Dropout Failure Experiments, and a call for Fork Donations

[dogman dan]

I wasn't thinking about the bolt breaking, but rather the effect Justin just demonstrated. The arms ability to pivot at the bolt allowing the lower portion to rotate, and then allow the clamp to slip down the fork. This effect may be less likely if the rotation slips the clamp up the fork? Presumably the fork gets thicker as you go up?

You can go crazy trying to think about all this, that's why try it and see is so valuable. Thanks Justin. 8)

Looking again at the #5 test, the arm is installed on the back, but it appears the rotation force was applied forwards. Now I'm going crazy thinking about it. Which way would the axle want to spin when throttle is put on? The wheel spins front to back, to roll forward. Does the axle tend then to rotate the opposite direction, back to front?

 

[Mark_A_W]

Which way would the axle want to spin when throttle is put on? The wheel spins front to back, to roll forward. Does the axle tend then to rotate the opposite direction, back to front?

The axle wants to spin the opposite way to the wheel.

 

[dogman dan]

Thanks, I can get stuff like that backwards in my mind so easy.

Hmm, then it would make sense to me to mount them on the front of the fork. At least the two piece ones. It would force the strap to slip up the fork, which might be harder to do if the fork gets bigger. The strap when installed rearward, could easily slide down the taper of the fork. On a one piece tourqe arm, It makes sense to me to put the arm to the rear. Then the strap is under tension.

All in all, a good reason to get forks with fender eylets and use them eh? Or weld the tourqe arm to prevent rotation at the bolt.

 

[dequinox]

Justin-

Do you still intend to do the aluminum dropout experiments? I am very curious about these in particular because I have a bike I would like to put a xlyte or 9c motor on...but I believe it has an aluminum fork. A lot of the folks on here say you can't use an aluminum fork at all for a front hub motor. Is that the case with all of them? Do torque arms make it possible at all?

Anyways I am vigorously looking for a new fork for my 700c bike and I would be glad to ship the old one to you for the aluminum experiments if you are still inclined to do them...

 

[justin_le]

Justin-

Do you still intend to do the aluminum dropout experiments?

Hey everyone, yes indeed. Unfortunately I didn't get a chance to do the aluminum fork tests before I left town on Dec 18th, but hopefully by the first or second week in January the experiments can resume. So hang tight for just a little longer. I'm pretty anxious to see how everything shakes down too!

-Justin

 

[dogman dan]

Me too. Just ordered a couple toruqe arms and c washers. Looking at mounting a 9c on a set of alloy rockshocks for 36v or 48v 22 amp use. Some data that I'm not commiting suicide would be nice to have. Still wouldn't run 1000 watts on a front fork that wasn't bombproof.

 

[will_newton]

I am very curious about these in particular because I have a bike I would like to put a xlyte or 9c motor on...but I believe it has an aluminum fork.

check it with a fridge magnet and you will know it it's aluminum or steel.

 

[dequinox]

It's aluminum, I just hadn't got around to checking it until now. Its a manitou magnum fork...pretty decent so I would hate to just replace it in favor of a lame steely fork. Although I intend to turn it into a cargo bike so maybe a solid fork wouldn't be such a bad idea.

I think what I will do is design and grind out some torque plates specifically for the setup...custom fit so that there are no adjustable (and therefore unable to slip) pieces. As for mounting, a full-on "pipe sock" should do it. Two halves of a pipe bolted around the fork above the dropout, with a tapped hole/ear for the plate to bolt to. That should beef it up enough I think.

 

[justin_le]

Hey everyone and thanks for holding out for the past few weeks. I am planning to get back in the swing of things with the torque arm and axle spinout testing this weekend. However, I was thinking about things and I'm not totally sure what would be the most informative way to proceed given that the forks are not in unlimited supply.

