updated web page on disk brakes and quick releases

Quite a lot more information has come to light since I first put up my page. I believe there should
be enough there now to convince any remaining sceptics, especially if you read the linked thread on
singletrackworld.com. The simple fact is that under braking, the quick release skewer is prone to
unscrewing and ejecting the front wheel on standard bicycle forks equipped with disk brakes. This is
not user error but a basic design flaw.

http://www.ne.jp/asahi/julesandjames/home/disk_and_quick_release/index.html#support

If anyone still disagrees with me after reading the page, can you explain whch bit you still
disagree with?

James

 

James Annan writes:

> Quite a lot more information has come to light since I first put up my page. I believe there
> should be enough there now to convince any remaining sceptics, especially if you read the linked
> thread on singletrackworld.com. The simple fact is that under braking, the quick release skewer is
> prone to unscrewing and ejecting the front wheel on standard bicycle forks equipped with disk
> brakes. This is not user error but a basic design flaw.

http://www.ne.jp/asahi/julesandjames/home/disk_and_quick_release/index.html#support

I have no doubt that your assessment of disengagement of a QR (or even nutted axle) wheel is
probable with a typical front disk brake located behind the fork having conventional "vertical"
dropouts. However, I disagree that this can unscrew QR's. I see no mechanism that would cause
rotation of the QR lever or the adjusting nut at the other end. I don't believe that loosening is
even part of the failure mode and that brake force alone disengages the wheel even when fairly
tightly held, as Lee Bower describes in the referenced URL above:

----------------------------------------------------------------------
"I have briefly experienced what you describe. A while back, I installed a 2002 Rock Shox Psylo on
my XC bike with Hayes disc brakes and a 6-inch rotor. One of the tests I often do to test fork
stiffness is to shift my weight back and lock up the front brake at about 5-10mph on asphalt,
causing it to skid (be careful if you try this). The first time, the front wheel quickly cocked to
the left, with the left side of the axle (same as brake) dropped in the slot. The lawyer tab
stopped further movement. Questioning how tight I had the skewer, I opened and closed it again to
what I would consider reasonably tight. I performed the test again, and the wheel again cocked
sideways. I continued to tighten the skewer with the same, albeit better, results. Finally, when
it took wearing a glove to reduce the pain of the skewer in my palm, I was able to stop the wheel
from slipping in the fork."

Lee Bower
----------------------------------------------------------------------

Jobst Brandt [email protected] Palo Alto CA

 

[email protected] wrote:

> I have no doubt that your assessment of disengagement of a QR (or even nutted axle) wheel is
> probable with a typical front disk brake located behind the fork having conventional "vertical"
> dropouts. However, I disagree that this can unscrew QR's. I see no mechanism that would cause
> rotation of the QR lever or the adjusting nut at the other end.

The skewer tension multiplied by the thread pitch generates an unscrewing torque. This unscrewing
seems to be an enirely standard failure mode of a bolted fastener when vibrational and slipping
forces are present.

I did put up a quote from www.boltscience.com and also a single anecdote where the unscrewing is
observed in some detail (there are many more similar examples).

I also note that many other bolts connected with the disk calliper do frequently come loose. Do you
object in principle to the idea of bolted fasteners loosening under vibration, or do you have a more
particular objection to considering the QR as one such possible case?

James

 

James Annan wrote:
> The skewer tension multiplied by the thread pitch generates an unscrewing torque. This unscrewing
> seems to be an enirely standard failure mode of a bolted fastener when vibrational and slipping
> forces are present.
>
> I did put up a quote from www.boltscience.com and also a single anecdote where the unscrewing is
> observed in some detail (there are many more similar examples).
>
> I also note that many other bolts connected with the disk calliper do frequently come loose. Do
> you object in principle to the idea of bolted fasteners loosening under vibration, or do you have
> a more particular objection to considering the QR as one such possible case?

A properly designed and tightened QR should be immune to this problem. There are two ways for a QR
to be loosened:
1) by the lever opening, or
2) by unscrewing of the nut from the QR skewer.

