Regaining frame aligment

Question

I bought a belt driven bike which had had the rear triangle of the frame moderately bent (i.e. bent but no tubes collapsed or cracked) – it was a gamble as to whether it’s fixable…. It has a Nuvinci 330 hub with disc brakes which requires a large 135mm dropout spacing.

The frame is made of steel and I’ve been able to bend it to near alignment. However, as I don’t have a reference bike to compare measurements from, I’m struggling to figure out whether I’ve aligned it properly or not.

Firstly I aligned it based on the string test (Sheldon web page and also a Youtube video) so that the distance between the string and the seat tube frame member was equal both sides. I got this right to within 1mm and was happy with myself.

But then I dropped in the rear hub and tried to pedal it. It turned but it was making a noise which wasn’t very happy and you could tell the hub wasn’t aligned with the pedal wheel, with the belt rubbing as it seated itself on the gearwheels.

So then I bent the frame some more and now turning the pedals works really well and looks like it’s all fitting nicely, however if I do the string test there’s around a 6mm difference in the distance between the string and the frame between the two sides. However if I measure distance from each rear axle end to the same point on the headset it’s the same distance within measurement error, so I think the axle is sitting perpendicularly to the line of the frame.

So the question is this. Does all of the above mean my frame is bent in some way where I just won’t be able to fix it? The bottom bracket is one of those split types so that you can tension the belt – it looks horrifically weak structurally to my eye and it may well be that the bottom bracket axis is no longer perpendicular to the frame plane which I doubt I could fix. Or maybe a 6mm difference in the string test is not enough to worry about? Or are some frames deliberately made so that the string test should not return the same measurement which would mean that the way I have it now is perfect?

Thanks in advance for any help.

PS. I am waiting for a rim before I can restring the rear wheel and this is taking a while to arrive – hence I’ve not been able to ride it to see how it feels.

Answer 1

A couple things here that could be tripping you up:

• Both belt drive and Nuvinci hubs can be noisy all by themselves. Belts frequently also make noise in the stand that vanishes when riding and vice versa. Getting belt tension just right is a whole other step and noise will frequently be the result when it's not dialed. Lubricating it is another piece. (I always use silicone cleaning spray, and yes this is a Gates-sanctioned thing). If the rear cog was visibly out of line with the front, then perhaps this isn't likely to be the problem, but worth mentioning.
• When you bent the frame around the second time and the noise went away, it could have affected the belt tension.
• If somehow the chainline (beltline) was incorrect with the parts that are on there, it could create the situation where it made noise when the alignment was right and didn't when it was wrong.
• Doing what you're attempting without dropout alignment tools or an improvised substitute is iffy. What they do is integral to the whole process of aligning a frame after it's been severely misaligned.

You mention that you're seeing some torsional misalignment between the BB and the front wheel. A way you could go about trying to fix that is taking the fork out, clamping the BB in a vise with some old steel cups, removing the fork and headset, and getting a long steel pipe that mates closely with the inside of the head tube to use as a leverage bar. Then to check the alignment (which you should do first) you can hold the frame up against a light source and eyeball the parallelism between the seatube and headtube. This isn't likely to have any bearing on rear torsional alignment though. You need dropout tools for that.

Something you can do is take a long, straight piece of square aluminum extrusion, like 80/20, and use it as a straight edge referenced off the bottm bracket. For example, clamping them together in a vise, measuring with a caliper between the seat and the straight edge all the way down their lengths, repeating for the down tube, then removing the straight edge, re-clamping the bb, and reefing on the tubes. If there is shell misalignment, something like this is how you'd fix it.

Answer 2

Well, it was quite involved but I managed to get the frame straight. I think it would be helpful for me to write the experience of how I managed it as it might save someone some time.

And the other thing is I did all this on the kitchen floor of an inner city appartment - so all of the fancy tools/garage were not available to me.

