Breaking spokes on GMAC laced in 20" cargo bike wheel

[EwanC]

Hi all,

I have a Yuba Sweet Curry longtail cargo bike. It has a 20" rear and 26" front wheel. I converted it with a GMAC (taken from another bike) in the rear wheel, laced into a DT Swiss U 663 rim and Sapim Strong single butted spokes (13g-14g; 115mm long as per Grin spoke calculator). It's laced in a single cross pattern with elbows inside on the drive side, and elbows outside on non-drive-side (this seemed to be the most common for 20" GMAC cargo bike builds in Grin's gallery). The wheel was built for me by a local bike mechanic who has lots of experience building wheels, but not much with hub motors (they're not very popular around here! everyone buys pre-built mid drives). He did comment at the time that he thought the spokes were a couple mm longer than ideal and did have to manhandle them a bit to build the wheel up. So there's a slight bend in them - visible in pics. Plucking the spokes results in a ~high soprano tone (I don't have a tension gauge and the spoke are too short to get a TM-1 or similar on there anyway). The wheel has ~1400km total on it.

I recently did a 500km tour with this bike towing my solar trailer (build thread if interested, though I owe you all an update!). I had my 18 month old + panniers on the back, and the trailer was probably 30kg. Plus me at ~90kg and the total system weight is probably 180-190kg. It was on the Alps2Ocean cycle trail plus some back roads. The terrain was mostly pavement or shingle roads/bike tracks, but there was ~30km of kinda rocky terrain at the limit of what I'd do on a cargo bike (included a couple pics showing worst terrain). When I got home, there were 5 broken spokes. I didn't notice them on the trip because they remained wedged up against the hub flange so made no noise. The spokes were all broken at the hub end, all on the drive side, and all were the "outermost" spokes in the pattern (not sure how to describe that, but you'll see what I mean in the pics).

I really like the bike + GMAC system as it's a hill climbing beast with great regen. So I'm keen to rebuild the wheel and make this work. Other than using slightly shorter spokes this time (I was thinking 113mm), does anyone have any advice for what I could do to make this wheel stronger when rebuilt?

Should it have all elbows out, or perhaps in an over-under pattern on the drive side? Grin's wheel building page mentions that if spokes are not snug against the hub flange, that can be a cause of spoke breakage. This is the case in my wheel on the drive side (see pics - visible gap between spoke and hub flange on drive side, but spokes are snug on non-drive-side). Though when lacing with elbows inside, isn't there inevitably going to be a gap between the straight part of the spoke and the hub flange? Yet it's common to build hub motor wheels with elbows inside, and even required for the All-Axle motor. So I'm missing something here.

Should I source and use a rim with angled spoke holes? So that the spokes don't have to bend so much (calc says spoke angle is 75°, not ideal). Maybe a Ryde Andra 40? I'd have to order one of those from Europe and I have no other use for the U 663 so that's not ideal, but if it'll result in a meaningfully stronger wheel then so be it.

Thanks in advance for any suggestions!

 

[E-HP]

Did he use the Sapim polyax nipples?

I used them in lacing my 24” rim with my DD hub, and still get some bend. Has to be worse for 20”. Eventually I may go to a offset radial lacing.

The edge runner long tail cargo build doesn’t appear to have any bend, or very little.

 

[Chalo]

It's laced in a single cross pattern with elbows inside on the drive side, and elbows outside on non-drive-side

That's almost certainly the reverse of what you want. The entire point of putting spokes all on the same side of the hub flange is to add or subtract lateral bracing angle. On a rear cassette hub, the bracing angle is always flatter on the right side, so to help equalize bracing angle and spoke tension, you put right side spoke elbows outside the flange and left side spoke elbows inside the flange.

Front hubs with provision for a disc brake are usually offset the other way, so you'd put all the spoke elbows on the left side in that case.

When bracing angle is very unequal, not only does it make the wheel weaker, it means that either the flatter side spokes are too tight or the more conical side spokes are too loose, or both.

Hub motor wheels are difficult or impossible to stress relieve by the more common squeezing method (as recommended by Jobst Brandt), and crossed spoke pairs must be pried with a tool (like Sheldon Brown) to accomplish the same thing. If this wasn't done correctly to your wheel, then that would account for chronic breakages. That all the broken spokes are on the right side is most likely a result of highly elevated tension on that side to support the wheel's exaggerated dish.

