Kontact

Right - which means that not "anyone who's ever cold set a rear triangle can attest to that". So far, I can't.

Ghrumpy

There's nothing I can think of that is necessarily or inherently self-centering when mismatching hub OLN and frame spacing.
The "danger" is more long-term, not immediate. It puts the side that bends more in a constantly stressed state. When subjected to the regular stresses of riding, it is likely to push that stressed area past its limits, causing it to fatigue more quickly than if it weren't pre-stressed.

This of course is also why dropout faces need to be parallel, which they won't be if they are being squeezed together more than a few mm. This puts the axle at risk as well as the frame. I'm not sure whether the JIS tolerance limit is more for one or the other, to be honest. I suspect it's more for the frame, if for no other reason than it's easy to replace a broken axle, less easy to repair or replace a broken frame.

As best you can tell. But how accurately are you able to assess that? I'm not doubting, just asking. The string test or Park F.A.G.-2 are only as accurate as their reference points, and head tubes and seat tubes are not always perfectly aligned to each other or the BB axis.

FWIW, I did say "may," and while IME it is more likely than not, that does leave open the chance for up to 49.9% "not."

Kontact

This is an interesting discussion, not a "big argument":
Considering that all four stays are the same length and made of the same tubing, they should flex in a very similar way. Considering that we expect bike frames to flex when we ride them, it would be problematic if they didn't do so in a roughly symmetrical manner.


Here you are suggesting that it is dangerous to allow something that is more flexible flex more. I don't think that is an accurate way of looking at things. If you have too pieces of spring wire of the same metallurgy, and one is thinner and twice as flexible, that really means that it can safely bend much further than the other before it gets to yield. It shouldn't be a bad this to flex something flexible


How accurate is accurate enough? We're talking about a flexible bike frame - at what point could a misalignment that we can't see or measure actually affect the way the bike rides or works?

In the case of cold setting stays, if the F.A.G.-2 says that the dropouts started out symmetrical, and we assume the frame was otherwise aligned, why wouldn't a final F.A.G.-2 measure of symmetrical be accurate? After all, the bending axis of each side is a line between the two bridges, which is roughly parallel to the seat tube that the F.A.G.-2 measures from. How could that sort of bending change that relationship enough to introduce a twisted final shape?


Most of the bikes people spread are 126mm steel road frames, and the vast majority of those have no chainstay clearance dents - so they really are highly symmetrical. It is reasonable that many of them will reach yield at pretty much the exact same amount of deflection and then bend a similar amount. I don't know how many people with 130 MTBs ever bother to take them to 135, but you're more to likely see assymetrical chainring clearance dents with those (if the dents actually matter).


I personally like spreading frames freehand because it doesn't stress the frame anywhere but where it should be - no 2x4s on the seat tube - and because it seems to usually work with the least effort, and can be individually corrected if it does not. People see this as crude, but it has always seems like the most direct way of applying the right kind of bending force. If you are measuring your incremental progress with a F.A.G.-2, you can tell if it is not spreading symmetrically before you have to un-do any of the bend.

Ghrumpy

Yeah, we're cool.
Sure. That's the ideal. But bike frames and their component parts are assembled to tolerances, not to ideals or assumptions. Tolerances are really what we're talking about here.
I can't argue with your example, but it don't think it really illustrates what I was talking about. For one thing, we're not talking about something twice as flexible, just slightly more flexible. More to the point, any asymmetry in flexibility (on a rear triangle anyway) is, as you just said, not by design.

Let me clarify that bike frames are more importantly meant to be elastic, not merely flexible. Under normal riding circumstances anyway. And elasticity has limits. That was the point of my stress example. Let's use one of your spring wires to illustrate. If you start with it at rest, then won't more likely yield to one side or the other if bent the same amount. But if you start with it already loaded in one direction, then bend it the same amount in both ways, then the pre-loaded side can be taken past its elastic limit much more easily.

Hard to answer. That's a personal decision as much as anything. From what I have seen, some people don't seem to notice or care that their bike is crabbing down the road like a Swing bike, while others are like Princesses and the Pea, and slight misalignments make it practically painful for them to ride. It's best to assume they will be the latter, and correct them as much as possible. If they don't care, so much the better.

Obviously that depends on your tools. The more relevant question is that, as I said, a measurement is only as accurate as its reference points. If they start out inaccurate, then it doesn't matter how precise your measurement tools (or corrections) are, as they will only carry those reference errors forward.

