Dropout Failure Experiements, and a call for Fork Donations

[ZapPat]

I hate to admit it, but I'm starting to eye my giant mtb, with the alloy 120 mm travel forks for a light gearmotor. Must resist temptation!

I've done quite a bit of riding with a cheap alloy front suspension fork, dogman (it's a low end 100mm RST I got for a buck). No incidents yet even with lots of off-road riding... but I do have two of Justin's 1/8th torque arms on there (it's a 9C hub). I even use regen all the time for braking and it has never loosened up yet after the normal initial break-in period of a few rides. Mind you, that bike also has a rear 9C hub too, so maybe this releaves the front fork from doing too much work by itself?

Thanks for the great tests, Justin!

[dogman]

Part of the concern about riding a motor on aluminum is the terrian I have in mind. Think going down a rock staircase with steps 16" tall. Not a place to go over the bars. A lot of the trails I see on you tube look quite cushy compared to my local desert rock piles with cactus.

But the forces a good tourqe arm can take do give me a lot more confidence in thier usefullness. I really had the opinon that they weren't so effective before.

Forks have been sent, it was a bit expensive but worth it for SCIENCE!

[jag]

#	Material	Thickness	Axle	Yield Force	Yield Torque
1	mild steel	1/8”		12mm	8.1 kg		24 N-m
2	stainless	1/8”		12mm	7.8 kg		23 N-m
3	stainless	1/8”		12mm	10.9 kg		32 N-m
4	stainless	3/16”		12mm	18.8 kg		55 N-m
5	stainless	3/16”		14mm	40 kg*		118 N-m*
6	mild steel	1/8”		14mm	12 -16 kg	35 N-m

These torque values aren't very encouraging. Many hub motors do 100+nm.

Besides the thickness, the material plays a role. Tools made from crome vanadium alloy are much stronger than mild steel. What's the max torque of a good quality 10mm wrench? Maybe better material in the torque arms would be a way to go.

On my ebike I'm lucky to have 8mm thick forged dropouts. Probably the torque washers and arms don't add much to that.

[justin_le]

Forks have been sent, it was a bit expensive but worth it for SCIENCE!

Hey Dogman, that's super. We also got a paypal donation to go toward these tests so thanks for that whoever it was.

Otherwise though, there hasn't been much response at all as far as aluminum forks go. There is the one unit that Dogman is sending, and another one from a pretty decent downhill MTB with shot seals that my room-mate donated, so that makes two at least. I'll be doing the rounds of bike shops around Vancouver over the next week or so to see what we can collected as far as discarded alloy forks that still have dropouts in tact, and expect to start doing all the spinout failure tests at the beginning of December. So sit tight!

[justin_le]

Nice work Justin. I'm really impressed with the effectivness of the thicker torque arms! I've been of the opinion that the thin ones were decorations, but clearly 3/16 makes a huge difference.

Well, I think the results more or less confirm what had generally been figured out the trial and error way: that torque plates made from 1/8" steel ARE pretty much decorative, as they don't in-and-of themselves have enough strength even with 14mm axles to resist the stall torque of most hub motors.

The effect of going to 3/16" was impressive, and I see no reason now why we wouldn't go further to 1/4". The reason I didn't get plates made from 1/4" stock before was that as the metal gets thicker, the accuracy of the bore made with the waterjet cutter gets worse and worse, so the tolerances suffer (we saw this even in the difference from 1/8 to 3/16, and had to hand-file a lot of them). However, since we've seen that a really snug fit isn't nearly as beneficial as the extra thickness, then it's not so much of a problem anymore!

Besides the thickness, the material plays a role. Tools made from crome vanadium alloy are much stronger than mild steel. What's the max torque of a good quality 10mm wrench? Maybe better material in the torque arms would be a way to go.

I will be picking up a 10mm wrench on my next hardware store visit and will do a spinout test on that too, since it seems to be a popular makeshift arm.

Justin

[Ypedal]

Likely different in different locations but here in Moncton the local landfill has a dumpster bin dedicated to bicycles, when full they crush them into a cube.. not sure how willing they would be to let someone do this but with a cordless sawzall or a regular with an inverter you could quickly hack apart the forks instead of wasting a bunch of time fighting with rusted bike parts.

[mwkeefer]

Dogman,

Earlier I noticed you relating your intended test to riding down a staircase of 16" stairs.

