TC1

Yes. To less than .01mm accuracy? Maybe. On a frame old enough to have been originally built at 126mm? I doubt it.

How so, exactly? Seriously -- it is possible that you know a trick for this that I don't. I don't claim to know everything. I am curious how one would perform this comparison to hundredth-of-a-millimeter accuracy.

At risk of being repetitive, how does one line these up to hundredth-of-a-millimeter precision?

I'm not saying that any of this is impossible -- but it is not as easy as is being suggested, to achieve the kind of precision that we are discussing.

Kontact 

I don't know what fractions of a millimeter pertain to. But the alignment of the dropout isn't a distance, it is angle, and that angle serves the hub and to a lesser extent the alignment of the wheel. A very precise angle is possible with these tools, The rest of the alignment is more than close enough with basic tools, but an alignment table can be extremely accurate.

The reason the dropouts need to be so precise (without the rest of the bike being that way) is to keep the hub retained and undamaged.

TC1

To further a point that I mentioned previously, I just grabbed two brand-new, never-used, never-welded, never-even-painted cast steel dropouts. I measured them with a decent digital caliper. The mounting surface thickness varies by almost a tenth of a millimeter at various locations, on each. The length of the dropout from weld-surface to axle-hole varies by a tenth, as well ( with reference to the previous point about the dropouts being precisely-located in space, with respect to the frame ).

The axle slots vary in width by a tenth of a millimeter, from one dropout to the other.

I just don't think people are being realistic about the limitations of the material in-question, and I seriously doubt that the precision being discussed is possible with simple hand tools, and if it is, achieving that is definitely not a rudimentary task.

TC1

We previously established that the dropout angle change from a 126->130 set is just a bit over a quarter of a degree, with common chainstay lengths. If we are attempting to square two planes separated by 130 mm, more precisely than 0.25 degrees, we must be able to measure to hundredths of a millimeter. Specially, if the leading edges of the dropout mounting surfaces are 65mm distant from our centerline, then a quarter-degree slant makes the rear distance only 65.07mm. And it is that 0.07mm that we'd be trying to adjust out, and that's no mean feat.

The term "very precise", is not.

In part, my point is that I doubt the dropout alignment is meaningfully-worse after a cold-set than it was before. In other words, it was already "close enough" and it probably remains such. And if it isn't, improving it with simple hand tools is asking quite a lot.

Kontact 

Have you ever done the stuff you're talking about?

Spreading frames is a crude process. The dropout may end up deflecting more than .25 degrees because leverage might have been applied to it, and that might mean that more of a correction is needed.

Every bike I've spread benefitted from having the dropouts aligned. They were all slightly, yet observably, off. Luckily the alignment gauge is also the alignment tool, so I quickly aligned them and moved on.

TC1

Spread a frame to incorporate a different hub width? Yes, many times. Attempted to square planes to within a tenth or so of a degree, with simple tools? No -- for the reasons previously discussed.

Geometry is a relatively mature field. As I said, with common chainstays in the neighborhood of 425mm long, the angle change resulting from a 126->130 set is going to be just over 0.25 degrees. If your set results in more, then you stretched your frame too far. Or, you were careless and you bent your dropouts.

If you disagree with that, respectfully, I invite you to explain the relevant geometry that can cause some trapezoids with equal sides to have different angles.

I don't like saying this, but I seriously doubt anyone can see a quarter-degree on a ~32mm plane with a naked eye, and no reference. And as I've said, I very much doubt that adjustment with simple hand tools would improve that. On the other hand, confirmation bias has been proven to be exceedingly powerful.

To be somewhat precise, in order to see what you are describing above, one has to be able to see that the front and rear of the dropout mounting surfaces are misaligned by one thin human hair ( 70 micrometers ).

Kontact 

You are not understanding. I don't know if that is because of inexperience or you aren't reading carefully.

Again: If you grab the dropouts with your hands and pull them apart until they are 130, they will not necessarily be the product of hinging from the BB end of the chainstay, like you are supposing. Imagine that what is happening is that the chainstay actually bent right in the middle. Now your dropout angle is .5 degrees, not .25. Now imagine that the process of pulling succeeded in bending the stays, but also tweaked the dropout itself because that what what you were pulling on. Now your dropout is off by a lot more than a degree.

