Originally Posted by
Road Fan
I want a few millimeters of accuracy, but I also want repeatability.
We seem to be in agreement on a few high-level points.
You say there are "super-accurate" methods that are well-tested, in the thread you cited. But a skim (admittedly a sketchy reading) through it does not show me any testing, and a few of the suggested techniques do not say it's tested. Have you tested any? What reference did you use to check if the results are accurate?
And why do you jump to the conclusion or at least suggestion that what I suggest is not tested? How do you know?
I can confirm the data given in some of the classic bike catalogs, Trek and Rivendell. I'm happy with that accuracy.
I don't see where I said you didn't test your suggestion.
What I said was that your suggestion is unlikely to be accurate because it relies on small degrees of angle measurement accuracy with a level. Additionally, it relies on getting fine angles measured repeatedly, and every time you have to re-measure a small angle (like with a level), you add a new possibility of error - which is like tolerance stacking. It isn't that you can't get the right answer from your method, it is just that there are so many opportunities to get the wrong answer and no controls to prevent that from happening - you are essentially eyeballing your test. If you had a surface that is useful for putting a level against it, why not simply measure directly from that surface without adding the step of leveling? Would the steerer and your measuring surface not remain parallel in any axis once you locked them together?
If you look in the other thread I tested FBinNY's string method and my pot and glass table method. My results for my method was within a quarter mm, so I think it could be regarded as solid. It also doesn't suffer from some of the problems I mentioned that cause tolerance stack.