With over 1.3 million Peloton connected bikes in circulation these days (and adding ~250,000 per quarter), they’ve quickly become (by far) the most common model of indoor cycling bike found in people’s homes. And as Peloton popularity soars, so does people’s desire to run non-Peloton apps on it – namely Zwift, but also TrainerRoad and others. In fact, I put together a post about this earlier this summer, showing how to install and run both Zwift and TrainerRoad on a Peloton Bike (it also works on the Peloton Bike+).
The challenge though is installing the software is only one piece of the puzzle. The second and arguably bigger issue is getting your Peloton power data to 3rd party apps. And as of today, the Peloton bike doesn’t do that. It keeps all of its data locked up inside with no standard ANT+ or Bluetooth Smart broadcasting. Further, there’s no way for 3rd party apps to control the resistance, simulating hills or intervals.
However, that all changes as of today with the launch of SHIFT on Kickstarter. Yes…Kickstarter – the same platform that Peloton themselves originally launched on many years ago.
How it works:
The first of a few different projects underway has launched on Kickstarter, this one allowing you to control your Peloton bike via Zwift (or any other app), letting it simulate the ups and downs of a course, while also getting power data into the app via standard Bluetooth Smart FTMS/Power/Cadence protocols, just like any other smart trainer.
Now, I don’t yet have a device, but have been talking pretty extensively to the SHIFT team for a number of months and they aim to get a prototype over shortly to try out. Until then, I mostly wanted to cover this in a quick post, namely as a way to funnel the 2,382 questions I’d invariably get over the next few days.
In short, the way SHIFT works is to place a removable device atop the red Peloton knob, sorta like putting a shoe upside down atop the knob. This device then figures out the calibration of your bike and is able to change the resistance on the fly based on what Zwift is telling it to do, such as simulating an 8% grade, or a 250w interval in workout mode. The device will work with both the Bike and Bike+, though it does appear to require two different pledges (despite the knobs being very similar). The difference at the knob level being the Bike+ is a true smart bike in that it can already hold and set various resistance values (again, up until now, only within the closed box that is Peloton).
In any case, again, walking through how this is supposed to work – and then I’ll circle back once I’ve got a unit in-hand. The Kickstarter is open for 44 days, so hopefully before then.
First, you’ll slide the unit atop your knob:
Then you’ll pair it up to a cadence sensor if you want cadence. It appears to only support Bluetooth cadence sensors, which is a bit of a bummer because that limits not only the cadence sensors you can use (especially if you have an older one) – but in the grand scheme of things isn’t a big deal. What’s notable here is that it WILL REBROADCAST your cadence from that sensor to Zwift (or any other app) as a cohesive single Bluetooth channel (I’ll show ya that in a second). Note, the unit has a touchscreen atop it.
Next, you’ll do a calibration procedure. Essentially they’re determining the power curve of your particular Peloton bike, relative to the number of knob twists. In my experience these two correlate well, however, the actual accuracy of a Peloton bike’s power is variable.
Meaning that while you are calibrating the SHIFT to your Peloton bike’s power, SHIFT isn’t trying to correct that accuracy of your bike. In all the testing I’ve done, the base Peloton Bike model is a bit iffy on accuracy. Peloton claims +/- 10% for that, and that seems in-range with what I see when we exclude intervals – but goes much worse when we include high-intensity intervals (which, is also in-range for what we’d see on a base non-smart trainer with a speed sensor).
Whereas, for the Bike+, I see substantially higher accuracy – crazy good in fact. For comparison, here’s two rides I did this week. At left is a ride on the base Peloton Bike on Tuesday, and at right a ride with the Peloton Bike+ today. As you can see the Peloton Bike+ is basically spot-on the accuracy of the power meter pedals for the vast majority of the ride, save a bit of drift towards the end (which, I’d need to validate isn’t the P2 pedals – hard to say here). Whereas the base/original Peloton Bike is…less than optimal.
In any case, accuracy aside – the next step is connecting the SHIFT to Zwift (or any other app). You can see below it’s broadcasting properly as a Bluetooth Smart FTMS smart trainer, a Bluetooth Smart power meter, and a Bluetooth Smart cadence sensor (all via a single profile). Separately, you pair your Bluetooth Smart HR strap. The single-channeling here is notable because it means this will work with Apple TV, which is limited to two Bluetooth device channels (+ the remote).