The problem isn't that .25 is an issue. The problem is that bikes don't bend in the ridiculously precise and predictable way you suppose. In fact, I'll guarantee you that the stay curves and doesn't just pivot at the BB lug. Geometry is a basic concept, so think carefully about what the dropout angle difference between hinging at the furthest extreme and bending along the length of the stay before you post your retort.

TC1

I specifically addressed the case of carelessly bending the dropouts -- I guess you didn't read all that carefully. I did assume that the operator was careful enough not to kink the tubes, but if we are talking about such a "crude" operation then clearly that operator is not capable of adjusting anything to a tenth of a degree -- which is my point.

You misunderstand how the calculation was arrived at -- but regardless of which, my point stands. You previously tried to sell a claim that the dropouts are precisely located in space with respect to the frame. That's clearly not the case if you are randomly bending your stays as you just described.

Adjusting a bent frame -- which is what you are now describing -- to alignment within one thin human hair of perfect, is challenging, to put it mildly. Probably almost no 126mm frames came off their assembly lines built to that tolerance.

bboy314

Here in the real world, when a frame is respaced, it almost always benefits from having the dropouts aligned with these - dropout alignment tools.

Kontact 

Dunno why you can't understand this. The stays aren't 'randomly' bent. They are bent and then checked against the rest of the frame. But no one has direct control of how they bend, just of where the dropouts end up.

And you made it clear you are basing the angle on a 425mm stay being moved 2mm to get about .25 degrees. Which is wrong, because the stay isn't going to pivot 425mm away from the dropout. It is going to bend closer to the dropout than 425mm, and probably in a curve. That makes the triangle a lot shorter than 425mm x 2mm, increasing the dropount angle.

https://www.calculator.net/right-tri...v=&x=Calculate

smd4

You realize we're talking about bikes here, not pocket watches, right?

TC1

That is basically my point. People here are talking about adjusting out a thin human hair of misalignment -- with crude hand tools. On an object that was almost-certainly not even originally-built to such a tolerance.

I'm observing that such an operation is quite difficult, and likely pointless, but I am fighting against confirmation bias, which is quite powerful. Even if the misalignment is several thin human hairs, the point does not change.

smd4

Gotcha.

Probably why I found it funny when 70sSanO suggested the index shifting of a 9 speed hub would suffer if forced into a 126 mm frame.

Kontact 

No, you're not. You're fighting your own fictional .25 degree number that you made up using the wrong information.

FBinNY 

Forgive me if I missed something, and if so, please disregard------

In all the "debate" about the 1/4 degree off parallel of dropouts, it seems that folks have lost sight of the reason. It has nothing to do with shifting or any noticeable performance issue. It's that if the dropouts aren't truly squared up and parallel they will cause the axle to bow when tightened. That means an unrelieved stress that can eventually cause breakage with threaded axles. It also effectively increases bearing compression (preload) possibly accelerating wear.

As to the tolerance, That depends on variables like the stiffness of the axle vs. the dropouts, threaded vs. non-threaded axle, amount of bearing preload, etc. Tossing it all into the hopper, shaking well and seeing what falls out, I believe that while correcting it would be nice, and I'd do so if I made the effort to cold-set the frame, it's not super critical if the dropouts were parallel all along.

TC1

First of all, why and how would I "fight a number"?

Second, I have said, all along: "9 mechanics out of 10 would probably make the alignment worse, rather than better, if they even tried to change it." and "Even if necessary, squaring dropouts after a cold-set from 126 to 130 on a 27" or 700c frame is anything but rudimentary. I have some quite nice metrology instruments, and it would take me some significant time and effort to align dropouts to a precision greater than a quarter-degree. I also doubt that frames other than very high-end models were even originally built with smaller tolerances than what we are discussing here. I further doubt that very many cast dropouts are even manufactured to sufficient tolerances to make this discussion meaningful." and "I'm not saying that any of this is impossible -- but it is not as easy as is being suggested, to achieve the kind of precision that we are discussing."