EDIT:
If the GMAC is only offset 2mm to the left, that is a pretty minimal amount of dish to compensate. But it still means that the all-left-side lacing you have is in the wrong direction. For a 20" wheel implementation, it's about the same to have all the elbows inside on both sides, or all the elbows on the right sides.

 

[Papa]

it's laced in a single cross pattern with elbows inside on the drive side, and elbows outside on non-drive-side

If memory serves, the Gmac is offset -2mm to the left of center. This is why I laced drive side elbows outside the flange, and left side elbows inside. Mine is laced single cross to a 20" DM24 .

edit: I just ran Grins spoke calculator. With your lacing, the spoke "Tension Ratio" (lower RH side of the calc screen) is 40/60. Mine is 50/50.

 

[amberwolf]

I don't have a full pic of my Grin-laced 20"-wheel GMAC (have to go take one) but if it is helpful, this shows the cover-side flange and elbows, and some of the rim/nipples/crosses

 

[EwanC]

Did he use the Sapim polyax nipples?

I used them in lacing my 24” rim with my DD hub, and still get some bend. Has to be worse for 20”. Eventually I may go to a offset radial lacing.

No - he used the standard Sapim Strong nipples (14mm long). Neither of us knew about the Polyax nipples. They don't seem to be available from the NZ Sapim distributor (neither are Sapim Strong 13g-14g single butted spokes - we have to get those from Grin). That seems like a good idea though given the 75° spoke angle so I'll order some in from overseas and use them. Thanks.

The edge runner long tail cargo build doesn’t appear to have any bend, or very little.

Do you mean this one? That does seem to have a bit less bend in the spokes, yeah, though some is still visible. I wonder if they used a 20" DM24? Grin says on their product page for that rim that it has "3D spoke hole drilling". That sounds like it might have some angle in the spoke holes, a bit like the Ryde Andra 40, to help reduce this bending problem with large hubs. It's not mentioned on the Alexrims product page though.

 

[EwanC]

That's almost certainly the reverse of what you want. The entire point of putting spokes all on the same side of the hub flange is to add or subtract lateral bracing angle. On a rear cassette hub, the bracing angle is always flatter on the right side, so to help equalize bracing angle and spoke tension, you put right side spoke elbows outside the flange and left side spoke elbows inside the flange.

Front hubs with provision for a disc brake are usually offset the other way, so you'd put all the spoke elbows on the left side in that case.

When bracing angle is very unequal, not only does it make the wheel weaker, it means that either the flatter side spokes are too tight or the more conical side spokes are too loose, or both.

OK, thanks, we'll rebuild it with the elbows around the other way - outside on drive side and inside on non-drive-side. Calc says that'll give us a 50:50 tension ratio instead of the 40:60 ratio it had before.

Hub motor wheels are difficult or impossible to stress relieve by the more common squeezing method (as recommended by Jobst Brandt), and crossed spoke pairs must be pried with a tool (like Sheldon Brown) to accomplish the same thing. If this wasn't done correctly to your wheel, then that would account for chronic breakages. That all the broken spokes are on the right side is most likely a result of highly elevated tension on that side to support the wheel's exaggerated dish.

Noted. I assume he used the normal squeezing-of-pairs method for stress relief when tensioning - will ask about prying them with a tool instead.

 

[Chalo]

OK, thanks, we'll rebuild it with the elbows around the other way - outside on drive side and inside on non-drive-side. Calc says that'll give us a 50:50 tension ratio instead of the 40:60 ratio it had before.

Doing that can only help, but 60:40 tension ratio is a lot better than most rear wheels. Tension imbalance of that degree is unlikely to be a root cause of your spoke breakages. I was thinking idly about your wheel yesterday at work, and it occurred to me that if none of the elbows-out spokes have broken, it's possible you could cure the problem by lacing them all elbows-out.

 

[PaPaSteve]

Here's an excellent e-Bike wheel building video.

As has been pointed out the spokes are on the wrong side of the flanges.
Another possibility for the failure could be the wider thickness of the hub flange causing the spoke breakage.
Since your being forced into a complete wheel rebuild try to straighten out the nipple angle too.
Nipples should be collinear to the spoke angle. (lying in the same straight line)
The normal fix to this problem is to file the nipple hole in the rim to the correct angle.
Because the DT Swiss U 663 rim has eyelets with the wrong angle the only solution, and this will sound brutal, is to use a hole alignment punch to bend the metal to the correct angle.
Or change to radial spoke pattern (no cross).
Picture of a hole alignment punch :

 

[offGridDownUnder]

use a hole alignment punch to bend the metal to the correct angle

Although my suggestion will sound odd at first, it's one I would try.

