Optimizing the Cantilever Brake

Optimizing the Cantilever Brake

Cantilever brakes have traditionally been the way to make brakes for bikes with wide tires. Invented by Nicola Barra in 1936, they were the first direct-mount brakes with pivots that attached to the fork blades and seatstays. This eliminated the flex that is inevitable with long brake calipers that reach all the way around the tire. Cantis greatly improved the braking, and the new brakes were popular on early randonneur and cyclotouring bikes. Decades later, they made their breakthrough into the mainstream with mountain bikes.

Cantis have many advantages. They are light – at 75 g, our Rene Herse cantis are among the lightest brakes in the world. The posts are standardized – you can mount almost any canti on almost any bike with canti braze-ons. They are reliable – there’s little to go wrong that a home mechanic can’t fix.

Despite all these advantages, you’ll be hard-pressed to find a new bike equipped with cantis. For mountain bikes, front suspension meant that there was no logical place to anchor the brake cable any longer, and it made sense to use hydraulic discs. As a plus, they had more braking power and required less hand force than most cantis.

For road and all-road bikes, centerpulls eliminate the one inherent weak spot of cantis: The pivots are located in relatively poorly supported parts of the fork blades and seatstays, which can twist under hard braking. This means that the angle of the brake pad changes, resulting in non-linear braking.

That is the theory. I wasn’t a big fan of cantis… until we started to work with Peter Weigle on a bike for the Concours de Machines, the competition for the lightest fully equipped all-road bike. Our goal was to build a no-compromise bike: reliable and fully equipped with fenders, generator lights, rack and wide tires. Yet our target weight was no more than 20 pounds (9.1 kg) – fully equipped, including the bell and pump. Peter was going to pull out all the stops with a featherweight frame and many modified components. Every gram counted – discs were out of the question, and even centerpulls didn’t look like the best option. I was willing to make some compromises in brake performance to reach our target weight.

So we began to look at cantis.

As we thought about how to optimize cantilever brakes, we realized that most cantis are – how to put it politely? – suboptimal in their design. Take the straddle cable: It’s made from the same thick wire as a brake cable, which makes too stiff to conform to the straddle cable yoke. (You can pre-bend it, but it’s never perfect.) When you brake, the first part of the lever travel only serves to tighten the cable, before can you apply braking force to the pads.

We calculated the stresses on the straddle cable and realized that a thinner cable would be strong enough. The only problem is that the cable is clamped to one brake arm. Every time you brake, the cable’s exit angle changes, and the cable flexes acutely where it leaves the clamp. Flexing weakens the cable until individual strands start to break, and the cable frays. The thick straddle cable is needed to prevent premature failure.

What if we allow both ends of the straddle cable to swivel in the brake arms? Now we can use a thinner straddle cable, which will always be pulled tight by the return springs of the brake. We no longer lose a third of the brake lever travel to flex before the brake squeezes the pads. Which means more braking power – plus we can set the brake pads further from the rim without worrying about the brake lever bottoming out against the handlebars.

Another issue of most cantis is the fiddly adjustment of the brake pads. You want the pads to have a little toe-in (the front of the pads should touch the rim first). Most systems to adjust the pads are hit-or-miss, because the pad angle inevitably changes as you tighten the bolts.

When you think about it, the toe-in only needs to be set once. The angle of the brake pads is determined by how the brake posts are brazed onto the frame. (There are always minor tolerances.) Rene Herse used his brakes only on bikes he built, so he simply bent the brake arm until he got the perfect toe-in. Once this was done, the customer never had to think about it again.

We realized that this was the way to go. Getting rid of all those spherical washers eliminates a lot of flex (and weight). And when it’s time to replace the pads, you don’t have to set the toe-in again, as the adjustment remains the same. (For riders who aren’t comfortable bending their brake arms, we offer a version with an angled washer that allows angling the brake pads without bending the brake arm.)

