and papers I've read do it wrong. They do it as experimental to theoretical. WRONG WRONG WRONG WRONG WRONG. These are experimental to experimental.

Wrong => error = (tested - theory)/Theory -- Assume 1 is Right (false assumption)

Right -> error = (A - B)/(average(A,B))

Why is this important? Well in empirical science you can only assume one is theoretical if and only if its much better than the unit under testing. I just proved you could get up to 14% error with basic math on a bad SRM Exakt install. So if claimed is 1.5% and simple math and app issues can cause 14%, it certainly isn't likely this is a 0.15% error meter!

Now here is the real rub. Of the 20 - 30 peoples high speed force curves on highly corrected proprietary Internal Powermeters that were never released to public because of 10x - 20x cost, many people produce radically different pedal strokes especially in radial direction. You know, that direction that we just saw could cause about 13.9% of the radial force to look like torque. Some people produced highly positive (Me, Ray Maker, a couple of the local "big" cyclists) on a pedal stroke and a few people produced almost none (couple of employees, few semi-pro). The tri-specific people were the strangest for radial forces. The European test riders and pro-continential people were also all over the map. I've done this at two companies now, it doesn't change. Radial force is the main enemy. I won't even get started on the inferior shear gage arrangement of the Look pedal. I have a video on IQ2 on my youtube channel with FEA on it.

So this means even if you install it poorly you could have a great reference IF you are an average radial force person, OR if you did a good job installing it and Me or Ray Maker rode it, then it could look really bad.

So now that we know that, lets circle back to the Right error calc. If we have to average the two meters because we cannot definitively say one is better than the other than how dow we know which is wrong? You can't. You basically are shouting "I'm biased"

This is the 3rd person principle. You always always always need a 3rd meter for any testing to be valid at all ever. Otherwise you can just ignore the results. See the 3rd test unit basically is the "odd man out" principle. If two different technologies with similar claimed accuracy match well and one doesn't, the one that doesn't is likely the wrong one. Now you can, sort of via a transitive property test known meters seperately. It's complicated but the transitive link must be equaly as good as the one you are trying to translate. So basically to translate a Quarq to P2Max in this test you test both the Quarq against something with good agreement and the P2Max with the exact same thing. However that 3rd thing must have had consensus of being good. In this case it's the Exakt which we know factually is high error prone and susceptible installation error and rider radial force curves. It's a non-starter.

Get yourself some different pedals, get a Tacx Neo and a Powertap, and then the data might be worth something but unfortunately right now you have pretty graphs that say "srm exakt" isn't that great. That's literally all you have.

User avatar
robbosmans
Moderator
Posts: 2833
Joined: Sun Aug 26, 2018 12:40 pm
Location: Central Belgium
Contact:

by robbosmans

Good point, I ordered an Xcadey spider based on the good reviews out there, now I am a little bit worried


Verzonden vanaf mijn iPhone met Tapatalk

elzilcho
Posts: 43
Joined: Tue Jan 26, 2016 7:39 pm

by elzilcho

robbosmans wrote:
Thu Jul 09, 2020 4:38 pm
Good point, I ordered an Xcadey spider based on the good reviews out there, now I am a little bit worried


Verzonden vanaf mijn iPhone met Tapatalk
Any followup on your Xcadey?

User avatar
robbosmans
Moderator
Posts: 2833
Joined: Sun Aug 26, 2018 12:40 pm
Location: Central Belgium
Contact:

by robbosmans

elzilcho wrote:
robbosmans wrote:
Thu Jul 09, 2020 4:38 pm
Good point, I ordered an Xcadey spider based on the good reviews out there, now I am a little bit worried


Verzonden vanaf mijn iPhone met Tapatalk
Any followup on your Xcadey?
Flawless, works great and the numbers seem to match my trainer so accurate to

by Weenie


Visit starbike.com Online Retailer for HighEnd cycling components
Great Prices ✓    Broad Selection ✓    Worldwide Delivery ✓

www.starbike.com



Post Reply