Right now, I've got a collection of 8 different alloy and/or suspension forks:



Unfortunately no two of them are identical. If we are mostly interested in comparing how good different makes and styles of fork dropouts are at resisting spinout, then what I should do is run the identical test on each fork, with an axle that has nuts done up to a consistent tightness. Then we could see if some alloy dropouts are considerably stronger than others, or if they are all mostly in the same band.

That will be interesting, but it won't give us any insight into how much improvement we get with torque arms, and how the spinout in the case of aluminum dropouts is affected by different nut tightnesses.

But, if I start doing tests on some forks with torque arms, others with loose nuts, others with tight nuts, we might get interesting data, but we don't have a control result in each case to compare it to. So we wouldn't know specifically how much the additional torque arm or nut tightness contributed versus not having it there. Overall much less scientific, but in some sense I feel that we'll gain more overall information this way regardless.

Opinions on what you want to see me do?

 

[GCinDC]

I'm curious how aluminum dropouts compare in strength, which would argue for identical tests...

But I'm also curious how they compare in different installations, ie, with different axle nut torque, under-tightened and over-tightened. (Is it possible to overtighten a nut on an aluminum fork, causing micro fractures, and leading to failure, or will the nut strip first?)

It seemed from your earlier tests that axle nut tightness was a key variable. I know most don't have torque wrenches, so it'd be great if there were some ghetto way of measuring it, so instructions can be given later, like: use a 6" crescent wrench and stand on it w/ 200lbs. when it stops moving, it's perfect... :lol:

I'd also be curious to know how two similar installations would compare in different temperature extremes!

But I probably won't ever use a front motor, so don't listen to me!

 

[dogman dan]

It's a dilemma. Mabye you could try to categorise them into groups that appear the most similar, and then do tests with one torque arm on one and two on the other. I won't be wanting to run with loose nuts so I don't care as much about that one. We know we don't want to run alloy forks with no torque arm, but is two really better?

Another good question to answer might be done with a high grade steel bolt, seeing if it's possible or likely to damage a dropout simply by applying too much tightness to the nut. It would be good to know if normal torque on the axle nuts might damage the aluminum dropouts. Lots of people ask how tight to make the nuts, and I can only tell em, as tight as possible but don't strip it. I tell em I make em tight, but use a short handled wrench.

 

[will_newton]

You could double your test results by placing the nut on the inside of the fork and testing individual dropouts. for ex. test one dropout with a torque arm and one without.

I wouldn't worry over it much. Just test one and see what the initial results are. If they are as feeble as claimed, then you may want to test them differently than if you find they are stronger than assumed.

I volunteer the Manitou I sent you to be the first victim!

Now, a brief farewell to my forks:

Goodbye old Manitou 3 fork, you traveled the mountains of Colorado, got eBayed to NC for a gentle retirement on my wife's cruiser bike and have been sent to Canada to give your life for science. The e-bikers of the world say, "Thank you".

Justin, when you test the Cannondale fork, just don't tell me beforehand...and make it quick and painless. That fork was one of my best friends for 16 years. Even after the Headshok died, I rode it as a rigid. We have been through hell and high water together. I only gave it up for the pleasure of an e-bike front hub on Surly steel fork. I think I'm gonna shed a tear or two when it's gone. *sniffs*

 

[el_walto]

My vote is for testing all the forks the same. I'm curious how much the forks will hold on their own. Take a picture of the size and shape of the dropout beforehand.

Maybe tighten them all to 60Nm? Or 90Nm if you think that is safe.

 

[davespicer]

But, if I start doing tests on some forks with torque arms, others with loose nuts, others with tight nuts, we might get interesting data, but we don't have a control result in each case to compare it to.

Not if the testing involved both dropouts... might it be feasible to test the dropouts on each fork separately - that is, one side at a time? There may be an obvious-to-everyone-else reason this wouldn't work, but figured it couldn't hurt to ask.

 

[justin_le]

might it be feasible to test the dropouts on each fork separately - that is, one side at a time? There may be an obvious-to-everyone-else reason this wouldn't work, but figured it couldn't hurt to ask.

