Wheel building for dummies

[Brucey]

wheels are recommended to be built (by shimano amongst others) such that the spokes that are doing most 'pulling' (whether by virtue of the action of the pedals or the brakes) are to be outside spokes,  not inside spokes. 

The idea is that in averagely-built wheels the outside spokes are more able to resist the increased fatigue loadings, and most likely the wheel is slightly more laterally stable if the outside spokes are loaded transiently than the inside spokes are too.

Thus the recommendation is that front disc wheels are built symmetric, outside spokes leading, and rear wheels are built skew, i.e. such that you have outside spokes leading on the LH flange and trailing on the RH flange.

If the wheel is built properly (i.e. with the spoke fit into the hub adjusted using washers where necessary, and with proper stress relief as per Jobst Brandt) then it doesn't make much difference; the spokes won't be breaking anyway.

In fairness there is are reasons for preferring the DS spokes to be inside trailing; under high torque the spoke crossings are not pulled towards the rear mech (as mentioned on Sheldon Brown), and if the chain unships it tends to jam less completely in the spokes too. Nonetheless I usually build rear wheels with the DS spokes outside trailing(pulling); if the rear mech starts to touch the spokes under load you should hear it before it tangles and know that there is an adjustment to be made.

Note also that the notion of a symmetric built disc-braked front wheel being maybe better is based on the premise that there is significant torque carried by the hub barrel. This often isn't the case, at least not enough to make much difference.

IIRC Zinn has written an article that is on velonews that discusses this rationale of spoking in disc braked wheels.

FWIW I usually build non-disc brake wheels symmetric so that the rear has outside trailing spokes both sides and with the front wheel it is also built symmetric but the direction doesn't matter. Note also that symmetric-built wheels are slightly slower to lace but show more uniform spoke protrusion/tension than skew-built wheels.

cheers

[RogerMusson]

Roger Musson also goes through the formula in one of the appendices to his book, based on the cosine rule, so I'm surprised that he ends up with different lengths, must be an assumption in the calculator somewhere.

The online calculator is new, and will be included in the next update of the book. However, the help guide within the online calculator tells you exactly what's going on. Here's what is written:


How the spoke lengths are calculated
The standard spoke length formula is used to calculate the left and right spoke lengths. This is pure mathematical geometry using the data you supply, and it calculates theoretical spoke lengths. These lengths need adjusting to account for spoke stretch.

1. Determine the tension in the left and right spokes. The formula for tension ratio between the left and right uses the hub flange offsets and theoretical spoke lengths. A tension of 120Kg is used for the side that represents 100%, and the other side adjusted down according to the ratio.

2. The spoke diameter is used to calculate the cross sectional area.

3. The elongation can now be calculated using standard engineering stress-strain equations. I've used a stainless steel Young's Modulus of 210000 N/mm2. This calculator is for stainless steel spokes, other types of spoke material will have a different Young's Modulus which will affect the elongation.

3. The lengths and elongations can be seen by looking at the log (once you calculate a spoke table).

[RogerMusson]

AIUI in his latest version he now allows for the elasticity of the spokes, and to keep things simple for us numpties he now gives a "minimum" length (from which you never round, and it is perfectly acceptable to round up by 2 mm) rather than the theoretical 'perfect' length from which you may round down by only a little ** and round up by a little bit more⁺⁺.

** rounding down by less than 1mm is usually OK, any more than that and lacing might get tricky
⁺⁺ rounding up by more than 2mm from the 'perfect' length and you run the risk of bottoming out the thread.

Yes, that's exactly how I intended it to be, couldn't have worded it any better myself, although I would take out the reference to numpties, since it's the logical way of presenting lengths.

Cheers,
Roger

[RogerMusson]

The spoke gauge is allowed for. This could/should make the spoke calculations more accurate, but in many cases this (combined with a slightly different ERD measuring and rounding regime from most others)

The method of measuring the erd which I describe is the standard way. Can you point me to the other calculators that do it differently.

just confuses people because they get different answers from when they use other spoke calculators.

Virtually all other calculators use the basic theoretical formula, although there's one out there that subtracts 0.2mm to match the old Wheelpro calculator. The only other calculator that does something different is the DT calculator.

  The corrections for spoke stretch are smallish, being typically about 0.5mm.

For a 294mm spoke, the approximate stretch due to 120kg tension is...
0.5mm for 2mm plain gauge,
0.6mm for 1.8mm butted,
0.9mm for DT Revolution and CX-Ray
1.0mm for Sapim Super spokes


Cheers,
Roger

[zigzag]

good to have the man himself here!

thank you for your contribution and a great tutorial - i've built and rebuilt quite a few wheelsets since, no issues whatsoever

[Jakob W]

 YACF, home to wheelbuilding gurus!

[Brucey]

In a nutshell; if Roger's method works out as planned and predicts an exact length, that should come to the nipple slot bottom.

