In this blog we have a detailed look at the bike ratchet from first principles to failures.
Starting with how the hub ratchet and pawls work, we inevitably veer off on a tangent or two. (The first being, pedal kickback.) Bringing things back on track we then look at the high loads involved in a bike ratchet. We then drill right down to the detail of small imperfections in otherwise perfect pawl performance. How these occasional exceptions can create enormous loads that damage and accelerate decline leading to the early failure of many hubs.
First let's start with a look at how a high performance fast engagement ratchet works.
This animation shows the ratchet of a KOM Xeno hub. Three unique features are:
Its larger diameter ratchet.
Disc Side Drive
Its intergration of ratchet and brake disc: Using the extra strength of the steel disc while saving weight overall.
The ratchet illustrated above uses 6 pawls in an offset arrangement of two groups of 3 pawls. That is each group of pawls is offset from one another by half a ratchet tooth. The idea is that first one group of 3 all click and then half a tooth later the other group click.
This has a couple of really important advantages:
There are twice as many clicks for one complete rotation of the ratchet.
When one group of pawls is about to drop (or click) the other group is half way between teeth all pawls engaged waiting ready to 'catch' the drive if the teeth ahead don't quite engage or slip back. A super reliable system that will never drop you onto the crossbar.
This slow motion animation shows how the two banks of three pawls click into position alternatively. One bank of three followed by the other. Â
The use of two banks of pawls offset by half a ratchet tooth with a large diameter ratchet with lots of teeth gives a nice fast hub. In the example above there are 120 clicks for each full revolution of the hub. For technical mountain bike climbing, in particular, once you have ridden a fast hub you won't want to go back. Some slow hubs have 20 or even fewer clicks per revolution of the ratchet. That on average would mean you have to move the pedals forwards 6 times as far before getting power to the back wheels (when compared to having 120 clicks.)
When riding, if you are having to pause pedalling to avoid rock strikes and then put in a quick power stroke, a fast hub can be the difference between cleaning a climb and not. As soon as you move the pedals the power is there where you want it.
Why not even more clicks?
When you see the design, it's easy to ask why not design more clicks? You could have 3 banks of two pawls and add another 60 clicks to the ratchet above or even push the boat out and have 6 groups of 1 pawl each acting individually. That would give loads more clicks, for sure, but at a terrible cost. We cover the horrors of occasional single pawl engagement later. While exceptionally clicky, incorporating that in the design would be a masive step backwards in hub reliability.
More clicks equals more flexibility?
Another thing about having fewer pawls engaged is the resulting increased load on the pawls. That's three times the load when you have one pawl taking the load previously shared between three. What this means is that while you get the next click sooner when you put the power down the extra load means everything flexes more and maybe your pedals stretch past where you would have been. The point is there are diminishing returns followed by big compromises as you reduce the number of pawls engaged in drive. 120 Clicks seems to be a pretty near perfect sweet spot. Fast engagement and very low flex for efficient power transmission.
Infinite clicks?
One of the unique things about the Xeno hub is that with its modular design you can swap out the drive type. This means that during development we are able to just drop in new drive systems for testing without the need for new hubs and wheels. We have been testing a silent drive system with no clicks for a few years now. Personally I love the silence; you can hear the tyres working (but also any rattles or creaks). In effect this drive has infinite positions of engagement but it does have some flex and makes the overall hub installation a bit heavier. Watch this space for when this product is ready for sale. (Or of course drop us an email if you would like to be kept updated or buy an early beta test wheel or hub. S@k-o-m.co.uk)
'Snake Oil' Kickback Claims
There is not much good about riding 'slow' hubs when mountain biking. However, that's not stopped some pretty dubious 'Snake Oil Salesman' claims.
Those trying to sell 'slow' hubs sometimes claim that they reduce pedal kickback. At best this is misleading but to see why probably requires a quick explanation of pedal kickback:
What is pedal kickback?
This is a pretty massive tangent we're about to go off on so, for those interested, we've written a whole blog about pedal kickback here. Thus we can get back on to looking at the ratchet torques and resulting loads.
A look at loads and torque
Getting back on track let's have a quick look at the loads that the hub, and more particularly the ratchet, has to deal with.