Another issue are the brake pads themselves: Many modern cantis use the same pads as sidepull brakes – they bolt right onto the brake arm. That means you don’t need special hardware (which is expensive to make), but there is a big disadvantage when the time comes to adjust for pad wear. With canti brakes, you don’t want to use barrel adjusters: As the pad wears, it hits the rim lower and lower, until it can dive under the rim as you brake hard. Then you lose all brake power. (On sidepull brakes, the pads moves the opposite way and can cut into the tire, but the long arms reduce this effect.)

You need to adjust the angle of the pad as it gets thinner, which means fiddling with the toe-in adjustment again… unless you have proper canti pads have posts – you simply open the nut and slide the entire pad closer to the rim. Tighten the nut again, and you’re done.

As for the brake arms themselves, you don’t want them to flex. When you pull on the brake lever, the entire force should be transmitted to the brake pads, with no power lost to flexing the arms. To achieve this, we optimized the shape using Finite Element Analysis (FEA) – a computer model that calculates the stresses. Blue and green indicates low stresses that will not cause significant flex. (The red dot is something known as an ‘edge error’ in the model, not a place where the stress is high.) The FEA also allowed us to pare the brake arms down to an absolute minimum, which is why they are so light.

Minimalist brake arms must be forged to align the grain of the aluminum with the stresses. This makes the brake arms much stronger than if they were carved out of a big block of aluminum with a CNC machine. (There’s also less waste with forging, so it’s better for the environment.) The down side: Forging requires very expensive tooling. To us, it was worth the effort.

While we were stripping weight off the brake arms, we thought about the adjustment for the spring tension that most cantis have. You only need to adjust the tension if the two springs are not equal. If we can make springs that have equal tension to begin with, we don’t need the tension adjustment. That saves weight and makes the setup of the brakes even easier. In fact, our springs wrap around the brake arm, since there is no room for a hole.

We also thought about the hardware. Usually, replacing steel bolts with titanium isn’t a good idea, since the bolts are dimensioned for the loads that they experience. The eyebolts that hold the brake pads are an exception – they have to be big so the brake pad fits through them. Making them out of titanium saves weight with no loss of strength. (We offer less expensive steel bolts for riders who aren’t concerned about a few grams of weight.)

In this way, we analyzed, calculated and rethought every part of the brake. And then we made each part from the best materials and to the highest precision.

Did all this effort make a difference? The first test came during the Concours in France. The final stage included a long mountain descent in pouring rain, with numerous hairpin turns. I was surprised how well the Weigle braked. The cantis were so powerful that they scraped the water off the rims within just a few revolutions of the wheel, and the braking action was powerful, linear and easy to modulate.

Before we release our products, we test them extensively, and so I mounted a set of the prototype cantis on my cyclocross bike. My Alan always had suffered from pronounced fork judder at low speeds – not really a problem when racing, but annoying nonetheless. I thought this was due to the stiff fork and flexible steerer, but to my surprise, the judder was gone with the new brakes. It appears that the flex and play in the pivots of the old cantis was the main culprit.

Since then, I’ve ridden many miles with our Rene Herse cantis, often in challenging conditions. I’ve realized that 97% of the issues I had experienced with other cantis were due to their design and manufacture, and only 3% were caused by the pivot location that is inherent in all cantilever brakes.

I still believe that for most production bikes, discs are a great option, and for a custom randonneur bike, I prefer centerpulls for that last 3% of linear braking action. But great cantilever brakes – with every part optimized and honed to perfection – are so good that I never feel limited when I’m riding with them.

Further reading:

More about the J. P. Weigle from the Concours de Machines.
• The full report from the Concours, plus Peter Weigle’s story of how he built this amazing bike, was published in Bicycle Quarterly 61.

Share this post

Comments (45)

  • Steve Palincsar

    My JP Weigle has a set of the Rene Herse cantilevers too, and I must say, they’re the nicest brakes I’ve ever used. Wonderful feel at the lever, and terrific braking performance. And unlike most cantilever brakes, these are simply beautiful – almost jewel-like.

