Originally Posted by
laura*
Try these numbers: ERD 598, 32 spokes, 3x. Left flange outside 35.9, left inside 32.45, right inside 18.85, right outside 22.3. (That's a 3.45mm flange thickness.) Keep in mind that the spokes lie _next_ to the flange which increases the angle and length difference, so add/subtract an extra 1.0mm (for 2.0mm spokes.) Thus calculate with 36.9, 31.45, 17.85, 23.3. I calculate a left side length delta of 0.65mm and right side of 0.40mm.
Large wheels, narrow hubs, thin flanges, and thin spokes (ie road wheels) will reduce the delta. The converse (ie mountain 26" wheels) will increase the delta.
Looking at its formulas, it simply calculates at the offset(s) entered. It does nothing to consider flange thickness.
That is because flange thickness really doesn't play any significant role, even with a whopper thick flange like yours. I think the problem is that you are trying to track the physical length of the spoke. Now this may sound like the obvious thing to do, but the snag is that spoke length is a little more complicated than that:
The length of a eg. DT Swiss spoke is measured from the top of the thread (obviously) to the inside of the the spoke head, not the spoke elbow. When measuring a spoke length with a ruler, the ruler and spoke therefore forms a triangle. Measuring the spoke length is akin to measuring "virtual top tube length" on sloping frames.
Here is a primitive diagram on from where one measures spoke length: from the X, where the spoke will support itself on the hub flange when the spoke head is pressed against the flange:
Code:

 
 X 




http://en.wikipedia.org/wiki/Right_triangle
Imagine that the A to B is the length of the spoke from elbow to thread, and A to C is the length of spoke from the elbow to spoke head. Then C to B will be the "virtual spoke length" which is the spoke length as advertised by DT Swiss on the outside of their spoke boxes.
The length of the spoke as advertised on the box is the "virtual length", or the hypotenuse on a right angled triangle. So a 250 mm spoke will always terminate 250 mm above* where it is supported on the hub flange, regardless of elbow length and flange thickness. Therefore stuff like how close the spoke lies to the flange etc, really doesn't matter regarding spoke length.
*At a right angle. Of course, one side of the flange is slightly farther away from the rim than the other, and will therefore "travel" slightly longer to reach the ERD. But the difference between heads in/out spokes is at max the difference that flange thickness causes, and since Spocalc measures from the middle of the flange, then effectively this deviation is cut in half. Eg. with a 4 mm thick flange the difference will be like adding/subtracting 2 mm to WL and WR in the formula. And since spoke length is actually measured from a point inside the flange, this will be a max deviation.
Playing around and adding or subtracting 2 mm from WL and WR (a 4 mm flange thickness) on a 26" (540 ERD) high flange (74 mm) dynamo hub 3X, 36H front wheel only gives me differences around 0.2 mm, rarely 0.3 mm, and that is a pretty extreme case.
Originally Posted by
laura*
I looked at one of the first wheels I built. Even amongst spokes that should be identical, the spoke ends are all over the place. That wheel was true with tension identical to within one unit on the Park gauge. Thus I think it is possible to true a wheel that measures fine but doesn't have the nipples turned to the "correct" spot. A small calculated length difference might be lost in the "noise".
We can certainly agree that the noise level is pretty high when using crude instruments like the TMS1, but still, it is not my experience that spoke ends are all over the place when relative spoke tension is close. My own measurements of 4 adjacent spokes plus the well built wheels I have seen seems to support that too.
One really ought to measure or even see a 0.50.8 mm length difference between heads in and heads out spokes if such a discrepancy existed. There can be several explanations for your observed data like spoke wind up, but Retro Grouch's explanation seems to be the best since there seems to be some kind of system in how your spokes deviate in apparent length.
Originally Posted by
laura*
Hmmm. I should bring a precision measuring caliper and measure the local bike coop's ERD tool to see where on a standard nipple it places the ERD.
That really wouldn't help, since all you would do is to measure that bike tools definition of what ERD is. The point is that ERD is a practical guideline that depends on the implementation of spokes and nipples. This works out in practice since most spokes and nipples are "standard" ones as defined by tradition. But take eg. DT Swiss's 14 mm nipples. With those nipples the wanted termination point for the spoke is 1 mm below the nipple slot, since ERD is the termination point, that will mean that the ERD for a given rim depends on the nipple implementation. So ERD isn't a fixed number for a given rim. In practice however, ERD is a fixed number assigned to a given rim, like when Rigida prints the ERD on the rim label, and people using 14 mm DT Swiss nipples just subtract 1 mm from the calculated spoke length.
Originally Posted by
laura*
Using the formula from Jobst Brandt's book, for the above wheel, the NDS spokes stretch 0.27mm (@ 60 kgforce) and the DS spokes stretch 0.44mm (@ 100 kgforce). If one started with spokes with a 2mm length difference, the difference will be only 1.83mm when tensioned. (Example: 292mm NDS spokes stretching to 292.27, and 290 DS spokes to 290.44.)
If a 2mm tensioned difference is needed, then the wheel needs to be built with spokes that differ by 2.17mm. Of course, such sizes are not offered for sale. This may necessitate starting with a 3mm difference.
Every DT Swiss spoke (admittedly not many lengths) I've ever measured has been a fraction of a millimeter short of the "marked" length. Using a precision caliper, I measured the DT Alpine III spokes I'm currently building with. Both lengths I bought are 0.3mm short. At 70 kgforce, according to the formula, these spokes elongate ... 0.3mm.
Due to an unfortunate mislabeling accident I was forced to measure many dozens of spokes in many different lengths. So I can say with some confidence that the many spokes I measured from both Sapim and DT Swiss came remarkably close to their advertised length as a general rule, and certainly not 0.3 mm to short. I suspect the discrepancy you measure are due to your measuring methodology. Here is IMO how to measure spoke lengths the correct way: take a metric quality steel ruler. Place the spoke head at the bottom of the ruler. With one finger press the spoke head "flare" up against the ruler, so that the spoke is lifted above the ruler and therefore forms a triangle together with the ruler when seen from the side. Measure the spoke length.

Regards