Lots of torque
An easy place to start is to just look at torque loads. A simple model is to first consider just the pedal crank with a rider standing on the pedal in the 3 O'clock position. This position is Peak Torque for most riders. Once you are moving of course you can't maintain anything like this torque for more than a fraction of a second. Fortunately 'normal' riding torques are usually only a tiny fraction of this peak torque.
This diagram shows the pedal crank. Assume the crank is 175mm from BB centre to pedal axle. If a 100kg rider was standing on the pedal that would push down with (approximately easy numbers)1000N force on the end of the crank arm (Red down arrow pedal end). This produces a torque of 175Nm. (1000N * 0.175m) (Curly red arrows) at the bottom bracket show the balancing torque provided by the chainwheel in the opposite direction. An equal an opposite upwards force of 1000N is provided by the bottom bracke bearings to support the weight of the rider. All the forces are thus in equalibrium in this diagram.
Load inflation
In recent years the ratchet in particular has had to deal with 'load inflation'. The torques applied to the ratchet have increased a lot. These days E-bikes can pedal a lot harder for far longer than any top athelets. Couple with that the increasing number of teeth on the largest cassette ring and its easy to see the job of the ratchet has got a lot harder. For example, if you take the diagram above and assume a 30 tooth chain wheel driving the largest mountain bike 52 tooth ring on the cassette the torque at the rear hub will increase from 175Nm to over 300Nm (175 * 52 /30 = 303). Admittedly this is looking at the upper limits, but most hubs are failing to catchup. On e-bikes it's worse; many hubs or ratchets expire long before the first new brake pads are required. Let's face it most riders don't want to check the brake pads and change the rear hub before riding.
What are the corresponding loads on the pawls?
What are the loads on the pawls? Well, when transferring a torque of 300Nm to turning the rear wheel means the loads on the individual pawls are enormous.
However, to keep the diagrams relatively simple (omitting chains and sprockets and their reduction ratios which can factored in later) let's just assume that the ratchet and pawls are attached direct to the bottom bracket to drive the chain wheel.
This diagram shows the pedal crank again but in this model it is assumed there is a ratchet and pawls on the bottom bracket with several equally spaced pawls. (Nearly all e-bikes have a ratchet or one way clutch here to allow the motor to drive the chain wheel while the pedals are stationary. For example in 'Walk Mode'.) As before the forces are in equalibrium but distances 'r' and 'R' are also shown.
In the diagram above the radius of tangential action of the pawls is shown as 'r'.
This diagram shows r the radius the torque loads are applied on a freehub body in a typical hub. This dimension is normally 15mm or less on a typical freehub.
The working length (between centres) of the crank 'R' is also shown on the diagram further above. The ratio of these two dimensions is very significant. Using the same numbers we assume that R=175mm. For many of the more robust legacy hub ratchets 'r' is usually about 15mm. Thus it can be seen that the load on the pawls would be 175/15 (almost 12) times the load on the pedal. If we use the same pedal load of a 100kg rider standing on one pedal you can see that the load on the pawls is well over a tonne. (1.2 Tonnes or 1.166 Tonnes more accurately)
Reduced loads in Xeno Hub
In contrast KOM's Xeno hub uses a larger ratchet and pawl diameter 'r' so that the pawls are further from the axis of rotation. This reduces the load carried in the example above from approximately 1.2 Tonnes to 0.8 tonnes. (Approximately one third reduction in load on the pawls.)
Fortunately, in either case, this high load is usually shared between several pawls. If we assume 3 pawls equally loaded then it can be seen that the load would be reduced to 400kg carried by each pawl in the legacy hub. These are still large loads for small pawls and remember that these calculations do not take into account the increased load, when in lowest gears, for those pawls in the rear hub where things get that much tougher.
Pawls are one of the smallest bike components but that one of the most important. You can't pedal without pawls. We could write books on the years of research we have done trying to improve pawl performance. (Not here, you'll be glad to read.)
Xeno Pawls are Massive?
A typical pawl on the left and a Xeno pawl on the right. The Xeno is clearly a bit larger in all dimensions but it's the increased number of teeth that is the most noticeable difference. The contact surface area of the Xeno pawl is thus far larger and therefore the pressure on those contact faces is far lower for the same pedal torque.
Xeno pawls are phsically larger than most bike hub pawls but they are massively larger in terms of tooth contact area. That is the important bit as it makes their contact pressures far lower.
So what could possibly go wrong?