Dope, this was suggested before and I wrote it off as not feasible with my setup jig for some reason, but I don't think I was thinking straight. Of course this is the ticket!



I took one of the eZee axles and turned one side to be perfectly round so it will just spin within the dropout and cause no torque, while the other side will get the full spinout test. This way with each fork we'll do a control on the first side (I think the 60 N-m nut will be about right), and then flip the test axle around 180 degrees and test the other side with whatever variable we want to experiment with.

The only slight issue I can think of is that on one side of the fork, a spinout will have a tendency to tighten the nut, while on the other side it would tend to loosen. But this is just one more experiment we'll be able to run!

It also turns out that most of the alloy forks will need filing first, their dropout slots range from 9.5 to 9.9 mm. So a bit more work yet but the tests should be taking place throughout the week.

-Justin

 

[Jukin]

I'm a newbie with eBikes, but have you tested, if this kind of locking washers: http://www.nord-lock.com/default.asp?url=2.16.37 can do some improvement at low nut tightening to the axle spinout. I have those installed in my eBike (front 24V250W with regen) and I haven't noticed any loosing of nuts, yet. I can send some pairs of washers to you, Justin, if you want to test these. (English is not my native language, as you can see)

 

[justin_le]

So the process of testing one single dropout at a time using the rounded axle worked great. I also changed some techniques on the video data recording process so that the data is more uniform and consistent for smoother looking curves.

And the results are interesting.

Test #6 was a Suntour brand suspension fork, with cast aluminum dropouts, 7.4 mm thick with a somewhat larger than normal 10.3 mm slot width.
View attachment 3


I decided with each fork to do one side with a nut tightened to 60 Newton-meters as the 'control' reference, and then the 2nd side to play around with variables. In this first case, I just wanted to get a sense for where the failure point was with a loose nut by comparison to the tight one. Here's a graph of what I got:



In both cases, the dropout failed catastrophically just before a 20 degree rotation of the axle. With the loose nut it rose steadily to just under 40 N-m, while the tight nut got to about 65 N-m before cracking.

The crack location in the first case with the tight nut was as we normally see, the front end of the dropout busted off. When I did the loose nut case, I actually was applying torque in the regen direction, not thinking it would really make any difference. What happened here is that the crack happened through the much thicker length of metal on the rear side of the dropout:

 

[liveforphysics]

Outstanding work Justin. I think it's equally important to notice the difference in failure mode between the aluminum fork vs the steel forks. The aluminum would be zero to minimal warning symptoms, then bam! The wheel falls off.

 

[justin_le]

The next suspension fork to test was an "Answer Manitou" brand.



What interested me about this unit is that the aluminum dropout was clearly CNC machined from an aluminum block rather than cast:



The dropouts in this fork were a lot thinner, just 5.7mm with no lawyer lip or thicker region, and the slot was originally just 9mm wide so I had to file it open to 10mm in order to do the tests.



The results are totally different than the thicker cast aluminum fork. Rather than rising rapidly to a peak torque and then cracking catastrophically, the the torque here was quite a bit steadier as the axle smeared the aluminum apart, and the dropouts never broke, they just opened up as we see all the time with steel forks.

View attachment 1


Even though it didn't fail catastrophically, the torque values are much lower than with the cast dropouts, peaking at just 16 N-m for a loose nut, and 37 N-m for a tight nut. The Suntour forks by comparison peaked at 39 and 62 N-m, respectively.

In both these cases, the action of having a 60 N-m tight axle nut versus a loose one seems to be about a 20 N-m improvement in how much the axle can resist spinout.

 

[justin_le]

My next fork to try was a 'Logan' brand, that had a very similar look and characteristic as the Suntour unit. Only difference was that the dropout slot was about 9.8mm, so I had to file it slightly to get the axle in:

View attachment 3

In this test, I wanted to measure the effect of having a torque arm in the system. So after doing the 60 N-m control side, I did the 2nd side using one of the Rev3 torque arms that has a 3/16" thick stainless steel torque plate for the axle hole. The nut over the torque plate was also tightened up to 60 N-m:



The results are at least a little encouraging. With no torque arm but a tight axle nut, the dropout cracked at 54 N-m when the axle had been rotated by a mere 6 degrees, quite a bit less rotation that the Suntour dropout handled before cracking ( 20 degrees), but a fairly similar torque value.



