I got a new trainer! Yay! At least, new to me. It's sorta new in that I got it from a store, but it was covered with dust and marked waaaaaay down and labeled 'no returns', but I GOT A NEW TRAINER! WOOHOO!
I've been on a wheel-on trainer since 2015, and while it still works just fine, it's definitely showing signs of age and wear. It's noisy as all get-out, and that is not helped by the different tire compounds Alastair and I have on our rear wheels (seriously: the first 5 minutes sound like someone is using 10-year old windshield wipers on dry glass). And that brings up another challenge: Alastair and I share it. So to do a race with him I have to move over to an even older dumb trainer and use plain ol' fluid resistance and a power meter. It's not ideal. Oh and the 2015 Kickr Snap has somewhere in the neighborhood of 15K miles on it, so the drum's worn down with a nice deep groove that really makes me question its power accuracy.
So anyway I'd been targeting a new trainer for a while, a wheel-off (or direct-drive, as the cool kids say) unit with greater accuracy and reduced noise and improved "road-feel" and, well, fewer YEARS on its clock. And then the pandemic put that desire in a choke-hold, because everyone who could actually afford a new trainer bought one, and deals evaporated overnight.
But I had a plan.
I'd created a short list of trainers that I found to be accurate and quiet and affordable enough to justify the expense, and I'd check local inventory at my favorite bike shops, because sometimes they actually have trainers in stock that aren't the top-of-the-line units.
My list comprised the following options:
- Wahoo Kickr Core at $899, +/-2% accuracy, 2000W capability, ballin' on a budget
- Elite Direto XR at $949, +/-1.5% accuracy, 2300W capability, cassette included (one less thing to buy!)
- Wahoo Kickr v5 at $1199, +/-1% accuracy, 2200W capability, the gold standard
Let's be clear: I was only going to consider the Kickr v5 if there were literally no other options and the store accepted payments in kidneys. But it made the list because Wahoo's products are pretty future-proof and enable some of the best integrations available in the trainer space. They're dorky, but nerd-cool dorky.
And before I went out the door to my local bike shops (LBS's to the cool...you get it), I figured I'd check stock online because ew yuck people gross. Yeah: still a pandemic going on, still no stock of the "affordable" options (I don't even mention some of the more expensive options, like the ones with built-in vibration, because apparently sitting on a trainer for hours isn't torture enough--you need it to vibrate!), so on Saturday, Alastair and I headed out with masks to do a quick round of bike-shop visits.
We never made it.
On a lark, I figured it might be worth dropping in to REI. They're kinda notorious for having wild deals on stuff and really not knowing their inventory very well. We'd been in the store for 5 minutes, and after asking if they had any trainers in stock, were ready to pull the plug when the dude showed us to the "trainer things" that were just front-wheel riser blocks (c.f. above about their knowledge of inventory) when *another* dude said he thought there was something bike-related in the "garage sale" area.
Do you want to guess what it was?
Well it wasn't anything on the list above, but it was a trainer with no box and a "no returns" tag. Specifically it was a CycleOps Hammer H2 with an accessory bag, an 8-speed cassette mounted, and a tag that further elaborated "used once". With that 8-speed cassette, I'm not surprised. Blech.
Oh, also? It was less than $500. Get the car, kiddo.
Now the deal with "no returns" is that, obviously, it's a gamble. And a big one, since this is a unit from almost 3 years ago that, while it still has its warranty card in the bag, mmmmmaybe won't be honored. But saving many hundreds of dollars wasn't exactly my plan, and you may know that I'm somewhat technically savvy, so OH HELL YES WE ROLLED THIE DICE.
Got it home with the quickness and immediately ditched the cassette, pulling a spare 11-speed 11-28 from an unused cyclocross wheelset (more on that in the next post!). Got the unit paired quickly and went for a quick...70-mile Zwift gran fondo. Nothing like a trial-by-fire.
That's kinda short-selling the experience. I did have some initial issues pairing the unit, so I rode the event pulling power from a separate power meter and used BLE to control the trainer's resistance. But OMG it was amazing! So quiet! So smooth! So much the better ride that I almost forgot I was on a trainer! Not even exaggerating: it was that much better.
But I knew it could be even better, because there's no reason I should have to use a separate power meter. A quick check of the H2's specs shows that it's supposed to be +/-2% accurate on power and somehow calculates cadence. So I tried it using all its onboard goodies, and...the numbers felt weird. I felt like I was working really hard, but the power numbers were low.