The first should not occur as a result of vibration since the maximum tightness occurs shortly
before the QR lever is moved to the fully 'closed' position. Therefore if it vibrates just a bit
open there will be a torque tending to close it again and subsequent vibrations will move it towards
the closed position. This is in contrast to the situation with a bolted connection where there will
always be a torque tending to unscrew the bolt until it is completely loose and therefore it can be
gradually unscrewed by vibrations.

The second mechanism will also not occur as long as the QR has sharp edges that dig into the dropout
when it is tightened. A vibration that tends to unscrew the nut from the end of the QR skewer would
need to raise these edges over the indentations they've made in the dropout. Again, there might be a
very small movement from a vibration, but the tension of the skewer pulling the edges against the
indentations will tend to move the nut back to its original position under subsequent vibrations. A
regular bolted connection doesn't have these sharp edges and any sharp irregularities would be worn
away by the turning motion as the bolt is tightened.

So even though I have once discovered that my QR was loose in the middle of a ride, I don't believe
a properly secured QR will spontaneously loosen from vibration. Of course loosening is possible if
the QR was never properly tight to begin with, possibly as a result of dirt or other material under
the QR making it feel tighter than was really the case.

BTW, thanks for your article on the problem with disk brakes. One person with whom I ride has this
type of brake on his road bike (with regular QR and no 'lawyer lips'). I mentioned your article to
him and he immediately saw the problem with the forces involved and is looking into ways to make his
bike safer.

 

James Annan writes:

>> I have no doubt that your assessment of disengagement of a QR (or even nutted axle) wheel is
>> probable with a typical front disk brake located behind the fork having conventional "vertical"
>> dropouts. However, I disagree that this can unscrew QR's. I see no mechanism that would cause
>> rotation of the QR lever or the adjusting nut at the other end.

> The skewer tension multiplied by the thread pitch generates an unscrewing torque. This unscrewing
> seems to be an enirely standard failure mode of a bolted fastener when vibrational and slipping
> forces are present.

That's true for any threaded fastener but thread pitches, even coarse ones are in the friction safe
mode even when lubricated. What unscrews them is incremental torsional fretting, something that
probably occurs with one sided brake torque on the fork ends. Still, the radius of the pressure
faces of the skewerlever and nut would cause the skewer to twist and possibly to have its thread
move in the nut, the nut remaining rotationally fixed with the dropout.

> I did put up a quote from www.boltscience.com and also a single anecdote where the unscrewing is
> observed in some detail (there are many more similar examples).

They describe vibration loosening.

http://www.boltscience.com/pages/vibloose.htm

However, the ratio of bolt length to diameter and diameter of but face is nowhere near that of a QR.
The QR has worked all these years because it is such a good "strain bolt" In fact the over-center
feel of the closure motion is an indication of how much strain there is in the axle-skewer
combination.

> I also note that many other bolts connected with the disk calliper do frequently come loose. Do
> you object in principle to the idea of bolted fasteners loosening under vibration, or do you have
> a more particular objection to considering the QR as one such possible case?

This all goes back to the basic problem that there are insufficient engiennering skills in the
bicycle industry where competition is fierce for low cost/weight/fashion instead of druability and
reliability.

----------------------------------------------------------------------

From: [email protected] (Jobst Brandt) Newsgroups: rec.bicycles.tech Subject: Re: What
is Loctite and were can I get it Date: 23 Jun 1998 17:54:08 GMT

Frank Krygowski writes:

>> Loctite on a new product is, with rare exception, a major design flaw. Properly designed screwed
>> connections allow enough elastic preload that the part in question that it will never unload the
>> thread in use. Therefore, the tread cannot unscrew. For this reason, critical joints used strain
>> bolts, or other elastic components, to maintain position. Screws that should hold, even if loose,
>> such as brake anchor bolts, should have elastic stop nuts. These have the disadvantage that they
>> resist turning on installation, but for such applications this tradeoff is acceptable.

> I disagree with the statement that Loctite use indicates a major design flaw.

> A successful design is one that achieves its objectives. If the objective is to fasten something
> in place and not have it loosen, what would be the advantage of a strain bolt over Loctite? One
> may be perhaps more elegant, in an engineering sense, but engineers typically don't go for
> elegance as much as for getting the job done.