Understanding how rear hubs work is one thing but how they fail is another. Many of the failures in rear hubs occur in components that are not necessarily directly responsible. This distracts from the true root of the failure.
I am not suggesting that poor materials and cheap components are not the major cause of failure in cheap wheels and hubs. I accept premature bearing failure is accelerated by water ingress and corrosion. I also accept that axle breakages are far more likely with poorly designed, flexible axles with stress risers in the wrong places.
But why do expensive hubs fail so often?
More worrying is why do expensive hubs fail so often? The trouble is that the most common modes of failure are not fully understood. It's a little known fact that much of the damage and destruction originates with ratchet and pawl issues though they don't always end up the most damaged. For many hubs it's the 'exceptions' to normal performance that cause the problems.
This is an image of a typical freehub inside a ratchet. In this case all the pawls are engaged to push the wheel round.
It's the exception that proves the rule
"It's the exception that proves the rule," has always seemed a slighly odd saying to me but in this case it almost fits the bill.
What happens when your ratchet is clicking and you start to pedal and only one pawl drops rather than the normal 3? However accurately you make a ratchet and pawl this invariably happens occasionally. It's a probability thing and the chances of it happening increase if you have broken pawl springs, dirt or any number of foreign bodies in the hub. We call this 'exception' single pawl engagement.
Single pawl engagement
In this image only one pawl at the top of the picture has dropped. When pedalling started the other pawls at 4 O' clock and 8 O' clock had not dropped and now there is no chance of them dropping until we go back to freewheeling. In the meantime the single pawl (at 12 O' clock) has to take all the load normally shared between three.
We've already suggested earlier in the piece that with single pawl engagement unexpectedly massive forces can occur. It's relatively straight forward to see that when a single pawl has to take the load normally shared between 3 pawls then its load increases 3 times. This is still very bad news if you are a ratchet or pawl tooth operating near the limit. But you won't feel so sorry for the pawls when you hear what the supporting bearings have to endure.
In this diagram only one pawl is engaged and has to supply a force 12 times the rider load on the pedals (as R =12r). Consider it like a seasaw or lever. A very large person can sit near the pivot on a seasaw but be balanced by a small child far from the pivot on the other side. (This diagram is NOT in equilibrium as we have not shown all forces for the whole system.)
The diagram above shows how a single pawl has to take 12 times the load of the rider standing on the pedal to counteract that torque. What is not shown on the diagram above is the extra force on the bearings and (in this diagram the bottom bracket) axle as well.
This is a similar diagram but with the missing load added to bring the system back to equilibrium. The bottom bracket bearing that was supporting just the rider load but now has to support the rider load and also provide an equal and opposite load to the single pawl. (With equally spaced balanced pawls this extra force is not required.) Thus the axle and bearing load increases 13 times from 100kg to 1300kg i.e. 1.3 Tonnes.
It's the load in the bearings that increases most: over 20 times is normal!
Before you see the diagrams it's not obvious but once you've seen them it becomes clear. A single pawl engagement coupled with some high pedalling torque may triple the load on the single pawl but what it does to the supporting bearings is on another level. You can see in the crank diagram above the load in the bearings and axle increases 13 times. When you factor in the added torque on the rear hub from being in lowest gear you can see the increase in loads in the rear hub bearings can jump to over 20 times normal during a single pawl engagement! For a bearing or axle that's destructive.
Can you see the evidence?
Once you know why the loads can get so high for single pawls some of the damage you see in ratchets becomes more understandable.
6 pawls should have been engaged in this hub. However clearly a high load single pawl has removed just one tooth ruining this ratchet.
Look at the ratchet tooth on the right of the photo. It looks like this tooth was almost forced into the next tooth by the extra load during another single pawl engagement.
Bearings hide it well
Damaged bearings hide it well. You can feel the damage but can't see it when looking at a photo. Seriously bearings are quite good at taking occasional loads much higher than their static ratings without catastrophic failure. Static load limit is defined as when permament deformation occurs. Usually when the ball permanently distorts by more than 1/1000th. So each time a load higher than limit occurs balls get permanently distorted, races dented or both. So it is often a cumulative decline in bearings rather than instant failure.
Axle failures: obvious once you see them
Axles can be repeatedly distorted and when they fail you often feel or hear it first. Usually you can't see a broken axle with the wheel still on the bike. However once you take the wheel off it's often pretty obvious.