You mentioned a rim you could buy that could help. As my own outlook is long term, I would buy that rim, and go ahead and try the solutions suggested above with your current rim. I would both learn more now, have a solution on it's way in case what I learn is "nope, that didn't work", and also have a strong rim in the closet for the future if it does work.

After the supply disruptions from COVID, and also due to economics, I think in terms of the next 20 years (and after that I probably won't be riding, but who knows?)

 

[Papa]

EwanC,
I noticed a peculiar pattern in one of your images - appears all of the failed spokes were leading spokes (as opposed to trailing spokes). Assuming the drive side spokes were tensioned higher than the LH spokes, is it possible that repeated, and/or abrupt engagement of regen exacerbated the spoke failures?

 

[EwanC]

Doing that can only help, but 60:40 tension ratio is a lot better than most rear wheels. Tension imbalance of that degree is unlikely to be a root cause of your spoke breakages. I was thinking idly about your wheel yesterday at work, and it occurred to me that if none of the elbows-out spokes have broken, it's possible you could cure the problem by lacing them all elbows-out.

OK, well if we laced it that way (elbows out on both sides) the spoke calc says that'd lead to 45:55 tension ratio. That's still an improvement over 40:60 and it'd mean all of the spokes are snug against the flange. Seems like we'd be addressing two possible causes that way instead of just one possible cause, good idea.

 

[EwanC]

Nipples should be collinear to the spoke angle. (lying in the same straight line)
The normal fix to this problem is to file the nipple hole in the rim to the correct angle.
Because the DT Swiss U 663 rim has eyelets with the wrong angle the only solution, and this will sound brutal, is to use a hole alignment punch to bend the metal to the correct angle.

Yeah the ~75° spoke angle isn't ideal and is causing spokes to be a bit bent. I'm keen to do something about it.
Would bending around the holes like that leave the eyelets intact? They look pretty thin, maybe they'd pop out if I bent the aluminium they're sitting on? I'd be reluctant to damage/remove those eyelets given manufacturers use them to allow the aluminium to be a bit thinner where the nipples tension against the rim.
Do you have any experience with the Sapim Polyax nipples mentioned earlier in the thread? They sound like a nice solution here - allow the nipple to come out of the rim at a greater range of angles, without needing to be so brutal to the rim.

 

[EwanC]

Although my suggestion will sound odd at first, it's one I would try.

You mentioned a rim you could buy that could help. As my own outlook is long term, I would buy that rim, and go ahead and try the solutions suggested above with your current rim. I would both learn more now, have a solution on it's way in case what I learn is "nope, that didn't work", and also have a strong rim in the closet for the future if it does work.

After the supply disruptions from COVID, and also due to economics, I think in terms of the next 20 years (and after that I probably won't be riding, but who knows?)

Yeah given we live at the bottom of the world and sourcing weird bike parts isn't easy, maybe that's a good idea! The Ryde Andra 40 isn't particularly costly either. Maybe I should also keep enough spare spokes on hand to do a rebuild, given I have to get those from overseas too (single butted spokes are weird apparently).

 

[EwanC]

EwanC,
I noticed a peculiar pattern in one of your images - appears all of the failed spokes were leading spokes (as opposed to trailing spokes). Assuming the drive side spokes were tensioned higher than the LH spokes, is it possible that repeated, and/or abrupt engagement of regen exacerbated the spoke failures?

Your observation is correct. That sounds like a plausible explanation for sure, but it'd be pretty hard to prove/disprove.
I'm running CA3-13v2S firmware on my CA, which only supports engaging regen via pulling the brake lever and then using the throttle to control regen strength. That results in a pretty smooth application of regen (I don't just pin the throttle), as opposed to back-pedal regen which makes it easy to just "slam" the regen up to max (I see my wife doing this - she has the latest CA3 firmware with back-pedal regen).
Conversely, I only have a cadence sensor for PAS which can fire up the forward motor torque pretty quickly if I "ghost pedal" in a too-low gear (it sees a very high pedal RPM and gives a lot of power in response). I'd have thought this would make me more likely to break trailing spokes - but I didn't break any trailing spokes at all.

Still, this seems like something I could address easily enough - use the CA3 ThrO->Down Rate setting to reduce the max allowed throttle slew rate downwards. That'd reduce how quickly regen could be applied since regen is just applied by throttle output falling below ~0.9V after all. Seems worth a try.