First off the video that made me think it's possible to try this is this one: Video

Unfortunately that video makes it deceptively simple as it only checks for alignment in the stays. And it uses an expensive tool. Instead I strung up the string between the dropouts and front of the frame as shown here String video. The string measurement was about 10-15mm difference between the two sides. This is material, but because nothing was crushed I thought I had a chance.

So I bought a piece of wood. The shape of the wood is important - the video uses a 2"x3" but the length is also important and he doesn't go into it much. The length matters because it needs to be able to pass past the front of the frame. So if you're doing this, buy a really long piece of wood and cut it once you see what I mean. Also, I found the way the video shows you bending the frame didn't work for me as it wasn't getting the frame to move much at all. Instead I threaded a much longer piece of wood through the frame the same way as in the video and laid it horizontally on the floor and then pressed the frame towards the wood - you can get significantly greater leverage than that the video shows.

However, the video also simplifies as bending the stays might well twist the bottom bracket casing which is not shown. To check for this, you need to measure the offsets between the left and right cranks and the seat and front tubes. I found it helpful switching one of the cranks so they're both at the same position rather than spun round 180 degrees as they are when riding. If the cranks are both turning in planes parallel to the main frame plane then the left right offsets will be the same. In my case after I thought I'd straightened the stays there was a good 6-7mm difference at the front tube which is again pretty material. The seat tube was thankfully aligned with the same l/r crank offsets - if that was gone I suspect the job becomes impossible as bending the bottom bracket to a precise angle becomes too hard.

I then bought a 5' scaffolding pole and some PVC pipe to use as a sleeve and make the scaffold pole fit exactly into the bottom bracket casing. Even with a 5' lever it took two people, one of which had to carry 18l of water to hold the frame down horizontally on the ground while the other pulled the end of the pole. It was super super hard, but believe it or not it managed to move.

So now you have control over all the misaligned bits - the stays and the bottom bracket. It's important to stress that there's a limit to how accurately you can regain alignment, and given that limit the operation becomes:

1. Measure l/r crank offsets at both seat and front tubes
2. Measure l/r string offsets
3. Measure dropout spacing
4. Write them all down and calculate the biggest error.
5. Apply a bend to that biggest error
6. Go to step 1 unless you're happy with the frame

It took me about 10 iterations total to arrive at a frame which is within 0.5mm aligned on all of l/r string offset, l/r seat tube and front tube offset and dropout spacing. It takes forever because every bend you do you have to set-up for a bend and then set-up for re-measuring everything. And sometimes you'll think you're bending but you won't have applied enough force to pass yield strength and other times you'll apply too much force and over bend. It's super important to write it all down because if you get confused you'll bend the wrong way and then you'll get annoyed and that will make you make even more mistakes and take even longer.

So anyways, in case you end up with a frame as badly misaligned as the one I bought there is hope. Just don't underestimate how long it takes - I probably spent a couple of days figuring all this out. But so far delighted with the resulting bike - it's as good as new.

Answer 3

As you mention that the bottom bracket shell has a system for tensioning the chain, I'm guessing that the rear wheel axle dropouts are not slot types that you can use to tension the chain and align the rear wheel.

Probably what happened is this: Using Sheldon's string method you got the rear dropouts equal distance from the center plane of the frame (the plane that the main frame triangle lies on), but the axis of the rear axle was not perpendicular to the center plane - either one dropout was further forward than the other, or one was lower.

If you fit a bare axle into the dropouts you can check to see that it is perpendicular the center plane. You can use an old threaded wheel axle (ask your LBS if they have an old one lying around, but make sure it's not a bent one), long 10mm bolt or section of 10mm threaded rod. With a spirit level set the frame up so the seat tube is vertical left to right then check the axle is horizontal. Also measure the distance from the bottom bracket axle center line to the axle center line on both sides to check they are equal.

Additionally, the dropouts may not be parallel with each other. There are tools for aligning dropouts but you can also use two 10mm bolts of about half the dropout spacing in length.