Controlling the dosage, the cleanliness of the joint and that it gets into the assembly at all, are
major production costs and headaches. As I mentioned, in machine design, products are rated in
motions during assembly with multiplying factors. Low scores are better in assessment of "design for
assembly" and adding any glue, filler or paste carries a high penalty. For this reason, fasteners
that do the job without glue are greatly preferred.

> 50 years ago, Loctite didn't exist. People invented a wide, wide array of devices to prevent
> loosening of threaded fasteners. If every instance of the use of lockwashers, jam nuts, special
> thread forms, etc. indicated a major design flaw in your eyes, you have high standards indeed!
> Higher than almost anyone else, in fact.

Loctite was invented as a fix for otherwise unreliable designs. Its main role is in aftermarket
applications for faulty machinery. There are applications where it is used in lieu of precision
cartridge bearing mounting by having a clearance bore whose gap is filled with Loctite. Even these
are better designed to not be glued.

> While I admit to being something of a retro-grouch, I see Loctite as a useful technology that can
> simplify life both for the designer and for the end user.

You probably mean the lazy engineer and the person who has to use his design. The term end user
evokes Scott Tissue, 400 sheets per roll.
----------------------------------------------------------------------

Jobst Brandt [email protected] Palo Alto CA

 

In uk.rec.cycling [email protected] wrote:

: I don't believe that loosening is even part of the failure mode and that brake force alone
: disengages the wheel even when fairly tightly held, as Lee Bower describes in the referenced
: URL above:

Sorry, just to be clear here:

You think that a QR won't fail via un-screwing, but are you saying that you believe that "brake
force alone disengages the wheel" (as Leee Bower descrbies) or that you don't believe it can?

Arthur

--
Arthur Clune http://www.clune.org Power is delightful. Absolute power is absolutely delightful -
Lord Lester

 

On 4 Apr 2003 07:50:14 GMT, Arthur Clune <[email protected]> wrote:
> In uk.rec.cycling [email protected] wrote:
>
>: I don't believe that loosening is even part of the failure mode and that brake force alone
>: disengages the wheel even when fairly tightly held, as Lee Bower describes in the referenced
>: URL above:
>
> Sorry, just to be clear here:
>
> You think that a QR won't fail via un-screwing, but are you saying that you believe that "brake
> force alone disengages the wheel" (as Leee Bower descrbies) or that you don't believe it can?
>
I wasn't going to get involved here as I'm not an engineer and engineering doesn't start from
"consider a two dimentional infinitely rigid fork"

However, it seems to me that if there is enough force generated to pull the hub right out of the
fork then there is enough to move it so that the nut starts unscrewing. What isn't clear to me is if
there is enough force to push the wheel back, allowing the nut to continue unscrewing with repeated
application of the brakes.

A fast descent on uneven ground is, presumably going to involve the highest peak braking forces and
also the peak forces pushing the wheel back into the fork.

Regards,

Tim.

--
God said, "div D = rho, div B = 0, curl E = - @B/@t, curl H = J + @D/@t," and there was light.

http://tjw.hn.org/ http://www.locofungus.btinternet.co.uk/

 

Peter wrote:

> The second mechanism will also not occur as long as the QR has sharp edges that dig into the
> dropout when it is tightened. A vibration that tends to unscrew the nut from the end of the QR
> skewer would need to raise these edges over the indentations they've made in the dropout.

A lot of modern QR nuts do not have any serrations, eg. Campagnolo Chorus. These are generally
designed for rim-braked road bike wheels with vertical dropouts, but some of the fancy aftermarket
ones sold for general purpose are like this too.

I've used QR's for years - with various levels of tightness - and have never noticed any
skewer rotation but did once find a lever completely undone in the middle of a ride for no
fathomable reason.

~PB

 

Arthur Clune writes:

>> I don't believe that loosening is even part of the failure mode and that brake force alone
>> disengages the wheel even when fairly tightly held, as Lee Bower describes in the referenced
>> URL above:

> Sorry, just to be clear here:

> You think that a QR won't fail via un-screwing, but are you saying that you believe that "brake
> force alone disengages the wheel" (as Lee Bower describes) or that you don't believe it can?

