How a Powertap hub works and why nobody talks about it

[@victorlwd]

dammn keith another great video as always i keep learning about stuff i never thought i would be intrested in btw you video's and voice are seriously like ASMR to me super relaxing hahaha keep up the good work buddy and count me in for the fan controller units seriously stoked about those

[@ArashFallah]

Keith these videos are awesome! Also smart fan controller is quite ingenious, even though I don't ride on Zwift or inside. Keep up the good work buddy.

[@ChrisCapoccia]

I've had mine for several years now. Had no idea about the potential indoor accuracy issues. I have a fluid trainer (JetBlack Z1). I let the wheel spin free and re-zero before I tighten up onto the roller each ride. I don't think that it's ever had a chance to slow down to something resembling zero torque during my trainer rides. and my indoor FTP tests on Zwift are actually slightly higher than anything I've gotten outdoors. The maintenance issues I've had are 1) bent rims turn into a wheel rebuild; 2) I broke the cap once trying to replace the battery and had to order a replacement

[@robertchung4914]

Very nice explanation, Keith. I like PTs for aero testing, even with the 1.26 second updates: you don't have to worry about drive train losses, the speed sensor is built-in, and I don't really care about cadence precision very much (I can back-calculate it from speed and gear ratio if I need to). NP and IF and TSS are slightly, but not horrendously, affected by longer time averaging; what they're more affected by is data sampling (early versions of head units for PTs and SRMs and the old Polar S710 chain tension meters used to lengthen recording time by systematic sampling of the data packets so, for example the PT doubled recording time by only keeping every other 1.26-second record and ignoring the other. SRM head units up until (I think?) powercontrol IV used to do a similar thing. That kind of sampling is much worse for NP/IF/TSS calculation -- but then, like cadence, I don't much care about precision with them.

[@kwakeham]

Can't really speak to the SRM point (I had a V and VI but never a IV in my possession). You're not the first but maybe about the 3rd or 4th person who's extolled the virtues of the powertap hub in relationship to the aerodynamics. I haven't looked into the why, but several companies doing pitot tubes like them. I don't know what, weather it's the multiple rotations giving a nice average between transmissions, or something inherent in their mechanical, aero companies like them more than even spiders.

[@robertchung4914]

Well, when we first starting looking at the performance of the Track Aero System on velodromes, people were bringing in their own power meters. The TAS turns out to be pretty demanding of good data quality so it became clear pretty quickly that some power meters were kinda crappy for this purpose but PTs were pretty robust -- and we didn't have to worry about drivetrain losses. I had theories about why this was the case but after a while I just recommended that people not use certain power meters for field testing aero drag if they were trying to nail down small differences. I'll sometimes help people if they have an interesting puzzle about their field tests, but my first two questions are almost always "how are you measuring speed, and what are you using to measure power" and if they don't have a dedicated speed sensor and/or a PM that produces good data, then their puzzle isn't that interesting to me.

[@allezsport11]

@@robertchung4914 could you be a star and recommend the power tap hub over a 4iiii precision left crank arm power meter please? I know you are incredibly well respected, if josh from silca says you know your stuff then 😊

[@robertchung4914]

@@allezsport11 First off, it was nice for Josh to say that but I'm just a guy who rides a bike. Second, everything is judged on purpose for task, and some tasks are well met with a left-only crank arm. Third, SRAM/Quarq have discontinued the PT hub, for which I am very sad. I started looking at the used market and last week I bought a used PT as a spare, but it's hard for me to recommend that over a device that has a future.

[@allezsport11]

@@robertchung4914 wow thanks for the reply! I do agree, and there is the connivence factor (swapping training, deep section and disc wheel) too. I’ve gone all in and gone for 4iiii. The only power tap I could find was 2nd hand and I would want to have it built into my wheels which is of course extra cost! All the best.

[@florentlonguet9149]

Hi, thanks for your video, do you know where I can find any Technical Specs of thispowertap hub ?

[@kwakeham]

Nowadays that'd be a mystery. Since SRAM acquired them and axed all the products over the years I don't even know where any of this stuff is available. Wayback machine maybe?

[@ARGENLABSRL]

Hi my power tap g6 consumes the battery in a week . Cna you me what tehe problem could be ? thank you

[@3eyedfishy]

You've shown the torque tube and hub shell as a single piece but CycleOps use to offer a replacement/repair service for the torque tube. So.. if those are two pieces how is the other end of the tube fixed in place to prevent errors due to movement? If they're a single piece then I suppose R/R involved receiving basically a whole hew hub back with your old freehub and axle?

[@kwakeham]

It's a simplification. It's more about highlighting the bearing that supports the torque tube so I could highlight that one pain point with the product which I've unfortunately experienced on every powertap. Last I checked it was a send in repair service. If I got that wrong let me know.