With the torque arm installed, the axle reached a peak of 93 N-m at 10 degrees of rotation. At this point, the dropout cracked resulting in the sudden drop in the graph. But there was still substantial spinout resistance coming from the torque plate alone, and as the rotation continued that reached a 2nd peak of almost 75 N-m

In the end, the failure mode on the torque arm was not the hose clamps or the 2 pieces sliding, but the axle simply spinning inside the cut plate. Both the threads on the axle and the insides of the torque plate were damaged.

View attachment 4

 

[justin_le]

The next to try was a Cannondale Headshok. This fork appears to use a piece of aluminum plate metal for the dropout that is beautifully welded (as Cannondale is apt to do) to the aluminum tubes. This contrasts to both the cast and CNC machined dropouts in the previous units. The dropout plate width was 6.7mm, a little thicker than the 'Answer Manitou' brand, but less than any of the cast dropouts:



I wanted to test the fender eyelet style of torque arm here, because I had some concern that the two pieces would tend to slide through the elongated slot until it bottomed out, allowing a fair bit of axle rotation before the full effect of the torque plate comes into effect.

View attachment 2

The results are pretty neat in the context of the previous graphs. 60 N-m axle nut on the control side showed a peak spinout torque of 54 N-m, with a fairly flat torque/angle profile as the axle was rotated and no cracking of the dropout.



With the torque arm in place, the peak reached a full 116 N-m with a large sweeping spinout curve.

The damage to the Rev2 fender eyelet torque arm was exactly like we saw with the double hoseclamp design, there was no 'relative motion' between the two pieces, and instead the entire failure as from the axle cutting into and smearing the 10mm slot in the torque plate.




This was the last test I've had time to run so far. You'll notice that the actual axle torque required for failure in these aluminum forks is in fact higher, in some cases substantially so, than it is in the thinner steel GT forks I tested in december. Each of the above graphs is for just one dropout plate, so multiply the value by 2 to compare it with the graphs of the steel dropouts.

There are quite a few other conclusions we can draw from all this, this most significant of which is that there is a huge range in the strength and spinout resistance of different types and brands of alloy dropouts.

There are still 3 more forks left, and I'd like with those to quantify the effect of the thinner 1/8" torque arms that are pretty common, as well as see how a thicker 1/4" torque arm would do. I'm also curious to see just how tight we can make the axle nuts on an aluminum dropout before the dropouts crack, with no inside washers, to see if this is really a concern or not.

Justin

 

[dogman dan]

Great work and thanks. Now that I'm running an alloy fork and double torque arms with c washers, with 1000 watts, I'm real glad to see that the torque arms really will help once the dropouts crack. Surely they will someday, but I'm confident now that the wheel will stay on the bike. Just need to make a habit of checking for a crack so I don't ride for months with one.....

 

[biohazardman]

Thanks Justin appreciate all the werq and sharing the results with us. I was surprised to see the axle and torque arm round out are either of them heat-treated? Still it took more than twice the torque to do it so proof that torque arms do werq.

 

[dogman dan]

Yeah, enough tourqe to round out torque arms may mean your real worry is how to land after flying over the handlebars anyway. As long as the motor doesn't normaly produce that kind of torque, I should be as safe as any front hubbie. It looks like a pair of torqe arms will be good enough for what I'm running, 9c at 48v 20 amps.

 

[chvidgov.bc.ca]

Would the results of single-dropout-at-a-time type testing be the same with two dropouts installed as per normal - ie. is the total torque applied distributed equally over the two dropouts, leading to a larger safety margin than the above tests would indicate? This would seem reasonable?