Spectactular failure.
Failure at a stress riser. These can sometimes seem almost like clean breaks.
Page under construction. Sorry to stop here. Hopefully we'be back on the case soon. ( It's raining again!)
The bearing at the bottom bracket end of the pedal would provide and equal and opposite force of 1000N to support the rider (Red up arrow at bottom bracket.) We know that if there was no chain the pedal would rotate the crank quickly to the bottom of its range. To stop the crank accelerating in a circular motion a torque is applied in the opposite direction.
Xeno secret sauce
In the Xeno the bearings are separated into two distinct groups that work completely independently:
'pedal bearings' and
'freewheel bearings'
Easy names: Pedal bearing work when pedalling and freewheel bearings when freewheeling. This has many rider benefits including, reduced friction, reduced bearing wear and improved performance.
This clip shows the bearings are rolling when you can see the bearing balls moving and black when the balls are stationary in their races. The pedal bearings are the smaller diameter bearings outermost on their axle.
Pedal bearings
These pedal bearings are positioned at the ends of the inner axle for even load distribution, to maintain good bearing alignment and maximum stability. When pedalling only these bearings are active.
The Xeno uses a larger 19mm diameter inner axle with its bearings located at the edges. The result is evenly loaded bearings on an axle that bends less keeping the bearings correctly aligned. Also importantly, when pedalling only these bearings turn. (The other bearings on the larger diameter drive shaft are only active when freewheeling.)
Free wheel bearings
The free wheel bearings on the far larger diameter driveshaft are only active when freewheeling. (The pedal bearings take a break and stop rolling when the freewheeling takes over.)
Only these bearings turn when freewheeling. The Xeno driveshaft is a much larger diameter shaft that runs over the inner axle enclosing the pedal bearings within. Moving the ratchet to the disc side allows the ratchet to be larger and stronger. It also frees up space for the bearing on the cassette side of the hub shell to be moved outboard taking the place of the ratchet. This rearrangement allows for more evenly loaded hub shell bearings (red arrows). (Compare the drive shaft with legacy hub shell bearings image, where one bearing in the middle of the axle takes almost all the rider loads.)
The Xeno driveshaft can be imagined a bit like a free hub body that has been stretched and extended the full width of the rear axle. The bearings within are thus moved much further apart. That makes the driveshaft far more stable on its axle than a short free hub body. But that is only part of the story. The freewheel bearings are also much larger diameter bearings.
Different jobs for different bearings
The larger diameter freewheel bearings run on the outside of the driveshaft supporting the hub shell. In the image above the arrows give an impression of how the normal riding loads are distributed. It might lead you to wonder why the bearing on the left is such a large diameter bearing, by far the largest of all the bearings, when it carries a smaller proportion of the total riding load. Well the reason for this is that this bearing has to do another key job; that of holding the driveshaft and pawl carrier central in the ratchet. The loads in this area can get very large especially if only one pawl is taking all the drive torque. On most hubs this happens infrequently, which is just as well as it tends to damage and ultimately destroy free hub and adjacent bearings.
Use a shaft up to the job
Some of the reasons for such a large diameter drive shaft have alreay been explained. One one of its main requirements is to transmit torque drive from the cassette side through the wheel hub to the brake disc side. Larger diameter shafts are far better and transmitting torque with minimum distortion. They are also far stiffer in bending so good at transmitting riding loads from the free wheel bearings to the inner axle and ultimately the bike frame.
But as mentioned above a super strong drive shaft is required in this area is so the hub can handle high torques even when applied unevenly for example if just one pawl engages.
Single pawl engagement
When only a single pawl engages it has to take all the load that would normally be shared with its fellow pawls. So if there are normally three pawls engaged when only one pawl drops it has to take 3 times the load.
One way to increase the strength of the pawl for high load situations is to make it larger with more teeth. (Legacy pawl on left KOM pawl on the right.)
The Xeno is designed to take these occasional high loads. Increasing the diameter of the ratchet reduces the load in its teeth as does increasing the number of teeth engaged.
The root of many hub failures
Single pawl engagement is the bane of many hubs and can often be traced back as the root cause of most hub, ratchet and axle failures. It is a massive subject and for those interested in a really deep technical dive and explanation it is covered in a whole separate blog.
What happens to cause the damage?