A string test or F.A.G.-2 both depend on the vertical alignment of both the head tube and seat tube to an imaginary plane that is perpendicular to the BB thread axis, and centered at the middle of the BB. Unless that is checked and corrected first, there's literally no way to know how accurate their measurements of the rear triangle will be. And any inherent misalignment will be exaggerated on larger frames. Oh, they also depend on the seat tube being perfectly straight (I've seen seat tubes warped from brazed-on water bottle and FD bosses.)
I can tell you from many years of checking and aligning steel frames that it is rare that all these line up perfectly enough for me personally to take string test or F.A.G.-2 measurements with much confidence.
I totally agree. But as I said, my experience prevents me from assuming the frame was otherwise aligned.
The dents do seem to matter. Probably so do things like inconsistencies in tubing wall thickness, heat application during joining, curved/bent shapes, inherent stresses, work-hardening, etc. Honestly, it is more rare IME that the two sides of a rear triangle behave in the same way than that they don't. I'm not sure I understand fully why that is in every case. But it is.

I agree to a point.
When you try to take a frame past its elastic limit, as when cold-setting, you have to move it something like ten times farther than you want it to end up. So a 5mm spread may require 50mm of displacement, maybe more.
Also, the structure will yield at its weakest point, which may not be anywhere near where you're actually applying the stress. It may not be on the tube at all, because you're stressing the entire structure.
So I prefer to focus that stress in a spot where there is less likely to be one of those aforementioned inconsistencies, namely the center of the tube, away from dents, curves, heat-affected zones etc. Pretty hard to do that freehand. I've spread more than a few 120 frames to 130 for restomods. Freehand bending becomes more risky in that case. So I prefer to use a Park FFS with the fulcrum on the center of the tube (with a firm but not hard plastic shield under it to keep from marking the frame, the kind that are often used on boxed bikes to prevent dents and scratches) and clamp the brake bridge area of the seatstays. And I always check the front triangle again after doing the rear.

Kontact

I think this is the point we are conflicting on. If one side moves more within the elastic range than the other side, it is because it is less stiff/more flexible. Both sides will move in proportion to their stiffness. Both are receiving the same amount of "work", it is just being used differently by each side. I just don't think that flexing the flexible side a greater distance while still within the elastic limit matters when the same thing is happening on the stiff side. They will each yield when they hit the amount of force that is the yield limit, not a distance. So I don't see a special danger if you don't cold set and one side of the stays flexes 4mm out and the other one only 1mm. Clearly the 4mm side is more flexible and can therefore tolerate the flex.



At this point you really aren't talking about spreading rear triangles by whether frames come aligned in the first place. My point was that the F.A.G.-2 was a reliable enough indicator of change in rear dropout location from what we assumed to be reasonable alignment in the first place. There is nothing about the bending movement that would cause the dropout to move in any plane outside of the bending arc.

But spreading the dropouts isn't part of process to correct frame alignment. It is just a change that references the alignment already there and makes a change that preserves the alignment but spreads the dropouts. The F.A.G.-2 seat tube reference is as good as any. No spreading process short of an alignment table is going to address a lack of alignment, except a frame with already off-center dropouts.


I didn't say that the dents didn't matter - I was pointing out that most of the bikes that get cold set don't even have them.


This is the first time I've heard of someone using an FFS to spread the rear triangle. And while I appreciate your interest in bending the stay away from its joints (even though that's exactly what your do when you align the dropouts), I fail to see how the FFS can offer the kind of evenness you are demanding when it can only bend either the chainstay OR the seatstay, but not both simultaneously. You would be better off putting a bar across the point on both stays around which you want them to bend, and then leveraging the center of that bar to affect both at once.

Personally, since we don't sweat that sort of thing when aligning dropouts, I don't sweat it at the other end of the tubes. Since the smaller bend will make more of a difference the further away it is, I would prefer to see the tubes bend close to the clamped brake bridge and BB shell - 2mm of bend over 370mm of stay is only 0.3° at the bridge - the minimal amount of yield "damage" to the structure. But really, I'm allowing the stays to bend wherever they most naturally want to (the joints, the butts, etc), which seems more natural to the structure than creating a curve where the tube is perhaps most resistant to bending. AND, if the tube is making its 2mm change from mid-span instead of the furthest points, you are doubling the amount of dropout bend it will take to align them.

In other words, a bend at 370mm (brake bridge) is a 0.3° bend there and a 0.3° bend at the dropout to counter it. But a bend at 180mm down the stay is a 0.6° bend which requires a 0.6° bend at the dropout to counter. That's twice the bending on four brazed joints for a theoretical decrease in bending near the other four. Hardly seems like a bargain.

Dan Burkhart 

When I spread dropouts, I don't concern myself with how evenly they are spreading, as I can deal with that after the fact.
Here's one of my long boring videos detailing how I deal with dropout spacing adjustment.

Ghrumpy

I wouldn't say we're conflicting, we're just talking about different things. I get what you're talking about. Your example is why bending a chainstay will also bend the seatstay it's attached to, because the seatstay is more flexible. And it's not a problem for that seatstay.

