Patrick “Tree” Miller is a tinkerer by nature, and his latest project caused so much pause over here that we had to dispatch Santa Cruz-based photographer Josh Becker to document the madness. Let’s check out Tree’s “Kroozr Brake” bike, a custom Retrotec that features a kickback-actuated rear disc brake!

The Kroozr Brake

A kickback-actuated mechanical disc brake. It offers all the power and consistency of a modern mechanical disc brake, with all the unadulterated joy of a traditional coaster brake.

Why Even Bother?

Honestly, this project was more of a “why not?” situation. This brake won’t make you faster, it won’t give you better bike handling fundamentals, and its service interval is a bit demanding. But crucially, it offers significant mass reduction compared to a clunky, traditional coaster brake – though it’s certainly no featherweight next to a standard hand brake. It also allows the rider to use any shifting system they choose. My current setup is for SRAM 12-speed or singlespeed.

The impetus behind this project was purely personal fulfillment, coupled with a desire to integrate modern Commercial Off-The-Shelf (COTS) technologies into a novel, enjoyable application. Back in college, I used to race in an international underground coaster brake bike league, the Kroozr $kid Nation, even participating in things like the Coaster Brake Challenge in Los Angeles. I loved that community, and as I got older, it faded away. So, this is my last-ditch effort to try to keep myself whole. I call it the Kroozr Brake as an homage to my community and the original brake itself.

A Note on Safety (Seriously)

Consider this an open-source manual to create a pretty unsafe brake; I absolutely do not recommend anyone use it. If my fourteen design iterations have revealed anything, it’s that this system exhibits multiple failure modes and demands a rigorous maintenance regimen to sustain optimal performance. But, if there are any coaster nerds out there, please feel free to learn from my mistakes. And if there are any adaptive cyclists looking for unique solutions, use this freely to create the bike of your dreams. That said, I take no responsibility for injury or death that may occur due to the use of this brake or technology.

This type of brake has been re-imagined a few times over the years. I’ve had some great exchanges with different framebuilders, each one recounting their own quest for a new coaster brake. This isn’t a new concept, but merely a novel system architecture or component integration. I’m sure others have prototyped similar ideas, and I want to celebrate them too! So please, feel free to share your designs.

Before we dive into the learning part of this paper, I want to extend a massive thank you to Hudski Bikes and their advanced R&D team, who significantly helped advance this project. Seriously, buy a Hudski; they make great bikes. And thank you to Curtis Inglis of Retrotec, a local framebuilder who helped me make dreams come true for a second time! Also, a huge thanks to Ethan at Heshel Machine Shop of Oakland for cutting so many parts for me and offering invaluable advice.

Riding the Kroozr Brake

After approximately 100 hours of operational validation, the system exhibits sufficient functional stability. The current configuration permits limited reverse pedal rotation for pedal-kick maneuvers, though this capability is significantly reduced compared to a conventional coaster brake. I recently finished racing the Wente 8 Hour, where my team did 8 laps and I completed 4, placing 32nd out of 45-ish participants. My lap times ranged from 58 minutes to 1 hour 9 minutes, which is pretty middle-of-the-road, maybe a little slower than average.

The bike is incredibly fun with the Kroozr Brake. Because of its minimal actuation displacement, I often find myself just pushing my frame forward to modulate braking force, then pulling back to accelerate. You can get into a bit of a “surfing” feel. This brake delivers significant stopping power and takes a few rides to get used to. You’ll intuitively dial in your actuation kinematics and find where you want to finish your pedaling before a downhill section, almost like second nature. Typically, I stop pedaling at 11 o’clock, preparing for optimal engagement by 9 o’clock pedal position. For aerial maneuvers or technical features, I often position the crank at 9 o’clock, allowing for an engagement window extending past 6 o’clock – providing a crucial error margin before wheel lock-up.

How This Thing is Built and the Learnings Along the Way  (Back to Front)

Caliper

I’m using a Tektro Spyre with a compression spring from McMaster (9657K361). This booster spring is critical because standard caliper return springs are calibrated for manual actuation forces, which provide insufficient tactile feedback under the significantly higher mechanical advantage afforded by leg input. You can still exceed the actuation travel, but there’s enough feedback to know when engagement begins. I tried a lighter spring, and at my weight (100 kg) and with a 175 mm crank arm length, it simply wasn’t enough. I haven’t tried any tougher springs due to concerns regarding cable anchor fatigue or catastrophic pull-through if forces get too high.

I’ve observed some wear on the caliper from the spring interface; I’ve used a single caliper for all recent revisions, so this is a known issue. However, the observed wear progression has remained within acceptable operational limits, so I’m running with it. In a perfect world, I’d design a caliper with an externally adjustable spring rate, a reinforced cable anchor point, robust lever arm geometry, and a greater cable diameter capacity. But part of my goal for this project was to stick to my design constraint of minimizing custom component fabrication, thus I made this work. (Paul Components, call me back! Let’s make dreams come true and make TENS of dollars – we could be hundredaires!)

As for rotor size, given the high potential input force from leg actuation and the desire for a delicate feel, I deliberately chose a smaller rotor diameter (160 mm) to minimize mechanical advantage at the caliper. Ideally, I’d even prefer a 140 mm rotor for further control. One nice thing about coaster brakes is you never experience “brake pump” in your rear brake hand; I always maintain a firm, full grip on my right hand.