However again, I think a bit of an opportunity is lost here to broadcast on ANT+, which tends to be more widely used by people running Zwift (or any other app) on an older laptop or desktop computer.
So with that all set up, the red part of the SHIFT will rotate your Peloton knob to simulate hills or interval workouts. Just like any other smart trainer in effect. In fact, it has a stepper motor, just like many smart trainers do:
Here’s the insides of it:
So how do you shift? Well, via digital gearing. Basically taking the exact same concept as what the Wattbike ATOM does, which is to create a set of virtual gear steps – 22 to be precise (also, the exact same number as Wattbike does).
This is where you’ll change the gearing to go up and down through that range. The downside is that’s on this display here, so it’s a bit tougher to quickly shift through, since you don’t have buttons on the handlebars.
I could see some pretty easy workarounds here though for SHIFT. First, they could easily leverage the Garmin Edge Remote, since that just follows an ANT+ open standard, and would be trivial to implement, allowing shifts up/down. Alternatively, they could also leverage any Bluetooth Smart remote, just like a camera shutter button. There’s also a standard for that too. The Garmin ANT+ remotes are slightly more ideal because they’re weatherized for riding already, so you don’t have to worry about sweat eventually killing it. In fact, the Varia remote (another similar variant) even has handy arrows that would work perfect for up/down.
In any case, once that’s done, you just Zwift as normal.
One thing to keep in mind is that officially Peloton doesn’t permit sideloading apps onto the Peloton bike. That’s why in their screenshots above they show it running on a separate display. Still, you can do the sideloading just fine as I’ve shown – if you’re technically inclined. But to be super clear, you don’t need to do that at all, and the base SHIFT product wouldn’t violate any warranty since it doesn’t dork with anything at all on the Peloton bike, it’s just akin to putting a towel atop the red Peloton resistance knob.
Still, for those curious, while the Peloton screen makes Peloton look beautiful, it doesn’t really make Zwift look beautiful, mainly due to the limited graphics processing power of the screen/display Peloton has selected. Basically, it’s just a giant Android phone, but not a very powerful one. That’s a logical decision, because there’s no good business reason to put a bunch of processing power in something that doesn’t need it normally. So, when you do use Zwift on a Peloton bike, the graphics aren’t great. Screenshots like this make it look better than the low frame rate experience actually is:
However again, there’s no reason you need to only use the Peloton bike screen, you could easily use a secondary larger display with something like Apple TV just fine.
Now mind you, this isn’t the only Peloton to Zwift project in the works. I’m aware of a few people in this realm, and another project, DFC, in fact just sent me their unit, and it’s making it’s way across the pond as we speak. That one doesn’t control the Peloton bike, but instead broadcasts your data out as standard Bluetooth Smart power/cadence signals, enabling you to then pair it up to Zwift natively. So look for my thoughts on that one soon too!
Going Forward:
Now, given this is a Kickstarter project, it means that there’s risk. Namely, you’ll spend ~$175 and won’t get anything in return. It’s a core reason why I don’t tend to write about Kickstarter projects unless I’ve got something in hand. In this case, there’s enough interest I figured I’d give my initial thoughts, and then follow it up with a hands-on post hopefully before the project closes in 45 days. But again, as always with Kickstarter – treat it like casino money – if you win – GREAT!, if you lose…shrug.
Right now their shipping timelines are March 2021 for the base Peloton Bike model, and May 2021 for the Peloton Bike+ variant. Given they have final prototypes, I see March as viable but tight. It’ll depend on how much they’ve accounted for the Chinese New Year holiday period in early Feb (which shuts down factories for weeks), and how much other fudge factors they have built in. But it’s not overly worrying of a date. Fwiw, I have put a Kickstarter pledge in for the Bike+ variant.
Still, I really like the direction they’ve gone here for the most part. Designing something that’s completely removable without any tools is ideal as a mass market type device. And the simplicity of it is also great, as well as the rebroadcasting of the cadence signal back out over Bluetooth Smart. Still, I’d love to see ANT+ integration here on both the sensing side as well as the broadcast side.
With that – thanks for reading!
Given how keen Peloton are on law suits, might they tyy try and intervene here?