I think that if you followed the thread (in rec.bicycles.tech) you would see that the probability of
wheel disengagement is supported by everything I have said. I have long stated that disk brakes have
problems that are not being addressed by their manufacturers and those who offer them on their
products. This includes misunderstanding of forces, torque, friction, and heat dissipation. The
whole concept.

The customer is the prototype tester.

Jobst Brandt [email protected] Palo Alto CA

 

Tim Woodall writes:

> It seems to me that if there is enough force generated to pull the hub right out of the fork then
> there is enough to move it so that the nut starts unscrewing. What isn't clear to me is if there
> is enough force to push the wheel back, allowing the nut to continue unscrewing with repeated
> application of the brakes.

That involves a different model. If the axle slips from braking force, then it will disengage unless
there is another retention mechanism, such as "lawyer lips", a barrier to further disengaging
motion. In that event, sliding up and own in the leeway of that attachment will most likely lead to
unscrewing.

My contention was that unless the wheel disengages (moves in the dropouts) the QR cannot unscrew.

> A fast descent on uneven ground is, presumably going to involve the highest peak braking forces
> and also the peak forces pushing the wheel back into the fork.

That is consistent with the model of a wheel retained by a lip at the edge of the dropout to limit
wheel removal to opening the QR and unscrewing it as "lawyer lips" require.

Jobst Brandt [email protected] Palo Alto CA

 

[email protected] wrote in message news:<[email protected]>...
> However, the ratio of bolt length to diameter and diameter of but face is nowhere near that of a
> QR. The QR has worked all these years because it is such a good "strain bolt" In fact the
> over-center feel of the closure motion is an indication of how much strain there is in the
> axle-skewer combination.

It's worked well for years on rim-braked bikes, but seems to struggle badly under the rather
different forcing of disk brakes (and vertical dropouts). But I see from your other message that you
do not seem to disagree on the fact or rough mechanism of failure, rather you don't think the
terminology is very precise.

James

 

[email protected] wrote in message news:<[email protected]>...
> That involves a different model. If the axle slips from braking force, then it will disengage
> unless there is another retention mechanism, such as "lawyer lips", a barrier to further
> disengaging motion. In that event, sliding up and own in the leeway of that attachment will most
> likely lead to unscrewing.
>
> My contention was that unless the wheel disengages (moves in the dropouts) the QR cannot unscrew.

I think therefore we are in agreement, which is a relief to me as no-one else has found any
substantive fault with the analysis.

The QR certainly moves frequently enough that such an unscrewing action is not too surprising.
Boltscience refers to it as 'vibration loosening' but the description in this case could perhaps be
labelled 'sliding loosening'. To the injured rider, these semantics are a minor detail.

Unortunately the director of R+D at a major manufacturer who read my web page has indicated that as
far as he is concerned, the lawyers have ruled that this is operator error and so he sees no reason
to be concerned with the design. He found no fault with my analysis either.

James

 

James Annan writes:

>> That involves a different model. If the axle slips from braking force, then it will disengage
>> unless there is another retention mechanism, such as "lawyer lips", a barrier to further
>> disengaging motion. In that event, sliding up and own in the leeway of that attachment will most
>> likely lead to unscrewing.

>> My contention was that unless the wheel disengages (moves in the dropouts) the QR cannot unscrew.

> I think therefore we are in agreement, which is a relief to me as no-one else has found any
> substantive fault with the analysis.

> The QR certainly moves frequently enough that such an unscrewing action is not too surprising.
> Boltscience refers to it as 'vibration loosening' but the description in this case could perhaps
> be labelled 'sliding loosening'. To the injured rider, these semantics are a minor detail.

Well it isn't vibration loosening which is caused by loss of preload under a vibrating load. This is
lateral displacement loosening similar to putting your finger on a plate on the table and sliding
the plate back and forth on a linear path... all the while with a slight bias in one rotational
direction. I think you'll see that even without a biasing torque ( from a thread slope under
pressure) there will be rotation. It is this mechanism that unscrews the QR, not vibration
loosening.

> Unfortunately the director of R+D at a major manufacturer who read my web page has indicated that
> as far as he is concerned, the lawyers have ruled that this is operator error and so he sees no
> reason to be concerned with the design. He found no fault with my analysis either.