I can attest that i have ridden my 26" Hard Rock not only down the Philadelphia art museum steps but, I rode down after I had ridden up!

This is with a Rock Shox Tora RS (with Lockout) I believe they are 100mm travel forks aluminum alloy (may be magnesium, not 100% sure) 100% unmodified (I ground the axle to fit the fork) and using nothing more than a single of Jason from ebikekit's torque arms.

Motor was 9C FH205 in 26" rim -after 100mi of breakin riding and re-adjustment
Suspension was set for max travel/give

Ride up required some tactical pedaling to maintain traction, and honestly was more like bunnyhopping 100s of curbs in a row - I guess this is how mountain bikers traverse rocks, I've seen videos but this was my first real attempt at anything resembling rockhopping - until then I had stuck with just crazy offroad trails. It wasn't super fast going up (adding a rear hub would have helped that considerably).

Coming down.... that was fun... About 4x as fast as going up, have no idea how quick but the suspension ate up the steps like they werent even there... actually it felt like a canter for anyone who rides hoarses, only way I know to describe it.

I did wear a helmet for this but didn't get any video... I only did the run once for a test (figured if somthing broke, like me... I didn't want that idiocy on video), police showed up before I could do a second run and I wasn't about to push my luck that night.

I ride the same bike now with 100s more miles on it and im still running through the woods, hopping (more like running over at a good clip) curbs, etc.

No fork fatigue or breakage (yet).

-Mike

PS: I will at some point... do the run again on video, I may have to do it at night though to avoid trouble... would be even better if I could get a cop to "dare" me, since they are all interested in the eBike and want to try it... I could play it off as if I was demonstrating how useful it could be to a bike officer? Actually... that might work. = )_

[mwkeefer]

Guys,

What types of Aluminum forks are you looking for to do testing on?

I know I can get you 3 or 4 models of Sr Suntour and some others I haven't a clue the manufacturer but are there some specifics your lookign for... models, years, with/without disc mounts?

Where do they need to be shipped to?

I thought I had posted already, I can and will source as much as possible for this purpose... less failure = less physical injury = better public perception of ebikes and better safety for all of us.

I also see the torque ratings above... kinda scary.

I can't remember the hardening level on the torque arms carried by ebikekits and justin from ebikes.ca but I have tested them well over 3000w without modification and now > 5000w with a JB Weld filler (to take all slack out of them - they had perhaps 1/2mm of play in the fit prior to the JB Weld fil) - it didn't crack either (I expected it might).

Hope it helps!

-Mike

[icecube57]

Im not plugging or anything but the test data is making me feel a little bit more at ease with my GM front motor. I have two ComCycle Torque arms. They are 3mm thick. I doubled them up on one side of the motor. This is a total torque arm thickness of 6 millimeters = 0.236220472 inches which is almost 1/4 inch. With my current setup fresh off the charger my dropouts and torque am see about 95N stall force. Although youve mention that torque arm snugness doesnt matter. Mines super snug. I almost had to file them. The axel is 14mm .

Have you did data to includes torque arm length. Some torque arms are long like mine and then some are very small that you can use the rack mounts to secure them. Also are there any weakness in using a solid arm vs an adjustable one... like the ones that have the any angle cog in the middle of them.

[steveo]

since we are on this topic; I thought i might show you my toque arm ..

These are from a set of steel trainning wheels ... the slot in them was perfect size for my x5 shaft...








-steveo

[justin_le]

Guys,

What types of Aluminum forks are you looking for to do testing on?

I know I can get you 3 or 4 models of Sr Suntour and some others I haven't a clue the manufacturer but are there some specifics your lookign for... models, years, with/without disc mounts?

Hi Mike, and thanks for the offer. I did the rounds to local bike stores yesterday and actually scored much better than I was expecting. In suspension forks, we've got vintage Rock Shocks (that are actually magnesium), SR Suntour CR850, RST 281R, an old BAM quality downhill fork, and a pretty cheap Logan fork. With these plus the units that dogman has sent and the various chromo forks collected previously I think we've got a good representative sampling.