    There are only two disadvantages I can see:
    1. Placement of the pivots needs to be precisely to spec, since these brakes don’t have a slew of adjustments you can make to compensate for sloppy workmanship. With a bike like a Weigle, of course everything is going to be spot on, so no worries there.
    2. They’re not cheap. On the other hand, the sort of bike they’re going to go on isn’t going to be cheap, either, and although they cost a good bit more than most cantilevers, the price really isn’t outrageous in the long term and the performance and appearance are worth the difference.

    January 18, 2021 at 12:18 pm
    • Jan Heine

      Glad you’re enjoying the brakes. Regarding the pivot location, there’s actually quite a bit of adjustment possible, since the pads slide in an out and rotate. I’ve not yet found a bike where the brakes didn’t fit. As to the price, the Herse cantis don’t cost more than other high-end bike parts. I recent times, cantis have migrated to the bargain basement level of bike components, but their quality (and reputation) has suffered as a result.

      January 18, 2021 at 12:44 pm
  • Richard Keith Gaunt

    With the spring wrapping around the brake arm is there much stress on the arm from spring pressure? Also, would this spot on the brake arm become worn from spring movement and cause a potential stress riser location?

    January 18, 2021 at 12:23 pm
    • Jan Heine

      We briefly thought about this, and it’s one of the places where it helps to have decades of experience. Rene Herse introduced his second generation brakes in 1945, with the same spring arrangement. We looked at bikes with the original brakes that have been ridden for 10,000s of miles, and it’s not an issue. The spring tension isn’t that high – all it does is straighten the straddle cable and pull back the brake lever.

      January 18, 2021 at 12:47 pm
  • Charles Breer

    These look beautiful. What are your thoughts about the applicability for retrofit on a tandem? I currently have V-brakes on a 20+ year old tandem frame and I’ve never been happy with the brakes, which require special pulleys. As part of an overall parts upgrade on the bike, I’m looking for brake options.

    January 18, 2021 at 1:08 pm
    • Jan Heine

      We’ve never run the modern version on a tandem, but our 1946 Rene Herse tandem has the classic version. I found that the braking (with modern pads) was adequate for mountain descents in the Vercors (Pre-Alps) of France…

      January 18, 2021 at 2:49 pm
  • Mark Guglielmana

    Just a small correction: Finite Element Analysis is always acronymed as FEA. FEM is Finite Element Method. Studying or analyzing a phenomenon with FEM is typically known as finite element analysis (FEA).

    January 18, 2021 at 1:16 pm
    • Jan Heine

      Of course, you are right. I’ll fix the typo.

      January 18, 2021 at 2:10 pm
  • Andrew Cheatham

    I chose the Rene Herse cantilevers for my custom rando by Johnny Coast after seeing them on Peter Weigle’s Concours de Machines bike and Johnny’s Sacramento NAHBS winner. I’m a fan of details done right and have enjoyed the most sublime attributes of many bike components for those reasons.

    These Rene Herse cantilevers really do add a wonderful bit to a very nice bike. I am often complimented by stunned riders who stop and ask about my bike and instinctively give it a lift in preparation for a weight comment and are surprised.

    Steve’s Weigle is more precious but I’ve learned much from reading, listening about and seeing bikes like his and those in the pages of BQ. Enjoy it more than my center pull rim braked and mechanical disc-braked bikes. That’s important to me.

    January 18, 2021 at 1:37 pm
  • Jerry Selwa

    Anyone with some experience making cantis that were originally set up for 27″ to work with 700C? I’ll try anything , even different brakes
    I’ve toured in Japan with the 27″ers, but was always uncomfortable with only one fold up spare.

    January 18, 2021 at 1:51 pm
    • Jan Heine

      In most cases, this is no problem. Jack Taylor even used the same spec for both 700C and 27″ frames. I’ve converted two Jack Taylor tandems from 27″ to 700C. The difference is just 4 mm… Check whether you can adjust the pads so that they end up 4 mm too low. If yes, then a 700C wheel will fit right in – with the added advantage that you’ve gained 4 mm tire clearance (at least at the top, a bit less on the sides).

      January 18, 2021 at 2:48 pm
      • Mike

        I converted my Borthwick from 27″ to 700C and there was no problem at all with the cantilever brakes.