Given no part of this interacts with Peloton, it seems pretty unlikely they’d succeed here. Not to mention the act this actually promotes buying a Peloton bike – namely for families that are trying to decide between the Peloton route and a Wahoo/Tacx/Stages/Wattbike type route (which, as I’ve found recently, is a growing number of people).
There are a variety of counter actions & generally annoying things they could do if they want to try to retain absolute control over the customer experience. Whether they would be worth the effort depends on whether they had plans future plans in that space.
They could also view this in a similar manner to how Garmin viewed Backtracker and their customers would see it in their own online store along with their color coordinated shoes & HRM.
I like living in a world with this kind of thing is possible. It’s just that I’d like the world to also include my original Pebble smartwatch, Pebble Core, and Bose Sleepbuds. Backtracker makes up for most of that.
If’s a good thing they didn’t call it SHIFTR or Wahoo would be complaining. 🙂
If you get tired of pedaling you could opt for this Kickstarter project: link to clmbr.com
No Ant+ probably means a generic ble chip instead of Nordic nrf52832 to save cost.
In chatting with them after the post, they’re looking into the viability of going dual via license upgrade module. Sounds like they’re likely on the nrf52 series.
In their promo shots, I see a raspberry pi zero, Adafruit pitft, some 3d printed parts, and other likely off the shelf lego type bits (small stepper motor, motor power supply etc).
Curious to see what their final production BOM would include.
One would think Peloton would create greater stickiness to their ‘platform’ by enabling or allowing 3rd parties to licence access to it. Additionally, they’d also create a greater ecosystem by allowing integration of their services onto 3rd party bikes (e.g. Keiser, Schwinn et al who often times produce a MUCH better quality spin bike than Peloton does). But this would involve an openness that they don’t seem to have. They’re wanting to be the next Apple, though in that case Apple looks about to launch their own Fitness service and with the behemoth they are, well, Peloton better watch out.
Peloton makes money off the subscription.
I would bet the bike is sold at close to cost, considering support, licenses, and R&D
But competitor spin bikes of similar quality is so much cheaper.
They’ve previously said on earnings calls that the Peloton bike as a standalone piece of hardware is quite profitable, and their financial statements (assuming I’m reading them correctly), seem to support that. For the last quarter:
Connected fitness products:
Revenue: $601.4M
Cost of revenue: $364.2M
Note: Subscription revenue is another line item:
Revenue: $156.6m
Cost of revenue: $65m
*Page 18: link to investor.onepeloton.com
(Other costs/revenue buckets are itemized as well)
Those numbers above for hardware match what I’d expect. Having taken one apart to every last piece, the Peloton Bike is actually a very simple piece of hardware from a cost and manufacturing standpoint. It’s basically just a big metal frame (exact same frame on Bike/Bike+), then an Android tablet display tossed atop it. They run two wires down through the bike for power on one, and to the flywheel on the other. The Bike+ also includes a secondary stepper motor system for engaging the flywheel resistance motor, but that too is designed so that it’s basically just a slightly different piece but otherwise fits onto the same Peloton Bike frame. Their efficiencies on this hardware are kinda incredible. Similar to Apple and Garmin in terms of reuse of hardware/components to minimize cost/manufacturing churn.
None of this is expensive to produce. Expensive to ship – yes, but not expensive or complicated to make.
Stephen wrote:
“Peloton makes money off the subscription.
I would bet the bike is sold at close to cost, considering support, licenses, and R&D”
I’d take the other side of that bet. Peloton was originally selling the bike for half of the current MSRP but it wasn’t a hot seller. They UPPED the price “to add value” and lo, it started selling. (I vaguely recall DCR mentioning this somewhere…?)
The Peloton (not “plus” version) is a generic, sub-$1000 spin bike with a cheap screen attached. (see “Bowflex EX1 or C6”)
Would be nice if they can get this to work with a normal (dumb) spin bike as well.
Totaly agree, what is the specifics that sticks to peloton bikes ? À spinning Bike looks close… would really be nice ?
I would think you need a power source…most likely pedals.
Still should be a lot less costly for pedals/bike/Shift if one does not already have a regular bike. Then a Snap/M2 would be the way to go.