Forks on floor.jpg (47.29 KiB) Viewed 1546 times

What would still be nice though is if we can get a couple suspension forks of identical model, so that we can test the same fork with/without a torque arm for instance, and quantify the different that is made without the different fork being a variable. Would these SR suntour models be of the same family as far as their dropout goes? I've already got the one CR850 so any more of those (or equivalents) would be really useful.

Justin

[Affliction]

I think it's great you are failure testing all these forks. One question tho, how are you going to account for the effect that regen braking has on the dropouts? If you tourque a dropout in one rotational direction only it will fail at alot higher torque than a fork fatigued at two rotational torque directions. This is why I stopped using electric braking. I estimate the failure rate at 10 times more if using E-braking. It was mentioned that thicker torque arms worked better but I think the effect has to do with more than the thickness. The proper torque of the nuts and a proper 10mm keyway is the key to this. If you hold the axle in only one axis of clamping force then it rips open the dropouts; if you clamp the axle down by the other axis by the axle nuts with improper dropouts then again it rips the dropouts. The solution to this is indeed 3 dimentional.
All metal is maluable so the goal is to prevent deformation on all axis. Think of a cube... if you deform two sides you have to increase pressure on the other 4 to correct this. A dropout is no different. Thickness prevents spreading; 1st axis. tourque of the nut prevents the same in the axis equal to the axle; 2nd axis. and the combined forces of the two prevent the axle from spinning out of the dropout; 3rd vertical axis and this is due to friction from the other 2.

[justin_le]

I think it's great you are failure testing all these forks. One question tho, how are you going to account for the effect that regen braking has on the dropouts?

Hey Ben, in this set of experiments we won't be accounting for that. I am very explicitly testing only the failure strength of the dropouts (and torque arms) when stressed to the limit in a single direction. The strength I am sure would be identical in the reverse direction. But the situation of regen in an ebike where you are repeatedly torquing the axle one direction and then the other direction, causing it to wiggle back and forth, loosening the nuts, impacting the torque plate. etc. etc. is a totally different set of tests, and may be subject to another thread here in the future.

[mwkeefer]

Justin,

Nice find on the assortment of stock shocks!

Let me know which you need dupes of (if you can't source them lbs).

You are dead on about regen too... the issue is more that the axle on most hub motors have the flats at 9.94mm and the keying of the torque arms and torque washers is 10mm... .06mm may not sound like squat but its the wiggle room provided by that small absess (or a hub which wasn't properly installed - square dead on, flush in the dropouts) that can cause higher dropout (front and rear) failure rate on regen equipped eBikes...

Since non regen is only loading up in 1 direction (for the most part, exception normal braking) of force if the torque nuts are tight enough it may slip that .06mm but in a proper install it won't because it's preloaded, tensioned, positioned - whatever into the loaded position and has no where to go under highest force (dead stop to WOT with load).

1mm of proper material, adequate in length (perhaps 3-4") and with a 9.95mm flat keyway - were talking hardened stainless here.. no more problems with any dropouts (methods, liveforphysics, ypedal, dnmun, myself, etc excluded - comeon were idiots running 3kw and up all resting on what 5mm of total overlap (the girth of total clamping surface holding shit together) to an alloy frame - ha

Now I'm thankful, to be alive (I must remember to stop thinking).

Happy turkey day!

-Mike

[patrickza]

With the unmodified tight fitting plate, the metal started yielding at 10.8 kg and pretty much gave way completely at 11kg. That's a tad over 30 N-m.

With the loose fitting torque plate that I filed larger so that there was a 0.2mm gap between the axle flats and the opening, the metal started yielding at 10.4 kg, though it didn't give way completely until almost 12kg.

So that's that, not much difference really between a tight as can be fit, and what the average person might whip up with a drill press and file. There is nothing wrong with making it super snug, but it's not AS crucial to the spinout strength as some posts would suggest.

Justin

This is a fantastic thread, I've had this question in my mind for quite a while now as I have a fraction of play in my torque arms and I've wondered what effect that had. Hopefully I'll be good with a 6mm arm on each side! Let me know if I can send a few dollars to your regular paypal account as for the cycle analyst, or if you have another one I should send it to?

[will_newton]

Justin, I have a Manitou 3 suspension fork with cast aluminum dropouts, if you are interested. It's old and doesn't work anymore and would be a good test of what has been deemed the worst kind of fork to use. Send me a PM with shipment info and I'll send it along at no charge.