Amusing that you are calling anything "fictional" after claiming to be able to see a thin human hair -- or even several of them -- of misalignment with your naked eye, and claiming that dropouts are necessarily fixed-in-space with respect to the frame and that you align your chainstays by eyeballing them against the seat and head tubes.

Those were all quite funny, so thanks for that.

zacster

All this argument is well and good, but I did this 20 years ago and the bike and wheel are still operating just fine. This is the post that started it: https://www.bikeforums.net/road-cycl...uld-i-get.html I was only looking for a front but took the pair and they are still on that bike. I did nothing to the frame but jammed the wheel in there.

TC1

I think the bit you are missing is just the scale. The question largely is, "What does 'truly square' mean?"

There are people here who are insisting that "truly square" means parallel to within substantially less than the width of a thin human hair. Or, to use perhaps a more-apt analogy, to within the thickness of a typical layer of paint. That would mean that all of the problems you describe above might manifest if the paint on one dropout is scraped-away due to use. Or, if one dropout acquired a little dirt. Or if whoever painted your frame made an uneven pass over one dropout.

To put it mildly, I doubt that any commercial-grade bicycle hubs are that sensitive.

Kontact 

Who is saying that?
.
Hubs axles cut through paint.

FBinNY 

Don't bring me into this argument, especially since I basically took a middle position that while checking dropouts would make sense if cold setting, it's not necessary if they were right beforehand.

FWIW I'm more than casually familiar with tolerance, both from design/functional and manufacturing conciderations. My opinion is that using a decent pair of tools, any attentive mechanic can get them parallel to someplace between 0.25 and 0.50 degrees. Personally, I used to import, and still own a pair with 4" faces, allowing either greater precision or similar precision with less squinting.

FWIW, working back from the tools that have long been used for this job, I'd venture that folks like Campagnolo and Shimano would consider a functional tolerance to be within that quarter to half degree range, and am happy to accept their opinion on that score.

bboy314

I’m seeing only you bringing up the “human hair” argument, and your math seems to assume cold setting involves bending stays perfectly evenly, with the BB shell as the pivot, which just isn’t really how that works out in reality. Take it from someone who’s cold set dozens of frames and aligned many more dropouts.

Kontact 

Again, you made up the quarter degree, and it is wrong. I never commented as if it was right. The degree change at the dropout is likely to be greater than that for all the reasons I stated, and your straw man won't change that.

You also don't understand how the alignment gauge works, and that it shows misalignment both where the faces of the tool meet (32mm) and in the way the shafts point at each other (65mm each).

Hopefully any rational person reading this will understand that the dropout alignment needs to be checked after manipulating the stays.

Kontact 

Not just the BB shell, he's assuming that the stay pivots from the center of the BB, which is where the chainstay is measured from.

The longest distance you'll get any bending is at the chainstay bridge and brake bridge, but the typical tapered stays make it much more likely that the bending will occur much closer to the hub where the chainstay diameter is smaller and therefore less stiff.

TC1

I was not attempting to bring you into anything against your will. You said 'Unless I'm missing something', so I replied with what that might be.

0.25 degrees is the previously-mentioned thin human hair, or layer of paint, and I think that's asking a lot with "decent" tools -- although that word will have different meanings to different people.

For the most part, I agree with your previous post and wasn't arguing with you. I just think the tolerance on "truly square" in this application is larger than most here seem to.

TC1

No, you didn't even understand what I was talking about -- which isn't real impressive.

Fine, let's pretend I'm off by 300%. Now it is 3 thin human hairs. Or 3 layers of paint. Does that change the point at all? No, it does not.

I understand how the tool works. Unlike you, I also understand that it does nothing to square those faces with the frame. 'Square to the frame' is different from simply being parallel to each other. That's precisely why I asked about finding the centerline plane of reference -- and no one has proposed a method for that operation that results in the necessary precision.

Sure, check away, if it makes you happy. Hork on them with crude tools too, if you like. But if you think you are improving the situation by eyeballing it, and bending things with crude tools, you probably have another think coming.