[@luigibirillo6629]

Correct me if I'm wrong. Direct drive trainers don't have strain gauges. Is this because of Powertap patents?

[@iamvillainmo]

Probably cost. I think when you have a standardized unit on the market, you can develop a pretty accurate power curve algorithm based simply on the RPM of the drive unit (like the kinetic inride)

[@whynotride327]

How do I turn the auto-zero on or off?

[@deadenear]

what model uses the 2 n style 1.5 volt batteries. ( battery looks like half a AAA )

[@carlosflanders518]

I thought powertap hub sampled at 60 hz and updated every 1s? The 1.24s interval was only for the older wired models. Am I wrong? Why is updating more often an advantage. Any reading average under 3 s is pretty meaningless on its own, I would have thought. The other advantage to powertap is that its speed update is via a mechanical relay. Calculating power should be more accurate than most crank-based PMs because the power=torque*rpm equation updates roughly 2* as often as for a crank PM (most update 1* per rev) and many crank pms use an accelerometer for rpm - which is never as accurate as mechanical. Never had a problem changing the bearings the couple of times I've done it. NDS bearing appears to be much more prone to wear because it's more open to contaminants. I've had 4 ptap hubs, 2 wired (1 new, 1 used) and 2 wireless (Ptap GS new and SL2.4+ used) and they've all been super reliable. In comparison testing they've been within 1% of each other. I think the beauty of the PT hub is that does a great job of isolation from crosstalk/parasitic forces, the very simple and effective auto-zero algorithm that doesn't have to worry about temps and the more accurate and frequent rpm measurement. The SL2.4+ freehub indeed suffered very badly from cassettes wearing through the splines. Had to select cassettes with full riveting. SRAM cassettes fit better than shimano and are less prone to spline punch through. Main problem I've had with the hubs has been poor design of the axle setup. Endcaps/washers continually coming loose and requiring adjustment. They've eventually got around to fixing this on later models but took them way too long.

[@kwakeham]

Internal sampling rate is averaged out. On ant+ it updates every basic power page at about 1.25s. This continues today with the latest g3 series. Keep in mind cranks update on a revolution so that's about once every 0.75s at 90rpm

[@carlosflanders518]

Still confused as to why updating more often is an advantage as long as readings are averaged. Wouldn't more accurate and more frequent rpm readings be more important and lead to better data? Hub-based rpm based on speed is inherently much less variable than cadence.

[@kwakeham]

Relative variance is similar if you're always producing some level of torque due to the momentum of the bike. Right now we don't have a use case for high speed data, and when we do spiders and hubs will not likely be the candidate for it's use. Leg muscle analysis will be but that will be on the independent leg measuring meters. For most people it won't be a problem, but when doing powermeter comparisons it means that there is only 80% the data of other meters. For an hour long ride I'll have 3600 data points with a Quarq, and only 2880 for a powertap where some of them are double recorded This means that shorter peaks don't match as well, and can cause skew of calculations like TSS and NP (though Energy correct recording has fixed this enough now, but early and low end cycle computers your TSS and NP Will not match as well when two head units record the same meter by potentially up to 5%). So the net effect of the problem is minimized now. As a data nerd and techy, if you record higher resolution you can always make lower res data easily. If you have low res data you can't get high res.

[@carlosflanders518]

Cadence can easily vary from 90 to zero in one second. Speed variation has got to be way way less, unless you're braking when you wouldn't be pedalling anyway. Lots of half pedal stroke data going unrecorded that should be picked up by a hub. TSS and NP rely on a 30s smoothing so can't really see extra data points with larger variances being any extra use. Just a fan of mechanical rpm measurements and measuring rpm more often than 1 pedal stroke.

[@tomgilbey3206]

@@kwakeham Keith, can you briefly describe the "Energy Correct Recording" you refer to?

[@phxrsx]

I'm interested in the fan controller.

[@kwakeham]

Well I hope you watched to the end and saw the sneak peek. More news soon. Very soon. It's functional and usable but there are only about 4 of these yet. Next two weeks is test test test. Code Code Code. AND then prep the mailer and order parts for 50 - 100 pcs.

[@aaronwhiting7725]

I didn’t hear you address the 2nd Part of your title - “why nobody talks about it”. Have they gone out of vogue, or are they no longer competitive in some way?

[@kwakeham]

Basically not much goes wrong in terms of mechanical (except water ingress and the bearing issue). Therefore very little testing and little controversy about it -- basically it's simple enough to have no inherent accuracy flaws. Now in terms of popularity it wasn't my intention to discuss. I suspect a lot of it is shipping and stocking. Wells are more personal than cranks, almost as much as pedals -- just more variety. The weight, the shipping, the changing of wheels have all been contributors to their slow decline. Convenience is king which is why pedals have overtaken just about everything.