[ZapPat]

I think it's great you are failure testing all these forks. One question tho, how are you going to account for the effect that regen braking has on the dropouts?

Hey Ben, in this set of experiments we won't be accounting for that. I am very explicitly testing only the failure strength of the dropouts (and torque arms) when stressed to the limit in a single direction. The strength I am sure would be identical in the reverse direction. But the situation of regen in an ebike where you are repeatedly torquing the axle one direction and then the other direction, causing it to wiggle back and forth, loosening the nuts, impacting the torque plate. etc. etc. is a totally different set of tests, and may be subject to another thread here in the future.

I wonder where you found that 10X dropout failure rate when using ebrakes, Ben? This figure seems very high unless you are talking about people using no torque arms at all, and/or not checking their axle nut tightness after installation.

I have been using very intense ebrakes on one bike for a couple months now and have had absolutely no issues with the dropouts, even with the front aluminum alloy suspension fork (but I don't go down 16" steps like dogman!). Mind you I use dual torque arms on both front and back motors - the simple ones come free on 9C hubs now anyways (two per hub), so why would anyone not use them?

One important thing I've found with aluminum dropouts is that there is an initial settling in period after installation, and so it's very important to check the axle nuts during the first few rides. I had to tighten mine up a few times, each time less and less, and now they stay nice and tight even when torque-abused back and forth. Steel dropouts tend to do this a bit too, but much less it seems.

[dogman]

The settling in I've experienced is the washer deforming a bit. It's inevitable since there is a gap in the bottom of dropout that the washer tries to squeeze into, and it seems to happen on steel too. Sometimes a washer that is slightly big settling into a cup in the dropout can spread the drops some. This can cause the settling to take longer. I hope that isn't what is happening with your aluminum forks, if so, you may have a micro crack in em now. Steel will just bend a bit and keep it's strength better. V shape drops, nuts that keep loosening, and axles that try to spit out as you tighten them are the red flag warnings.

I think the effect of regen is meaninless if the fork, washers and motor fit properly. Tight in one direction is tight in both directions. Able to wiggle is loose, and just as bad with or without regen. No data or experience to back up this statement, just my dog sense. Tourqe arms of course, can save your ass if loose does occur!

[Jaybird]

Hi I am totally new to ebikes I have just bought this kit http://shop.crystalyte-europe.com/produ ... 249&page=1 I was hoping to install it on my Dahon Jetstream P8 front forks I would be grateful for any advice I was wondering if the torque produced by the motor would be to much for the Jetstream P8 front suspension forks or would I better installing it on a Dahon folder with ordinary forks many thanks for any advice

[Affliction]

I wonder where you found that 10X dropout failure rate when using ebrakes, Ben? This figure seems very high unless you are talking about people using no torque arms at all, and/or not checking their axle nut tightness after installation.

I have been using very intense ebrakes on one bike for a couple months now and have had absolutely no issues with the dropouts, even with the front aluminum alloy suspension fork (but I don't go down 16" steps like dogman!). Mind you I use dual torque arms on both front and back motors - the simple ones come free on 9C hubs now anyways (two per hub), so why would anyone not use them?

One important thing I've found with aluminum dropouts is that there is an initial settling in period after installation, and so it's very important to check the axle nuts during the first few rides. I had to tighten mine up a few times, each time less and less, and now they stay nice and tight even when torque-abused back and forth. Steel dropouts tend to do this a bit too, but much less it seems.

I came up with the 10X figure as a guesstimate partly from my own experience with e-brakes and accounting for the fact that people tend to be inherantly lazy. And yes this was for no use of torque arms. Nuts will loosen at even the slightest bit of back and forth movement. e-braking creates the possibility of this movement where driving the axle in only one direction does not. Thus my increased calculation of failure.
Alot of hubmotor kits don't come with torque arms and unless you didn't know any better you wouldn't think of getting any.

[dogman]

Re, jaybird.
The wattage is pretty low, 250, so it should be fairly safe even if the forks are alloy. But the install needs to be absolutely perfect for it to work. Alloy forks often have a cup on them for retaining quick release hubs that get loose. The washers may need to be modified to fit into the cup, not span over it leaving space underneath. And get a tourqe arm. If you get the tourqe arm from Justin at Ebikes ca you can get a special washer from him for solving the cup problem.