        January 19, 2021 at 10:57 am
  • kevin lindsey

    Will these fit all canti bikes? I ask because a set I ordered from Paul’s Engineering did not fit on the rear of my 1984 Specialized Expedition (although the fronts went on fine). Paul subsequently said that the older bikes had a slightly different spec that is not compatible with current canti brakes.
    Thanks,
    Kevin Lindsey

    January 18, 2021 at 2:06 pm
    • Jan Heine

      It really depends. Older bikes often had canti spacing that was closer than the current spec, and that is no problem with these brakes – you just need to slide the pad all the way in. In some cases, you’ll have to cut a bit off the pads to make them thinner (or use thinner pads, or pre-worn ones). The height of the posts is a spec that hasn’t changed since Rene Herse introduced his first brakes in 1938, but some frames aren’t quite to spec. Small variations don’t matter, but if it’s a lot off, then you just have to see with trial and error. Generally, post-style pads allow far more adjustment than screw-on pads, which can only slide so far in their slot.

      January 18, 2021 at 2:46 pm
  • Matt Gilkey

    Is there still the possibility of a Rene Herse front cable hanger? I have these brakes and love them but the hanger seems to be a weak point in the system. I have tried several different hangers but they all end up having a surprising amount of flex.

    January 18, 2021 at 2:19 pm
    • Jan Heine

      Our new cable hanger is in production right now. There are many steps to it, so it’s taking a while. As you mention, when you strengthen/stiffen/improve everything else, the cable hanger often becomes the weakest link.

      January 18, 2021 at 2:42 pm
  • Stuart Fogg

    My takeaway is the execution counts as much as or more than the concept. I’m sure you guys could make a great coaster brake!

    January 18, 2021 at 2:33 pm
  • Thomas Pirko

    I presume that the tubeset was Kaisei Superlight.

    January 18, 2021 at 4:10 pm
    • Jan Heine

      You mean the bike? Peter Weigle never talks about the tubing he selects, but I suspect there’s plenty of Kaisei in that bike…

      January 19, 2021 at 9:15 am
  • Jim Kramka

    I would love a detailed troubleshooting guide for these brakes. I have them on my Soma Grand Randonneur frame and they squeal horribly. I’ve scared many pedestrians, fellow cyclists, and even drivers! I’ve tried the flat washer and the angled washer to set toe-in correctly. I’ve tried both the salmon and black Kool Stop brake pads. I’ve moved the pads close to the rim and away from the rim but I still have squeal.

    January 18, 2021 at 4:44 pm
    • Jan Heine

      I’m sorry to hear about your squealing brakes. Brake squeal is like clicking noises on a bike – it can be hard to exorcise.

      If the toe-in is set correctly, I’d look at the fit between canti posts and brakes. If the canti posts are too small in diameter or distorted during welding/brazing, you can get vibrations that will cause the squeal. While that is fixable (machine the posts so they are round, make undersize bushings for the brakes and hone them to fit), the cost will very high – machine shop time is very expensive. You might also try a hack like Jacob indicated in his comment below: flaring the posts slightly to make the fit tighter, but that’s also not something undertaken lightly.

      January 19, 2021 at 9:20 am
    • John Duval

      If you are riding in dry conditions, like the southwest US, you might try Shimano pads in post mount holders. After much tinkering, the salmon 4 dot pads still squealed so much that the bike was unsafe to ride. Black 4 dot pads worked adequately, but the squeal was still embarrassing. I have tried other inserts, such as Jagwire, Swiss stop and a few others, all of which were prone to squeal and had much less power. The Shimano dura-ace inserts have been silent and powerful, even when not toed in or aligned perfectly.

      January 19, 2021 at 9:01 pm
  • Benjamin Van Orsdol

    Are they polished?
    Also, is there a close up of the chrome braze on on the inside of the chain stay? Assuming it’s the chain hanger, but probably done with some extra special JPW styling. It’s super high up

    January 18, 2021 at 7:36 pm
    • Jan Heine

      The brakes are polished – the raw forgings are gray… The chainhanger is mounted very high for Rinko. That way, the chain is tensioned and unlikely to come off when carrying the bike.