High torque pedalling, combined single pawl engagement results in a very large uneven load perpendicular to the axle that has to be resisted by axle and bearings. It is for this reason that the drive shaft and bearing are so large in the Xeno (and why the two groups of bearings, pedal and freewheel, operate totally independently.) In the Xeno the pedal bearings are protected from these high loads by the drive shaft and its bearings so that they can continue running silky smoothly.
Xeno super strong shaft
The Xeno uses a far larger drive shaft and bearing to safely handle these loads.
The difference is striking when you compare the axle size (shown below) and wall thickness of the axles between the Xeno and legacy hub.
On the left is the KOM Xeno drive shaft that the large bearing runs on. On the right is a typical axle from a legacy hub (15mm OD 12mm ID.) The thin little axle only has about 4% of the stiffness of the Xeno drive shaft. Or to put it another way the Xeno dimensions on the left make the axle 25 times stiffer and of course many many times stronger.
Its a similar story with the bearings too.
On the left is the KOM Xeno single bearing whose job is to keep the pawl carrier and drive shaft central in the ratchet even when the going gets tough. On the right is a typical bearing from a legacy free hub struggling to do a similar job.
Xeno: Performs better and lasts longer.
So that's a look at bendy axles. I hope it gives an insight into why the Xeno performs better its axles don't break and its bearings last many times longer.
We intend to do an anniversary blog, looking at the first KOM prototype hub that was actually fitted to a bike, almost exactly 6 years ago. The bike used to do a lot off off road, took me over the Alps and Dolomites to Venice (Italy) and is still used, normally several times a day as a commuting bike. The hub is still running on its original bearings so it should be interesting to have a look at how things are wearing.
Consider first a evenly loaded legacy hub
Once again the FE model can help visulise the loading with its exagerated images of distortion.
This is a view of a typical hub with three equally spaced pawls sharing the torque loads.
Above shows what a ratchet should look like when engaged. 3 pawls, all engaged and evenly spaced around the ratchet. When torque is applied the pawls try to stand up and the ratchet is pushed outwards as well as being turned.
When a large torque is applied the pawls are loaded and exert an expansion load on the ratchet as well as transmitting the torque. That is the ratchet or hub body has to contain radial loads. However the axle is not distorted by this load.
This FE model has a window cut in it so that we can look at the way loads on the axle affect its shape as well.
Looking at the side it can be seen that with torque applied the free hub stays central with pawls even spaced around the ratchet. Also looking through the round window, cut into the model it can be seen that the axle remains straight.
It can be seen that when drive loads are applied evenly with equally spaced pawls around the ratchet that no uneven loads are transmitted to the axle or bearings by the torque.
Next consider an assymetric load on a legacy hub
Let's now look at what happens if say only one of the pawls engages.
In this model just one pawl is engaged. This might happen if for example the other pawls did not quite drop.
This is a very bad scenario for hubs and hopefully only happens occasionally and if so when you are pedalling very gently. Let's have a look at what happens if you happen to be pedalling hard:
In this model just one pawl is engaged. The side view through the window is very revealing.
The axle has to trasmit some horrible loads between the free hub bearings and the hub shell bearings. You can see how the axle is also bent by the offset loading of the single pawl.
Running the model when just one pawl is engaged. The side view through the window is very revealing.
Many hub axles are so flexible that when the loads get high the axle bends so far that the pawl carrier is forced into the ratchet teeth opposite the engaged pawl. If you look at an old Free hub body you can often see the telltale marks. This probably gives a good indication of order of events that lead up to the hubs failure.
Why are the loads from single pawl engagement so damaging?
When say three load carrying pawls are equally spaced around the ratchet they share the torque load and the resultant linear force on the axle is zero. The pawl loads have to have equal and opposite reaction forces which are transmitted from freehub to axle via the freehub bearings and then from the axle to the hub shell via the main hub shell bearings. With that sort of bending applied and then rotated, as the wheel turns, the axles have such terrible loading that they tend not to last very long.
Of course its not just the axle; the small bearings suffer too as they are horribly overloaded.
We have a couple of intereting stories in the pipeline hopefully out pretty soon:
Does it sometimes feel like you left the parking brake on when you arrive at the top of a climb? We will show you why you might be right.
Christmas brake? We are nearly always testing brakes and have lots of interesting news to update you on.