 

[PaPaSteve]

Yeah the ~75° spoke angle isn't ideal and is causing spokes to be a bit bent. I'm keen to do something about it.
Would bending around the holes like that leave the eyelets intact? They look pretty thin, maybe they'd pop out if I bent the aluminium they're sitting on? I'd be reluctant to damage/remove those eyelets given manufacturers use them to allow the aluminium to be a bit thinner where the nipples tension against the rim.
Do you have any experience with the Sapim Polyax nipples mentioned earlier in the thread? They sound like a nice solution here - allow the nipple to come out of the rim at a greater range of angles, without needing to be so brutal to the rim.

Sapim Polyax nipples do have a nice radius on the underside of the head, I don't think it helps here.
It's been my observation the problem is caused by the eyelet diameter being too small for the nipple diameter.
The nipple angle is constrained by the diameter interference.
In your case there are 2 things that work against the ideal nipple angle :
During rim manufacture the nipple angle for a 20" rim is drilled at the same angle as a large diameter rim, your build needs more angle.
Larger diameter of the e-bike hub forces an extreme spoke angle.

If bending the rim to achieve a better angle seems too extreme a fix, another option could be file the hole to the correct angle and add washers under the nipple head.

 

[offGridDownUnder]

we live at the bottom of the world

Nah, mate. We live at the top of the world. They're just all wrong about that, too.

 

[Papa]

papasteve,
While I agree that nipple/spoke angular alignment is important, it is not, IMO, the core cause of Ewanc's multiple spoke failure.

1.) While a number of factors undoubtedly exacerbated the multiple spoke failures, I am reasonably convinced the key culprate was non-uniform spoke tensioning. Over half (of 9 total) leading spokes failed on the higher tensioned drive-side. Why the failures on drive-side only?... because the drive spokes were tensioned closer to the spoke's ultimate failure limit. I'm assuming the 5 failed spokes were tensioned slightly higher than the remaining 4 that did not fail.

2.) Unusually high torque capabilities of the GMAC (over 105N·m @ lower speeds).

3.) Increased GVW forces greater torque demands on the GMAC (and, of course, increased tension on the leading spokes during regen)). As GVW increases, application of regen must also increase. And it's easily deceptive, because it doesn't 'feel' any stronger than when unloaded... even though the torque applied at the wheel can easily double on decents. On an unloaded conventional bicycle, contact patch grip becomes the limitation.

Spoke tensioning on wheels expected to survive unusual conditions, demands uniformity and precision. This is especially true on wheels that are built on offset hubs. Tension Ratio of 50/50 is strongly preferred, if at all possible, because it guaranties uniform loading on ALL spokes.

 

[PaPaSteve]

papasteve,
While I agree that nipple/spoke angular alignment is important, it is not, IMO, the core cause of Ewanc's multiple spoke failure.

1.) While a number of factors undoubtedly exacerbated the multiple spoke failures, I am reasonably convinced the key culprate was non-uniform spoke tensioning. Over half (of 9 total) leading spokes failed on the higher tensioned drive-side. Why the failures on drive-side only?... because the drive spokes were tensioned closer to the spoke's ultimate failure limit. I'm assuming the 5 failed spokes were tensioned slightly higher than the remaining 4 that did not fail.

2.) Unusually high torque capabilities of the GMAC (over 105N·m @ lower speeds).

3.) Increased GVW forces greater torque demands on the GMAC (and, of course, increased tension on the leading spokes during regen)). As GVW increases, application of regen must also increase. And it's easily deceptive, because it doesn't 'feel' any stronger than when unloaded... even though the torque applied at the wheel can easily double on decents. On an unloaded conventional bicycle, contact patch grip becomes the limitation.

Spoke tensioning on wheels expected to survive unusual conditions, demands uniformity and precision. This is especially true on wheels that are built on offset hubs. Tension Ratio of 50/50 is strongly preferred, if at all possible, because it guaranties uniform loading on ALL spokes.

The shortest distance between two points is a straight line.
EwanC pictures show an obvious curvature in the spokes demonstrating "not a straight line".
So physics states, due to the curvature, the spokes will not be at full tension when in use.
It should be expected that this condition will create some cyclic loading and unloading to the spokes.
Cyclic loading, over time, always finds the weakest link.

 

[amberwolf]

If it's helpful, here's the full pics of my gmac laced up by grin. Can't say how well it would hold up under my usual heavy-cargo usage, as I didn't keep it on the sb cruiser that long due to controller issues and noise from it being in a small wheel.