And do you know what that means? It means it's time for science! Which I did today! Because science!
I happen to own two bikes with onboard power meters. One is my primary road bike, a 2011 Blue Axino with a Stages left-side-only crank-based power meter. Let's call it a Stages "Gen 1", since there's no generational data on it. It's old, is kinda my point. And it's gotten wonky before. In 2017 I was seeing some of the most amazing power numbers of my life until I did a mid-summer calibration and saw an almost 30% drop. Booooooo. More recently it just stopped doing nice stuff altogether until I did a factory reset on it, but it's been my primary power source for Zwift racing and outdoor riding December 2015.
I also have a TT bike, a 2016 Swift Neurogen, with a Quarq D-Zero crank-based power meter. The Quarq requires far less care & feeding, includes fancy bits like temperature compensation (you don't have to calibrate it because it got cold outside) and left/right power offset (which leg is stronger?). I don't often use it on Zwift because with the wheel-on trainer the disc wheel is horribly loud, the TT position is weird and makes me look up up up at the monitor, and the front end of that bike is super wobbly (are all TT bikes like that??).
So back to the science! Today I picked a route on Zwift and rode it twice: once with each physical bike that has an onboard power meter. Each ride had the same settings in Zwift, from "trainer effect" to chosen in-game bike. Each ride was also dual-recorded, with Zwift capturing the CycleOps H2 power and cadence data over ANT+, and my Garmin Edge 520 recording the bike's power meter and cadence. Heart-rate data was recorded from the same Wahoo Tickr for each and would serve as the anchor for data overlays. Firmware was updated for the H2 to latest (I didn't catch the version number).
The goal for both rides was a general Zone 2 effort, with a 300-ish watt climb in the middle and an unspecified 15-minute effort through rolling hills to get SIM mode doing its thing. In both cases, at 14:30 into each ride I stopped, calibrated the onboard power meter, and performed a spin-down calibration on the trainer. Temperature in the riding space was maintained at 69F throughout.
With both rides done, I pulled the .fit files from Strava for the Garmin recordings and uploaded them into data analysis sets at zwiftpower.com, and the results are amazing.
First up, the TT bike: Quarq D-Zero vs CycleOps H2
Ok, 5%-ish deviation across the ride. Not too terrible. Kinda on the limit of acceptable, but let's look at pre- and post- calibration.
Well that's just ridiculous. It was within 2.x% before calibration, and drifted apart after. That's not how "calibration" is supposed to work!
Oh that is not better. Oh yikes. Overall an 8% deviation between the two power sources. Generally speaking, +/-3% is considered good, while +/-5% is acceptable. Anything outside that is...a problem. The trainer seems to think I am doing less work. But wait! That big double-gap in the first 1/3 is the calibration, and it definitely looks worse before that than after! And it is: 11.6% vs 7.1%, but still way out of acceptable limits for racing in such a competitive community as Zwift.
So somehow calibration hurt the TT bike and helped the road bike. To see how that trend aligns with the overall results for the ride, let's take a look at critical power curves for both. Now bear in mind that the calibration was performed at ~15 minutes, and all efforts leading up to that had been Z2, so nothing on this curve shoud really reflect anything pre-calibration.
Again, first the TT bike:
Oh now that's neat: it seems the Quarq evidently can't handle burst power very well (not that it was a lot of burst power, but bear with me), but tracked fairly accurately on shorter intervals, deviating pretty substantially with longer duration efforts. In fact it seems the ONLY thing the Quarq and Hammer agree on is efforts ~1m or less.
But then there's this, from the road bike:
The numbers are pretty amazingly similar, even if the graph looks a little different. TBH I'm probably prouder of the fact that I was able to turn in such similar rides than anything else, but we're here to compare technology and not marvel at my consistency (seriously, though: those are pretty consistent!).
So what's the takeaway? The new trainer is amazing, but it under-reports power. I have 2 power meters that agree far more closely with each other than they do with the trainer (except for 1s, they're within 3% all the way down, even factoring their differences and using the comparative values from the trainer to extrapolate what they "would have been" in an exact-match ride). So it seems that I'll be relying on onboard power meters in Zwift races for the foreseeable future. And it's a bummer, because I was hoping Alastair could use the trainer with his non-PM bike to improve the accuracy of his rides, but it seems to be no more accurate than the old Kickr Snap, even if it is hella quieter, smoother, and nicer to ride.
No comments:
Post a Comment