That's too bad because others will have to suffer injury and ultimately get suitable legal
representation to bring the bicycle industry to action. This whole subject is part of my complaint
with the technical abilities of the industry.

Jobst Brandt [email protected] Palo Alto CA

 

James Annan writes:

>> However, the ratio of bolt length to diameter and diameter of but face is nowhere near that of a
>> QR. The QR has worked all these years because it is such a good "strain bolt" In fact the
>> over-center feel of the closure motion is an indication of how much strain there is in the
>> axle-skewer combination.

> It's worked well for years on rim-braked bikes, but seems to struggle badly under the rather
> different forcing of disk brakes (and vertical dropouts). But I see from your other message that
> you do not seem to disagree on the fact or rough mechanism of failure, rather you don't think the
> terminology is very precise.

The QR was not invented for this application. It is the disk brake folks who have brought in an
unruly child that cannot be tamed by a QR. The QR is just fine doing what it is supposed to do.

Jobst Brandt [email protected] Palo Alto CA

 

Peter <[email protected]> wrote in message news:<[email protected]>...
> A properly designed and tightened QR should be immune to this problem. There are two ways for a QR
> to be loosened:
> 1) by the lever opening, or
> 2) by unscrewing of the nut from the QR skewer.

I'm not talking about mechanism 1, I agree it is basically impossible.

> The second mechanism will also not occur as long as the QR has sharp edges that dig into the
> dropout when it is tightened. A vibration that tends to unscrew the nut from the end of the QR
> skewer would need to raise these edges over the indentations they've made in the dropout.

It's not the vibration but the substantial direct pull of the disk brake which is the dominant part
of the problem. This pull is demonstrably sufficient to frequently overcome the indentations.

> So even though I have once discovered that my QR was loose in the middle of a ride, I don't
> believe a properly secured QR will spontaneously loosen from vibration.

To put it bluntly, Russ Pinder was equally sceptical until recently. He is now paralysed from the
chest down.

James

 

James Annan wrote:
> Peter <[email protected]> wrote in message news:<[email protected]>...
>
>>A properly designed and tightened QR should be immune to this problem. There are two ways for a QR
>>to be loosened:
>>1) by the lever opening, or
>>2) by unscrewing of the nut from the QR skewer.
>
>
> I'm not talking about mechanism 1, I agree it is basically impossible.
>
>
>>The second mechanism will also not occur as long as the QR has sharp edges that dig into the
>>dropout when it is tightened. A vibration that tends to unscrew the nut from the end of the QR
>>skewer would need to raise these edges over the indentations they've made in the dropout.
>
>
> It's not the vibration but the substantial direct pull of the disk brake which is the dominant
> part of the problem. This pull is demonstrably sufficient to frequently overcome the indentations.

That part I agreed with - and said so in the part of my post that you deleted.

You seemed to be saying that QR mechanisms would be subject to the same type of loosening as regular
bolts which can incrementally loosen through relatively small vibrations that are sustained for a
long period of time. I was just pointing out that QR mechanisms that include serrated edges (which
not all do unfortunately) are not subject to this type of gradual loosening. A sudden jolt from a
disk brake application that suddenly moves the axle an appreciable distance in the dropout is an
entirely different mechanism.

 

The more I see on this the more I find the defense of the status quo stranger than fiction. Why are
writers trying to say that it cant happen? What motivates writers to claim that disk brakes as
currently offered are not a hazard?

The mechanism has been clearly stated, the forces have been identified in magnitude and direction,
and credible descriptions of failures have been presented. What's going on here! There is no easter
bunny. Believe it!

Jobst Brandt [email protected] Palo Alto CA

 

[email protected] wrote in message news:<[email protected]>...
> The more I see on this the more I find the defense of the status quo stranger than fiction. Why
> are writers trying to say that it cant happen? What motivates writers to claim that disk brakes as
> currently offered are not a hazard?
>
> The mechanism has been clearly stated, the forces have been identified in magnitude and direction,
> and credible descriptions of failures have been presented. What's going on here! There is no
> easter bunny. Believe it!

Thank you for an eloquent summary. I hope you don't mind if I quote it on my web page.

James