 

[EwanC]

Thanks for your thoughts everyone.

Sapim Polyax nipples do have a nice radius on the underside of the head, I don't think it helps here.
It's been my observation the problem is caused by the eyelet diameter being too small for the nipple diameter.

After buying some Polyax nipples and trying them in my U663 rim: yes, that's absolutely the case here. The hole diameter (really eyelet ID) combined with rim thickness limits the possible angles the nipple can exit the rim at. Even with the rounded Polyax heads, it's the diameter of the nipple "body" limiting rotation. So the Polyax nipples aren't much help in my case.

I also emailed Grin about my spoke problems, and Robert their wheel builder responded. Reposting here with his permission, because I'm guessing some of you will be interested in what he had to say!

I'm just going to go bullet here:
1. What is the actual e.r.d.? did you measure it or rely on the mfg page which says 383mm and gets you those numbers for the spoke length.
Always measure the actual rim. See my videos
2. That said, they're still going to be somewhere around that length so the problem of "manhandling" is still there. With all due sympathy, spokes have to be manhandled sometimes.

I had relied on the manufacturer's ERD spec. When I measured it as per Robert's video, it was more like 380-381mm. So that explains the spokes being slightly too long.

3. Note the articulation angle on the calculator. It's 74.9 degrees when using the 383mm e.r.d. This is 0.1degreee less than my minimum (not much, to be fair). A lower articulation angle (or conversely greater than 15 degrees) is where we see rim damage with normal tensions, and really awkward builds. On smaller rims plus large motors, I will sometimes switch to radial lacing. The problems with radial on this sized wheel are almost non-existent. Many of our hubs, the Grin AllAxle included, use paired holes on the motor flange and are "Zero Cross" laced, NOT radial, and solve these building problems inherently.

However, all of that aside:

4.The right flange of the GMAC leaves no significant space between it and the inside gear (largest cog, lowest gear) on the cassette. This is a liability if the chain unships to the left (derailleur adjustment problem, frame alignment problem, bad shift) catches the exposed spoke elbows and jams, ripping everything apart. Not uncommon and very expensive.
BECAUSE of that, all elbows are "in" on the right.
And that means all the elbows are off the flange and unsupported.
I build all of these with a tiny brass washer under the head which pulls the elbow in closer to flange, supporting it much better.

^^ This explains why Grin builds GMAC wheels with the elbows in on the drive side, even though that makes the tension ratio a bit worse. I copied Grin's practice of having the elbows in on the drive side, but didn't use the washers - so that helps explains spokes breaking.

5. So, rebuild the wheel with the washers (easy to find on online wheelbuilding sites) and I would do it with the pulling spokes on the outside of the cross (opposite what you have now). It should be rebuilt at this point. Check the rim for damage.

6. I can't check the tension from here but; I suspect that it's low. That rim should have 120 - 130KgF pulling on each spoke, based on its weight and construction. That spoke is too short to fit most (maybe not all) tensiometers.. How did your mechanic check it? The bend towards the nipple may be unavoidable, but bends in the middle usually pull out at that tension. If I'm wrong, I apologize in advance.

My mechanic answered that yes, his tensiometer is too large to fit on such small spokes and he tensioned it by feel. So insufficient tension can't be excluded. Though he is an experienced wheel builder, so his tension "feel" judgement probably isn't too far off. I asked Robert how he checks tension on such small spokes:

In answer to your mechanic's question I use a Pillar tensiometer that has a 50mm pin spacing. I also use a fairly sophisticated tone system and formulas to measure in hertz. . The tonal accuracy is probably within 1% or better. The accuracy of the tension derived from the tonal measure is more like 10%. Not perfect but it keeps me from blowing up wheels that I can't measure with the gauge. Bear in mind that Park Tools' official position is 25% difference between the tightest and loosest spoke. That reflects reality for many rims. When possible I can check against the Wheel Fanatyk gauge.

My number 5 and 6 would be my highest probability reasons for your spoke breakage (at the head). The uneven tension and the articulation angle are the less likely reasons for this problem. (please remember I haven't seen it in the flesh). Making the sides matching (or at least closer together) tension is fine, but I don't really start worrying about a wheel this size until it's worse than 75/25. I might loctite the looser side. That changes as the rim gets larger, but even then I don't worry that much.

We've now gone ahead and rebuilt it in the U663 rim (radial this time - I already had the right spokes on hand for that). If that fails, I'll try again with a single cross build in a Ryde Andra 40 rim (which I now have on hand).