[Affliction]

Justin, I had this idea in the past but I never got around to making a stamping die.

The flat surface where the axle rests against the dropout opposite the nut is always smooth on hub motors.
I've frequently seen on regular bike hubs where the mating surface on the inside of the dropout has "teeth" on the hub to bite into the frame for retention. The metal on hub axles is soft enough for easy rethreading; I repair stripped axle shafts all the time for Ridemore.ca by rethreading from 12mm 1.25 to 12mm 1.5 thread pitch. My idea is to make some hardened die's with the negative spline and using a press, imprint teeth onto both sides of the axle where it would meet the frame. Having two die's would make this a 1 step 30 second process you could do to motors before shipping.
Please stamp a few axles this way and test the improvement with properly torqued nuts.

[justin_le]

Justin, I had this idea in the past but I never got around to making a stamping die.
The flat surface where the axle rests against the dropout opposite the nut is always smooth on hub motors.
I've frequently seen on regular bike hubs where the mating surface on the inside of the dropout has "teeth" on the hub to bite into the frame for retention. <snip>
Please stamp a few axles this way and test the improvement with properly torqued nuts.

Hey Ben, this is a SUPER idea! I have a feeling that a radially grooved imprint on the ends of the axle could go a really long way in locking the axles against spinout. And unlike most torque arms, this method intrinsically has no slop between forwards and reverse torques, so it could prove especially beneficial with regen systems.

The only slight problem is that a lot of the newer motors (BMC, eZee, Nine Continent) have pretty small diameter 17mm axles. That doesn't leave nearly the same amount of contact surface as the Crystalyte 400 / WE style hubs (20mm) or the 24mm axle diameter on a 5304. See image below for the comparative difference.

Shoulder Diameter of different hub motor axles

Axle Diameters.jpg (45.58 KiB) Viewed 1237 times

I'm doing all these tests with the 17mm eZee axles, where the effects of a knurled end would be the least, but I'll do it all the same. I should be able to score a set of radial grooves in the end of the axle in order to do this experiment, and then we can see if the improvement is worth investing in a die for or not.

Justin

[justin_le]

Went to a sports junkies (a used/liquidated sporting goods store) the other day, and got this:

7 Steel GT Forks

Yellow Forks.jpg (22.49 KiB) Viewed 1236 times

That's seven IDENTICAL GT steel forks. The paint is a little dinged up, but it doesn't look like they have ever been used on a bicycle, as there are no signs of paint wear or chipping at the actual dropouts where the wheel would have been installed.

Closeup of the dropout

Yellow Dropout Closeup.jpg (23.51 KiB) Viewed 1348 times

The shop tech was interested in what I was up to (turns out he has an ebike himself!) and helped me get the entire lot for just $25.

Anyways this means that we can do a set of super controlled experiments quantifying the exact effect that different torque arm designs, axle grooves, nut tightnesses etc. have on the spinout resistance of a motor axle, at least as it applies to steel dropouts. I still don't have any exact duplicate aluminum suspension forks, but probably the lessons that we get in the steel case would apply more or less.

[justin_le]

I also finished making the axle clamp and lever arm tonight for performing the actual spinouts. The eZee axle has a keyway in it, (well, two actually one for the stator and another for the freewheel), so I just cut a circular notch in two pieces of 5/8" steel bar stock with a matching groove for a short piece of keystock:

eZee Axle Clamp Arm, pieces.jpg (15.75 KiB) Viewed 1234 times

eZee Axle Clamp Arm, assembled.jpg (12.52 KiB) Viewed 1234 times

So everything is in place and I was going to start the tests on all the steel forks this weekend.

But then I realized it would make sense to get more than just a single data point with each test. Previously we looked at the load cell and recorded what seemed to be the maximum value that showed up as the lever arm was pulled all the way down. But when there was a torque plate installed, there wasn't a clearly defined "failure" torque, at some point the dropouts clearly started to give, but it took an ever increasing amount of torque to make the axle actually spin around a full 180 degrees. So it was a little bit subjective.

I'm going to see what we can setup to generate a full spinout profile for each experiment, so we can see the torque on the 'Y' axles plotted against the axle rotation angle. This will be a lot more informative, comparing for instance how steel gives while alloy cracks, and will eliminate some subjectivity as well, but it'll take some time to figure out the best way to set things up.

Justin