But.... if there is an error of any kind (eg in measurement or arising because of variations in the fit of the spoke in the hub) the predicted spoke length could easily be too short, which is usually (IME) a worse result than too long.

cheers

[zigzag]

apart from a spoke's elongation, i reckon there's another ~0.5mm to account for in a hub shell hole when the tensioned spoke makes indentation when it "settles", also the elbow of the spoke gets straightened a bit

[Brucey]

apart from a spoke's elongation, i reckon there's another ~0.5mm to account for in a hub shell hole when the tensioned spoke makes indentation when it "settles", also the elbow of the spoke gets straightened a bit

it depends what the fit is like. I recently did some tests that indicated that the fit (within reasonable bounds, not as large as those I have seen wheels built with) could easily make a difference of +/- 1mm or so.

If you take steps to ensure a really good fit (in which the spoke elbow is supported as well as it can be by the hub), then you could easily need spokes that are at least 0.5mm longer than with an average fit.

cheers

[ElyDave]

Roger,

thanks very much for the comments, very useful.  I've got two wheels built so far based on the book and the advice on here.

[whosatthewheel]

In a nutshell; if Roger's method works out as planned and predicts an exact length, that should come to the nipple slot bottom.

But.... if there is an error of any kind (eg in measurement or arising because of variations in the fit of the spoke in the hub) the predicted spoke length could easily be too short, which is usually (IME) a worse result than too long.

cheers

It depends... too long can jam in the nipple thread, which means one has to think in terms of thick rim washers. Only spokes that don't jam due to the butting being just at the end of the threaded section are Sapim D-Light... not cheap though, but I routinely use them on all my wheels, as they work well for my weight.

Too short can be corrected using DT Swiss longer nipples 14 or even 16, but I agree it is a cosmetic solution more than anything, as effectively a short spoke will place more load on the nipple, regardless of how well it engages with the threads

I have always been for the "round up" approach, rather than the "round down", unless I use very thin spokes which stretch a bit further... Laser, Revolution and such

[Brucey]

FWIW I recently carried out some tests in which I deliberately wound an unlubricated  brass  nipple on about 2mm past the point at which the threads ran out on a 14G stainless steel spoke.  There is a lot more detail (which I have described elsewhere) but in a nutshell

1) the spoke suffered no detectable damage (as inspected under a microscope)
2) the nipple just lost the tops of a few threads (it subsequently span on to another spoke just fine)
3) the loss of spoke thread engagement was a relatively small proportion of the whole; with good quality parts there is little risk of thread stripping until the nipple is wound on a fairly ridiculous amount
4) the torque required (even without any lubrication present) was much smaller than would be required to shear or permanently twist most spokes.
5) once fitted thusly, the threaded portion of the spoke no longer sees such large bending stresses inside the nipple, so the chances of spoke fatigue in the nipple are reduced.

So the consequences of 'crunching' are usually not great, in practice. With this in mind, I'd suggest that (with some calculators) shooting for spokes that finish flush with the top of the nipple may be a better route; it helps avoid the possibility of a small measurement error giving a predicted spoke length that is too short.

cheers

[whosatthewheel]

With this in mind, I'd suggest that (with some calculators) shooting for spokes that finish flush with the top of the nipple may be a better route; it helps avoid the possibility of a small measurement error giving a predicted spoke length that is too short.

cheers

Some rim manufacturers already factor that in, when they provide the ERD, problem is one never knows which ERD they are quoting... the one where the nipple is counted as 10 mm or the one where it is counted as the all length 12 mm. It would be best to call them differently (e.g. ERD10 and ERD12) to avoid confusion.

I've never tried jamming deliberately the spoke, always assumed it would result in twisting of the spoke and damage to the nipple... interesting to see it is doable with no major consequences

[Brucey]

IME (and Roger makes the same recommendation BTW) you should always measure the rim rather than rely exclusively on manufacturer's figures.

 Many manufacturers quote NSD (nipple seat diameter) as if it were ERD, and this obviously yields predicted spoke lengths that are 1.5 to 2mm too short. In addition there is the possibility that there are batch-to-batch variations in rims, too.

cheers

[whosatthewheel]

True,

however, the batch to batch difference is a bit of a myth... I think over the years 2012-2015 I must have built at least 300 H Plus Son Archetype rims and they all had exactly the same ERD.

There might be rims that stayed in production for decades and went through several iterations with small variations of ERD, as a result of changed extrusion dyes

[Brucey]

Not a myth!  I have had rims of the same make and model that have varied appreciably in diameter (and tyre fit). Not all companies exert the same kind of QA on the rims they make.

BTW the extrusion profile (and the ERD) of the Archetype has changed during production.

cheers

[whosatthewheel]

BTW the extrusion profile (and the ERD) of the Archetype has changed during production.

cheers

I don't think I've ever handled any of the first (allegedly lighter) generation Archetype. All the ones I have built were 470-480 grams and 592-593 ERD measured adding 10 mm per nipple.

[Brucey]

The archetypes that I have built personally (nothing like as many) and seen built (many more) have worked OK using an ERD (checked by measuring) of 595mm. With good quality DB spokes (sapim or DT) the spoke ends typically finish mid slot to 0.5mm protruding.

 IIRC the change in extrusion profile (which I think thickened the spoke bed on the inside, thus increasing the ERD slightly) appeared in rims starting sometime in 2013. I've measured between 460 and 490g for the current rims, some of which variation appears to arise from the way in which the rims are finished.

cheers