I'm talking about a residual stress caused by an imbalance of stiffness or alignment between two members that should not have that imbalance. I think I can safely say that all steel frames have them. Most of the time they aren't big enough to cause problems. But if the stress is big enough, it can result in accelerated fatigue at the weak point of the structure. Might be the frame tube, might be the dropout, might be the weld or lug. Might be the hub axle. I've seen frames that literally tore themselves apart because of these residual stresses. It's not just a theory.

That's really the relevant point WRT why the mismatching of the OP's hub OLN and dropout spacing is a problem. Some of this other stuff is more academic. Not that it's not interesting and worth discussing, but I don't like to veer too far off the thread's original topic.

Again, it's a matter of tolerances, both mechanical and personal. Clearly the F.A.G.-2 works for most people and most shops most of the time.

This is the academic part. I brought up alignment because the F.A.G.-2 assumes correct front triangle alignment, and I don't. Let's say you want to spread a bike from 130 to 135. To preserve the original alignment, you should only move each side 2.5mm, right? If you just spread the rear triangle to 135, and only afterwards check and correct the dropouts' relative location with the F.A.G.-2, you have no idea whether you're changing the alignment of the frame. You should use the F.A.G.-2 to baseline the original dropout locations, which might or might not be exactly equidistant from the F.A.G.-2's indicator. Is that what you do?

Maybe you're right, but I don't know that I can make such a statement. My experience includes bikes from earlier than the 1990s, which might be very different from yours.
And in any case, don't get hung up on the dents, because as I pointed out there are plenty of other factors that might cause two sides of a rear triangle to behave differently.

I think we both do what we think is better. Just to be clear, when I'm talking about "better" I'm not talking about better than you, I'm talking about better for the frame.

First off, the bend that I introduce into the stay is very gradual. It's not a dogleg, but as gentle a bow along as much of its length as I can work with. I will often work the FFS at a few locations along the stay to create that curve. Usually the seatstay just goes along for the ride, but occasionally I will have to work that the same way to avoid putting a bend right at the brake bridge, especially with bikes going from 120 to 130. (One old English bike I did was originally 4½" or 114mm rear spacing, and the owner wanted 130. I couldn't get it there without risking kinks in the stays, so I stopped at 126.)
I try to avoid putting the bend close to the ends of the tubes because they are already the most highly-stressed part of the structure. I'd prefer to let them stay at rest. They are also a heat-affected zone, and I try to avoid putting a bend in a HAZ, because who knows how hot that area got. Too hot and tubing becomes more brittle, leading more easily to cracks. That's what I'm trying to avoid by putting the bend in a lower-stress area of the tube.

That's the problem I have with letting the bend "naturally" happen, because IMPE it may naturally happen in the worst place. It's one thing for a framebuilder to align a freshly-built frame and let the bend naturally happen. It's another thing to take a frame that's been ridden and work-hardened, perhaps hard and/or for decades, and force it to a different shape. Who knows what you're stirring up. (WRT framebuilding and alignment, there's a cool video out there showing a Japanese builder "witching" his rear triangle into alignment with a torch, heating one side of the middle of the stays and letting the heat cycle move the frame where he wants it to go. That's kind of what I'm doing with a tool instead of heat.)

As far as dropouts are concerned, that's a totally different matter. Flat ones anyway allow the kind of alignment required to square up the faces at least a couple times. It's not that it's not a problem, I've broken a couple dropouts on my own bikes. Took about 20 years though.

Kontact

The F.A.G.-2 is an alignment tool. You use it to center rear triangle to the front triangle. It will not fix a twisted front triangle, but it will usually help detect if the head and seat tubes are out of plane. It mainly exists because it is relatively easy for the thinner stays to get off center compared to the stout main triangle from getting wonky.

When I'm using the F.A.G.-2 I do an initial read. If the bike appears square, then I go ahead with spreading the rear, using the tool to monitor how it is spreading.

If the initial read is off, then something is odd with the frame. If the tool is sitting flat on heat and seat tube, then either the seat tube isn't straight or (more likely) the rear triangle is off center. At that point I check the seat tube with a straight edge and the rear triangle with a known dished wheel. So far, it as always been the rear triangle off center. If it isn't grievously off center I would then align the triangle and change the spacing at the same time by putting each side where it should be for alignment and spacing.



But do you get what I'm saying about the dropouts? The closer you put your stay corrections to the dropout, the greater the number of degrees of bending you'll have to do at the dropout to realign it. Greater bends do more damage.

A better tactic would be to clamp the stays just behind chainstay bridge and bend around that point. That's a largely unstressed section of tubing and is far enough from the dropout to make that transition as small as possible. A very gradual S-bend rather than a tight one on the stay/dropout joint.