      January 18, 2021 at 10:04 pm
      • marmotte27

        I’ll make a note of this for my next bike. On my current one, the chainhanger is useless.

        January 18, 2021 at 10:50 pm
      • Steve Palincsar

        My Weigle has two chain hangers, one in the conventional low position, and one high like yours for Rinko.

        January 19, 2021 at 10:58 am
  • Tim

    Alex Singer’s front hanger appears to have contacted the upper bearing cup, which blocked the flex. Very simple design.

    The old Ritchey front hanger (early 90s) is pretty flex-free, in my experience.

    January 18, 2021 at 8:01 pm
    • Jan Heine

      That is a smart idea that we’ve also thought about – although I’m not sure it was entirely on purpose, since on my Singer, the hanger doesn’t contact the headset cup. It would also work only with certain headsets. With a Chris King, if you did that, you’d lock the steering under hard braking, since the upper cup is fixed, and the cone is turning…

      January 19, 2021 at 9:15 am
      • Tim

        Interesting, I have never used the CK headset so, wasn’t aware of this. And all but one of my bikes have threaded headsets so my familiarity with threadless is limited.

        One aspect I have noticed is that headset cable hangers vary widely in the length of the lever arm- a long lever will flex more, all other things being equal. The Ritchey hanger is short (it doesn’t touch the headset at rest, I realized later, until the brake is pulled. Clearance is about .01″ or the thickness of a high E guitar string, so the flex is stopped then).

        I have a modern Tektro hanger on my allroad bike (from 1996, a Riv All-Rounder. We just didn’t have suitable tires then) which is pretty rigid by virtue of being thick. The lever arm is medium, I would say, but many bikes won’t have enough head tube to fit it.

        My old steel Mafac hanger was very long and flexy.

        January 21, 2021 at 10:41 am
  • Adam

    20 years ago my fork bent backwards (not much, maybe 15-20mm measured on wheel shaft level) during fast descending on country gravel road. As result, front canti brake locked the wheel so fast, that I landed face on the road faster than I realized I am falling. 2 Hospitals and 7 hours long cosmetic surgery saved my face.
    I will never ever again use any canti, despite all respect for you Jan.
    (And do not tell me “it was fork that failed , not a canti…”.)

    January 19, 2021 at 1:14 am
    • Jan Heine

      That is a scary accident, and I’m glad you’ve recovered. Even so, I’d like to talk to the investigator who concluded that the brake locked onto the wheel and caused this. It’s hard to imagine that the non-linearity of cantis is powerful enough to send you over the bars. Or was it that the fork hit an obstacle and bent backwards first, and that locked the wheel? In that case, I can see it, but in that scenario, an accident is hard to avoid. If the tire hits the down tube and goes on either side, the fork has to turn, and it’s inevitable that you’ll lose control. And if you run a carbon fork that can’t bend that far, it’ll probably break from that impact…

      January 19, 2021 at 9:13 am
  • Matt Walker

    What is the recommended method for bending the arms for toe-in?

    January 19, 2021 at 2:13 am
    • Jan Heine

      Mount the brake, remove the brake pad and eyebolt, put something on it to protect the finish (I use an old piece of tire casing), and take an adjustable wrench set to just the right distance. Then pull or push. There was a detailed, step-by-step article in Bicycle Quarterly 68 with photos. We may republish this here soon.

      January 19, 2021 at 8:58 am
  • Pete Chesworth

    I have had a bumpy canti journey. My BDB Pelican came with Tektro CR720s – ok but prone to shudder and sensitive to pad wear. Moved to Shimano CX70 which simplified things but weak on power. Am about to migrate to V brakes but struggled to find ones with the appropriate arm length that “thread the needle” between rack and fender (a big advantage of cantis is not worrying about that). Also require new levers. The other big canti variable is the levers. Shelton got it right in his mechanical advantage article. Yet for those of us who like tinkering, cantis are appealing 🚴🏼

    January 19, 2021 at 3:25 am
    • Jan Heine

      Bad cantis are like all bad brakes – very frustrating. But the same applies to discs, dual pivots and even centerpulls. Rene Herse cantis work with most levers – you get a little more power with modern ‘aero’ road levers, and a little more pad clearance to the rim with classic or ‘canti-specific’ levers. I prefer the modern levers – the pad clearance is always enough since the straddle cable doesn’t bow as on other cantis, so you have less lost motion.