Another thing to note: these brakes have a high propensity for unintentional wheel lock-up. Please, do not build this brake if you’re going to hit the trails and you’ve never dug before. If you want to rip up the streets and blow through all the pull-offs at your local bike shop, by all means, SKID2LIVE – LIVE2SKID. But extra caution needs to be taken on singletrack. You can also detune the system to mitigate the risk of skidding entirely – an adjustable skid threshold feature that was accidentally discovered! No trails were harmed in the creating of the photoset.

Cable and Housing

Segmented housing, under the extreme compressive loads generated, demonstrated a catastrophic failure mode characterized by segment delamination – it pops like popcorn! It was probably the funniest error I’ve made on this bike. The first segmented housing explosion took out a frame protector rubber ring with it; I assumed I just lost it, but couldn’t figure out how. The failure signature was initially elusive, requiring extensive field testing and diagnostic rides at Tamarancho and the Wildflower Century.

I thought it was cable slip, or something else settling, because when they pop, there’s no evidence. They just disappear. Until one day, I handed the bike to a Paul Components employee in Chico, and I watched him pop one; it was caught right in the line. I felt so relieved. I’m currently running Shimano BH-9000 housing and Shimano cables. I keep an extra cable with me when I ride. I’ll discuss this later, but you need to tune the brake to have a physical backstop to prevent cable breakage. Despite these measures, the inherent high-force environment means cable fatigue and eventual failure remain a potential failure mode.

Bottom Bracket and BB Shell

The introduction of the T47 bottom bracket standard proved to be the enabling technology for this integration, providing the necessary volumetric capacity within the BB shell for the brake’s components. My earliest attempts involved an extra-long spindle and mounting the mechanism on the non-drive side outside the shell. I also considered utilizing an e-bike frame’s mounting points, replacing its drive system with my brake setup, but once again, this fell outside my design constraint of minimizing custom component fabrication.

It took about five years for me to fully comprehend the T47’s potential, but I got there. Achieving optimal cable routing and minimal friction necessitated the cable’s close proximity to the non-drive side of the frame, requiring localized modifications to the BB shell to prevent component interference between the bearing sleeve and lever collar. While I operate this system with confidence, it’s important to note that these structural modifications remain unvalidated by formal testing protocols, so play at your own risk.

Nomenclature

• Bearing Sleeve: The component upon which the roller bearing clutch rotates.
• Roller Bearing Clutch: The “magic” of the one-way bearing mechanism.
• Lever Collar: The business end is responsible for actuating the brake.
• Booster Spring: The foot-feedback spring is located in the caliper.

The Magic: The Inner Workings of the Brake

The key component here is the roller bearing clutch. If you want to know how they work, watch this video, but the fundamental principle hinges on its unidirectional action. During forward pedal rotation, the clutch disengages, resulting in minimal rotational drag (akin to adding a low-friction bearing). However, upon reverse torque application, the rollers engage and bind with the bearing sleeve, which then translates into linear actuation of the lever collar, thus pulling the brake cable. It’s as simple as that: you pedal backwards, it pulls the cable.

The Major Issues

• No Backward Rolling: You can’t roll the bike backwards because the rear wheel’s freewheel mechanism is functionally bypassed, preventing reverse wheel rotation. So, no landing fakie for this rig! But with a brake line routed down the steerer tube, you can have a shifting/dropper post bike that can still barspin!
• Backstop Calibration: Precise system calibration is paramount. The brake must be adjusted so that at maximal pedal input leading to wheel lock-up, the lever collar makes direct contact with the frame, serving as a physical backstop. The high forces generated by a single leg are sufficient to induce cable failure, but thankfully, wheel lock-up typically occurs before the cable snaps. So, my solution was to ensure it locks up as close to the frame collar as possible. Housing compressibility provides a valuable compliance factor, accommodating minor calibration deviations. And if you want to ensure you don’t tear up trails, you can even adjust your brake to never lock up – this inherent characteristic can be exploited as an adjustable skid threshold feature!
• Cable Integrity: Cable snap, pull-through, and housing failures represent constant system integrity concerns. Consequently, your safe operational velocity should not exceed the stopping power afforded by a standalone front brake. It’s simply not safe to rely solely on this.

Assembly: The Nitty-Gritty

I don’t want to bore you with every single detail, but for those ready to get their hands dirty (or greasy), here’s the condensed version. For the full, exhaustive breakdown, don’t hesitate to shoot me an email!

First up, you’ll need to machine an aperture into your bottom bracket shell using a vertical mill, about 15 mm from the NDS, 15 mm wide, and 35 mm long, terminating above the draw line of the cable. Next, have a skilled machinist fabricate two critical components:

• The Lever Collar – I opted for 6061 aluminum for this.
• The Bearing Sleeve  – This component demands hardened bearing steel for its durability.
• The one-way bearing – found here

The assembly process begins by installing your drive-side bottom bracket bearings onto your crank arm. Then, apply green Loctite to the bearing sleeve and install it. Insert your crank, and seat the bearing either with a press or, if you’re feeling old-school, a carefully applied hammer. Once seated, pull your crank spindle back.

Attaching the lever collar to the one-way bearing requires a tolerance fit. I experimented with countless mechanical binding methods, but the most effective approach was a thermal expansion fit. Essentially, you’ll use a scientific oven to heat the collar; once sufficiently expanded, it should slide onto the bearing with minimal effort.

In my specific build, I had to remove some material from the bottom bracket shell to prevent the lever collar from rubbing and ensure free rotation. If I were to iterate on this design, I’d reposition the braking tab to the absolute leftmost point of the collar to avoid this interference altogether.

Finally, after dropping your collar/bearing assembly onto the sleeve, attach your non-drive side bottom bracket. At that point, you’re pretty much ready to roll!

If you have any questions or would like to discuss this further, please comment below.