Raleigh Headset 26 tpi 1" steerer _ Help!

[Brucey]

I would recommend extreme caution when welding steerer tubes together; this is the one tube you can't easily inspect on a bicycle, yet it is also the one that is most likely to try and kill you if it fails.

Things like plug welds can actually be counterproductive, since they are chock full of residual stresses and can be ideal places for cracks to start.

Most framebuilders wouldn't touch that job, leave alone using the method you propose.

Might I suggest that

a) you would certainly be better off if you just fitted a 24tpi headset to you existing steerer
b) ditto if you just replace the fork
c) if you want to go 1" A-head (almost as much of a blind alley as 26tpi IMHO) then the best (and easiest/safest) route is simply to bond (using epoxy resin) a 7/8" piece at least 3" long into both the steerer and the extension. If you do this right the joint should have a shear strength of about 10 tonnes, and ought to withstand the bending loads OK provided the join is near the top of the steerer.
d) whichever route you use for an A-head conversion (or other extension) it would be very prudent to use a long bolt of some kind that goes all the way to the fork crown. This way the A-head preload isn't trying to tear the steerer apart and should the extension fail, you may retain some semblance of steering control long enough to bring the machine to a halt.

cheers

[joshua3]

Thanks Brucey,

I can't see what I am proposing would be any worse than this?
I am going into this with my eyes open, I will definitely use your suggestion of the
star nut beneath the insert tube also. I understand your concerns about stress concentrations
and crack propagation, but having welded up a locost 7 and working as a stress engineer
for BAe Systems, I feel qualified to do this. I cannot bring myself to trust epoxy for this
job, unless I was to perform a load test on it first.

Thank you for your thoughts

James

[kylecycler]

... I actually got one of the ETC 26 tpi headsets, but curiously it still didn't screw on to the threads, even though it screwed quite happily all the way on to two 24 tpi non-Raleigh forks I have. Don't understand that.

that is actually a pretty good sign that the ETC headsets may be wrongly listed. I don't know that for sure but that'd be my best guess. You ought to be able to check the thread pitch, using a gauge or another part with the same pitch.

If you can get hold of a 26tpi axle (eg a 3/8" x 26tpi axle which is a common size for rear nutted axles and some fronts, or a SA three-speed axle which also has a 26tpi thread) then you will be able to sit the axle in the headset threads. If the threads are the same it will mesh perfectly and sit square. If the threads are a different pitch, it won't quite; you will be able to rock the parts slightly and the parts won't want to settle in any one place even if you are squeezing them together with finger and thumb.

cheers

The ETC headset's box says 26 tpi; a 24 tpi headset was also listed but it certainly doesn't seem right. I think I've got a 3/8" x 26tpi nutted rear axle from a 1980s Raleigh Pioneer so I'll dig it out and try what you suggest. Thanks.

I've heard of 26 tpi fork threads being re-cut to 24 tpi, but I presume you wouldn't advise that? I asked at the local bike shops but none had a threaded headset cutting tool, which surprised me. They were neither interested nor keen to help, but if it compromised safety I can understand that.

I think for now though I'll put it back on the back burner - I've got an awful lot to sort and it's a long way down the list. Frame and fork need refinished anyway, and I want to do that right. I've waited this long, another few years won't make any difference - it's always something to look forward to.

[Brucey]

Thanks Brucey,

I can't see what I am proposing would be any worse than this?
I am going into this with my eyes open, I will definitely use your suggestion of the
star nut beneath the insert tube also. I understand your concerns about stress concentrations
and crack propagation, but having welded up a locost 7 and working as a stress engineer
for BAe Systems, I feel qualified to do this. I cannot bring myself to trust epoxy for this
job, unless I was to perform a load test on it first.

A good 'off the shelf' long bolt solution is to use a 'Headlock'. You may not be able to get a SFW into the base of your extant steerer because the ID tapers inwards near the base. If you are fitting a long bolt it may as well go down all the way if you can do it; a welded fitting with a female M6 thread (eg a connector nut) and a short length of tube at 90 degrees can be used with a crown bolt or brake bolt at the bottom of the steerer.