      January 19, 2021 at 9:00 am
  • Benjamin Van Orsdol

    Are these only available with the 4dot pads? Doesn’t Kool Stop offer more thoughtfully designed pads which perform better than this retro style? Salmon color either way, but why the 4dot preference??

    January 19, 2021 at 5:47 am
    • Jan Heine

      We tested all types of pads, and the 4-dots offer the same performance as smooth pads. We like that the thicker pads last about 2-3 times as long as thinner pads. If you don’t like the 4-dots, you can fit any post-style pads. If you want to use the same holders, Campagnolo-style pads from around 2000 fit. Kool-Stop makes them in salmon, too. But you’ll have to double your budget for brake pads…

      January 19, 2021 at 9:02 am
  • Jacob Musha

    I still need to try this brake. After a frustrating amount of effort, I’ve exorcised (almost) all the demons from a Tektro CR-720. Think of the worst judder you can imagine at almost every speed. Toeing in the pads was a temporary solution. They’d wear, and problems would return. Then I hand-filed brass washers to the exact thickness to take up the length slack of the brake/post. Better, but not fixed. Finally I over-tightened the brake bolts enough to bulge the posts slightly, enough to allow them to still pivot without binding. Now they don’t judder and only give an occasional squeal under very hard braking, just like my Paul Neo-Retros on another bike. A drilled-stem cable hanger might not be helping but I like the simplicity and stiffness. I probably should’ve just installed a different brake, but I didn’t want to let this one “beat” me.

    Funny, I never had judder on a slew of old bikes with bargain-basement 1990s Shimano low-profile cantis… V-brakes are fantastic on my MTBs and cause zero problems, but I’m not a fan of the big clunky drop-bar V-brake levers.

    January 19, 2021 at 8:53 am
    • Jan Heine

      Play in the pivots is one of the biggest problems with canti brakes – it exacerbates all the problems that we then attribute to the brake itself. Same thing on my Alan – new brakes cured all kinds of things that I thought were due to frame/fork or inherent in canti brakes.

      January 19, 2021 at 9:04 am
  • Reid Harding

    Love the Rene Herse cantis. I do find that oval cross section fork tubes have a greater effect on the non linearity of cantis versus round fork tubes. The comparison made between my Norther Lyon with Kaisei fork tubes and my Jo Routens with round Forbur fork tubes. The bike with the oval fork tubes has some brake squeal during heavy braking on steep descents where the round fork tube bike exhibits no brake squeal on the same descent.

    January 19, 2021 at 10:31 am
    • Jan Heine

      There are many things that can affect brake squeal, but you are right – a round fork blade resists twisting better than an oval one. Fobur tubing also wasn’t very light, so the blades of the Routens probably also have thicker walls and are stiffer.

      The Kaisei ‘Toei Special’ fork blades are optimized for centerpull brakes. Even so, the J. P. Weigle that you see in the photos doesn’t have any brake squeal…

      January 19, 2021 at 10:55 am
  • Joe Kendrick

    Any chance that the front rack on the Weigle bike can be purchased? It’s a beauty.

    January 19, 2021 at 1:40 pm
    • Jan Heine

      You can buy a similar rack from Peter Weigle – but you need to buy the bike with it! Seriously, that rack is custom-made for that fork, and it wouldn’t easily fit on other bikes.

      January 19, 2021 at 2:18 pm
  • R

    What do you do to add toe-in when the brake pad wears to where it no longer has toe-in?

    January 20, 2021 at 8:24 pm
    • Jan Heine

      The pad wears pretty evenly, so the toe-in doesn’t just disappear. Usually, the pad is worn out before the toe-in is gone. If the pad does wear into a wedge shape, you can remove it from the holder and insert in the other direction.

      January 20, 2021 at 10:24 pm

Comments are closed.