I must be missing something, but why on earth would anybody buy a Peloton bike to then use it with Zwift? If somebody plans to use Zwift, that person has obviously figured out that it’s cheaper and better than the Peloton software. So I would assume that this same person also knows that any hardware is better than a Peloton bike…
“I must be missing something, but why on earth would anybody buy a Peloton bike to then use it with Zwift?”
It’s super common, and namely is budget based. Essentially, a family decides they want an indoor bike of some sort, but perhaps one person wants more of a KICKR Bike, and the other wants more of a Peloton bike. Finding middle ground is hard.
As for better hardware than a Peloton bike – also depends. Frankly, the accuracy & power stability on the Peloton Bike+ (Plus) is putting to shame some of the smart bikes I’ve been testing.
I think you nailed the analysis on that Ray.
There’s a gap right now: the difference in price between a full-on smart-bike ($5000CAD) and a smart trainer ($1500CAD) with a cheap to decent bike frame ($500? $1000?) attached.
Oddly, a Peloton bike seems to be one of the few things in the middle. (and I say this as someone who never had any interest in buying a Peloton bike)
I’m still holding out hope for a “Smart Bike Lite” from Wahoo, Tacx etc.
That’s pretty neat. If they ever support bikes other than the Peloton like the IC4/IC8/C6 we currently have, I’d pay $200 for this without a second thought.
Would it be possible to use a set of power meter pedals to provide the power data and using SHIFT just to control the resistance?
Yes, they discuss this a bit in their Kickstarter. However, it’s offloading that to Zwift. Nothing is wrong with that per se, but it’d be slightly better if they did calibration against the PM for that power curve. That’s something they could add in software down the road.
Any reason the calibration couldn’t be done against a PM, e.g. Vector 3? Obviously would prefer a direct connection over calibration but seems like an improvement over the base Peloton.
Any chance we’ll get to see the Peloton review(s?) soon?
Always a chance!
Actually, it’s why I’ve started riding Peloton again. Just want to get it knocked out now that things are a bit quieter.
Awesome! You, sir, are both a gentleman and a scholar. Have been looking forward to those reviews.
hope the cost is correct.. vs just buying power pedals.
@DCRAINMAKER how does the one sided pedal power work for this interaction??
The cost is correct, you can already buy it (it won’t ship yet, but you can place the order).
There’s no need for power meter pedals in this case. A single-sided power meter pedal will probably be more accurate for the base Peloton bike, but probably not as accurate for the higher end Bike+ would be.
If this works it would be a much better overall zwift experience than just throwing power pedals on
I for one, am very excited about this project and have already backed it. I originally got into cycling with the peloton, but as my interests have become more triathlon related I have been more excited about apps like zwift and such, which can offer longer rides. While I still do very much enjoy the peloton, only their power zone rides naturally fit into the training. This device saves someone like me a significant sum of money and space (not having to set up a smart trainer next to peloton) to train. The accuracy issues, while bothersome, do not actually change the paradigm so much because I generally find that you should train against yourself…
Still, they seem to have a lofty funding goal. I’m not sure enough peloton people are eager to jump onto another subscription.
I am right there with you on the reasoning and am concerned the funding goal seems very high. Hopefully they market it like crazy for people to see.
I really don’t understand how bike + trainer + Peloton streaming service isn’t better (less expensive, more flexible) than a Peleton bike.
Because Peloton doesn’t let you get power (or resistance) data into their platform without their bike.
Will the peloton + variant be backwards compatible to the gen1 bike..? That would be cool especially if there’s was ANT+, then I imagine there could folks who have a bike at home travel and use the device (when world is normal) at hotels etc (assuming they have a dongle/laptop or iPad) to run zwift and pair with the bike.
It sounds like the goal is to have them backwards compatible. I think they’re going to update a bit there soon. The unit was designed as modular, and the only piece that’s different is the little component that ‘cups’ the Peloton knob. I asked how that would work if someone bought a regular one, and one day wanted to upgrade to the Peloton+ (or vice versa), and that was core to their design that the cup is swappable to other units (in case there’s a Peloton+++ or whatever).
Does the resistance control on the bike+ still work when the display has zwift loaded onto it and is running that program?
In other words, if you have a bike+, load zwift on it, and then turn zwift on, will turning the knob still change resistance?