FWIW Locost 7s are designed so that few, if any of the welds are put into fatigue in bending; most of the structure is highly triangulated and the welded joints (if welded all round) are overspecified in such a way as they are extremely unlikely to cause trouble unless they are truly dreadful welds. Even then, a single weld failure is unlikely to cause an accident, and most of the critical welds are on view so can be inspected easily.

A steerer extension is a completely different kettle of fish. If you want to do a proper job, a butt weld with dressed beads both sides, inspected both sides for lack of fusion, cracks etc would be a pretty good arrangement; without stress concentrations such a weld ought to be quite fatigue resistant. Even with that I'd use a long bolt BTW.

I've tested typical adhesively bonded joints with 'perfect' shear loading and such joints (using half-reasonable epoxy resin, and half-decent prep/assembly) take about one tonne per square inch to fail when they are new. (You can easily test this for yourself by epoxying nuts and bolts together, and looking at the breakaway torque you get). With a 3" overlap on both sides of the joint, and a long bolt in use, the chances of losing steerage with a bonded sleeve joint would be about zero IMHO.

cheers

[joshua3]

True Brucey, but the lower wishbones see significant bending loads and I have never seen any of these fail.
Regarding a fatigue failure, if we were considering alloy tubes I would be in total agreement, but steel alloys
which are subjected to a cyclic stress levels below the elastic limit rarely fail due to fatigue even with stress raisers.

Regards

James

[Brucey]

... steel alloys
which are subjected to a cyclic stress levels below the elastic limit rarely fail due to fatigue even with stress raisers.

If only!

In another thread I recently reported that in just a few thousand miles you could get many millions of stress cycles in a steerer. SCFs for undressed weld beads, lap joints etc are typically listed as ~x10 to x12 and the service stresses are superimposed on a residual stress distribution left over from welding. The residual stresses are basically already yield magnitude.

Most steels have a fatigue limit (ie essentially infinite fatigue life) if the cyclic stresses don't (anywhere, not even locally) exceed about 1/3 of yield (in a normally stress-neutral specimen). You will be very far away from that with the design and fabrication method you propose; even without allowing for the effect of residual stresses, because of the stress concentrations, you would have to keep the service stresses below about 1/30 of nominal yield if you wanted a decent fatigue life.

cheers

[joshua3]

Where are these stresses occurring in the steerer? Are you modelling it as a simply supported beam cantilevered at the end where the wheel axle is? I am joining the steerers at the furthest support (from the load) so I fail to see where the bending loads are coming from.

Regards
james

[Brucey]

the bending loads occur because of the road and they still occur in the upper parts of the steerer whenever there is any weight bearing on/attached to the handlebars. In addition, riding out of the saddle (or an enthusiastic 'handlebar wrestling' style even in the saddle) imposes very large stresses in a steerer too.

cheers

[joshua3]

The steerer will see a shear reaction with my proposal, but very little bending.


Regards

James

[joshua3]

Yes I can see handlebar loads acting on the steerer but these bending stresses (at the upper bearing) are tiny compared to those caused by the road acting on the steerer at the lower headset bearing.

[Brucey]

you appear to be ignoring both the effect of residual stresses and that of stress concentrations.

You have both.

cheers

[joshua3]

I don't think it is possible to quantify the residual stress from the welding process. My analysis is very pessimistic as it assumes a built in cantilever, the steerer isn't.

[joshua3]

corrected error

[Brucey]

residual stresses after welding are always yield magnitude; 'tis the nature of the beast.

You will also have high SCFs.

I don't think you are being anywhere near pessimistic enough TBH.

cheers

[joshua3]

oh well, thank Brucey, we shall see. What epoxy have you tested and found to provide the values quoted in your earlier message?
Thanks
James