[@deadenear]

@@kwakeham bingo. The weight is why they are gone.

[@rogerbravacaster326]

I use a shimano 11v cassette with campagnolo 11v shifters, and works 100% perfect.

[@kwakeham]

Having watched the high speed tests for this, I can assure you it's not... unless you're on EPS. EPS sets the distance on cogs and assume constant which the Shimano is. HOWEVER if your on a mechanical Campagnolo the spacing is off 0.11mm per cog which results in a whole cassette sweep error of 1.21mm, however if you adjust mid cassette this means its only off about 0.6mm. This is detectable by pro's but most people, like yourself probably find it fine. Perfect is subjective when riding. When in a lab where a robot shifts it tens of thousands of times counting misshifts, that 0.6mm makes a huge difference. Maybe not to you but an elite where an extra 100ms of lost torque transmission is important. EDIT: Those numbers are wrong! it's only 0.07, so that makes it 0.77mm or 0.38 off. Which is, as you mentioned nor perceivable by most.

[@rogerbravacaster326]

@@kwakeham I use the normal cable Super record shifters and campagnolo rear derailleur.

[@kirylkiryienka6910]

It's good for You. But there are people who don't know where to spend their money, so they need this products)

[@kirylkiryienka6910]

Have You heard about Shimano?

[@kwakeham]

In what regard? I have a whole video were I FEA their crank design to show strain gage placement flaws in their 12 gage.

[@davethegayguy420]

I cannot get the iphone to recognize the Powertap, and I cannot find a Powertap App, any hints Keith?

[@robinseibel7540]

Quarq shows Campagnolo free hubs for PT hubs in stock.

[@kwakeham]

That's good for Americans. But when I was in the EU it was painful a year ago, and in Canada it still is. Quarq.com is effectively a non starter because it doubles the price due to shipping. Quarq.com can be a shipping death sentence to Canada. The 49 dollar hub is 56.92 USD to ship to Canada. Depending on regional taxes (5% - 15%). Luckily that's expedited UPS. If it was ups standard that would be 16.75 CAD just to process the tax, oh and then that is taxed! So IF UPS process it correctly (In Canada that's a coin flip) that hub will be 152-160 CAD or 116-120 USD min! IF UPS makes the mistakes they sometimes do, there is no recourse, and can be about 200 CAD or 150 USD for a 50 dollar hub. Le sigh.

[@robinseibel7540]

It's too bad the free hub isn't some other manufacturer's design, one that could be purchased outside of the Quarq/PT pipeline.

[@kwakeham]

OR somehow we could sort out cheaper shipping in Canada. It's really big with only 2/3 the population of California.

[@robinseibel7540]

I don't know if that can't happen after the Canadian Bacon incident of 1995.

[@iMoriMori]

Think PT use novatec freehubs? Recently changed my 10s to 11s shimano. Though a weightweenies post from 6 years back says the campy are not up to spec.

[@jaro6985]

I always thought they were shunned due to not capturing drive train losses, which should cause a 2-4% inaccuracy. But this page is showing good agreement: ridefar.info/2018/01/power-meter-comparison-stages-crank-versus-powertap-hub/ Not sure if they are making compensation given the average drive chain/speed/etc. or if that user just got lucky. Now we just need someone to do a powertap cassette, chain, and that will cover about everything..

[@kwakeham]

I think you're phrasing might be off here. A powermeter doesn't capture drive train loss. It's measuring either before or after. This comparison though resulting in numbers in the ballpark is using a method of measuring half the legs used before drive train loss and comparing to something that actually measures total power after drivetrain loss. The usefulness of this is suspect even if it aligns with expected behaviour.

[@jaro6985]

@@kwakeham Yes before or after is a much better way to phrase it. I agree its not conclusive data, I found another comparison here which is better: ggtri.wordpress.com/stuff/power-meter-comparison-pt-p1-pedals-pt-hub-stages/ about 4-6% as expected from drive train loss, so no compensation. Its just something that your average "uninformed" buyer might see 1.5% accuracy and not consider the extra 4-6% on top of that that someone experienced as yourself accounts for. Also if you are a zwift competitor then that is an instant 5-6% disadvantage over anyone using a pre meter.

[@johntrussell7228]

These things are garbage. Everyone I know that has had one got rid of it after a year.

[@kwakeham]

I think every brand has had people who have had bad luck. Sounds like your experience sucks. I've heard of people with dead / water ingress issues. I haven't so that's why I reserved my analysis to the setup and the bearing issue.