I’m unsure since it is no longer a direct physical connection (screw knob) and instead an electronic one (knob rotates some sort of rotary encoder (?) that sends signal to bike+ electronics which then control the motor that acts upon the magnetic resistance unit).
Zwifting on the stock display tablet is not ideal of course, for the reasons you mentioned. Just curious as to what the setup and usage would be like.
Great post, as always. Here’s my favorite sentence from this one: “First, you’ll slide the unit atop your knob.”
🙂
Hi Ray, have you noticed anything strange with the speed data coming out of your Peloton rides? My data on the bike+ seems to oscillate constantly while power and cadence are constant on a steady state ride, don’t think it was like that on the standard bike (which we’ve upgraded from). thanks
I backed this, but I can’t see how they will get to their funding goal. Hope I’m wrong, but selling 1200 or so seems a bit optimistic.
Wow does that look familiar….
link to github.com
For what its worth, this project already fits on a wide variety of bikes too.
HMC Magnetic Bike
Schwinn Felt resistance
Wow this looks awesome, great that you figured it out, don’t have much experience think this would be easy to make for an amateur. Checked out your link didn’t seem too bad.
It’s not too bad but I do not have any build reports from units in the wild. Tracking my github stats, I suspect there may be a few but it could also just be Chinese companies investigating if there’s anything to mine.
I’m currently working on an easier to build yourself “kit” for people that want to experiment but I don’t have production plans. My research indicated that once liabilities were taken into account, the profits weren’t worth the risk for a commercial venture, hence the open source on github.
It’s been in design/test for over a year now, but I’m only one guy working on it for fun in my spare time, so it’s been a slow go. As they say, hardware is hard. Every time the “very last issue” is fixed, two more pop up. That said, all of the core functionality is stable. It works reliably when paired to Zwift as a “smart trainer” with Zwift using sim mode. The current issues (hopefully solved soon) are random crashes (~once an hour) when using the unit as a bluetooth bridge. (using it to compile bluetooth data from multiple other sensors and present that as one bluetooth device to Zwift). It would be easy to drop that functionality all together, but running power meter data through the unit is what enables ERG mode.
I’d really appreciate it if some software guys came along that would like to help do some polishing. If anyone with that talent has some spare time and reads this, I’d love to hear from you!
Exactly what I wanted ! Would love to do it ! Will try to do it 🙂
Exactly what I wanted ! Would love to do it ! Will try to do it 🙂
Ray, I assume that that with this device, one could also use it to broadcast power and cadence to TrainerRoad side loaded onto the Peloton – and then avoid needing a power meter pedal or crank? If not, what am I missing?
New to the whole smart trainer scene, but we recently got a peloton bike+ which takes up all available space in the apartment for anything else.
Would the SHIFT + the mentioned project from DFC used together be the ideal auto-resistance & power+cadence combo? Or is the DFC project power output essentially the same as what the SHIFT would provide to zwift and therefore you would be better off with a cadence sensor only?
Wonder if he got the product yet to test? I got an email from them that they could be opening for purchase soon. If I could use with plugging into iPad with Zwift then perfect.
Happy New Years from the makers of DFC!
2021 is an exciting year for us. The new pre-production board is in fabrication this week. With it comes new features that will promote the use of DFC beyond just a bike data transmitter. The updated design will make the hardware open to the community to add features and expand its capabilities over time. We’ll have more details on this at launch.
Pre-orders will start in the coming weeks. We can’t wait for you to use DFC on your future rides!
So the DFC device mentioned in this article that you were testing recently looks an awful lot like a Raspberry Pi + Gymnasticon open source project solution. I have that at home and it works, even my Karoo2 picks it up. This is the project:
link to github.com
I used the released image with my Raspberry Pi 4 and obtained the necessary daisy chain of cables to both intercept the power/cadence data as well as feed it back to the tablet and convert serial to USB port on my rPi 4. Here is the original pull request for the Peloton support:
link to github.com
I’ve since sold my Peloton and used the money to order a Wahoo KICKR v5. Might find a use for the Pi someway. It’s really cool to be able to broadcast the BT LE signals. I imagine it might be possible to plug the Wahoo ANT+ USB adapter into it and write some software to radiate the BT LE signals. But I’m unclear if Wahoo’s USB ANT+ adapter is receive only or can also transmit.
Sean