Rear axle alignment deep dive
02-01-21, 03:34 AM
#1
Senior Member
Thread Starter
Join Date: Dec 2019
Posts: 956
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 321 Post(s)
Liked 263 Times
in
212 Posts
Rear axle alignment deep dive
Doug suggested in another thread that establishing alignment during build was worthy of a thread...
These are my thoughts on it in case anyone is interested. Obviously I'm a total amateur but I do have a method that now seems to work pretty reliably for me (after a good deal of angst and/or luck on the first two or three frames) without having a high precision jig.
The questions are: what can be wrong in theory, what matters, what can you measure, and what can you fix and how?
There are 3 possible translational errors. The axle can be too far up or down, fore or aft, or left or right.
The first two are things you get as close as you can with a ruler and angle finder while fixturing but they won't need fine-tuning and don't matter all that much. They will alter the BB drop and seat and head angles a small amount but these won't have any dramatic effect. Left and right translation however is important-- it will put the two wheels in different planes and mean you're crabbing down the road.
There are a further 2 possible rotational errors: roll and yaw. Whether the rear wheel is tilted (roll) or angled as though it were steering left or right (yaw). Pitch is not an issue because the axle is symmetrical about that axis. Roll and yaw are very important.
So 3 errors that matter. What can you measure? The string method measures the translation error directly. Putting a known true and correctly dished rear wheel in will, assuming no translation error, reveal roll errors (wheel is closer to one SS than the other) and yaw errors (it's closer to one CS than the other). The wheel magnifies these errors so it's a pretty good way to check for them. When checking with a wheel always try it backwards as well as forwards in case it's not as correctly dished as you thought.
What you can fix? The easiest thing you can do is bend the stays open or shut a little bit more on one side than the other. This changes translation and yaw at the same time. If you bend the TT and DT a little you can translate the HT left or right instead and this will correct translation without affecting yaw. But this is much harder to do.
You can change roll very easily if you haven't attached the SS yet, just by bending one CS up a little and the other down. If the SS are on there your only option is to file the top of one of the dropouts a little. Likewise you can change yaw without changing translation by filing the front and back of the dropouts a little.
If the dropouts are horizontal track-ends, you never need to file the front and back-- you are adjusting the yaw anyway every time you fit the wheel. If they're those old-school slots that go up and backwards like all bikes used to have you can adjust roll and yaw every time you fit the back wheel (with small changes in fore-and-aft and up-and-down translation which don't matter). Vertical dropouts can be filed a bit. Thru-axle dropouts cannot be touched.
It should not be necessary to file dropouts on a new frame you're making but can get you out of trouble. Don't attempt a thru-axle for your first frame.
Because bending the CS changes yaw and translation at the same time, don't try to measure or fix yaw until you have set the translation right. Rig up string (or use a straight-edge) and get the two dropouts equidistant from the centre-line first. Now if the wheel is too close to one CS than the other you're going to have to get that file out.
On a new frame it's critical to fixture up the rear triangle with high precision. The BB and rear axle must be parallel, the dropouts centred and the CS must be exactly the same length. And I mean exactly. If the two dropouts are the same length (which they usually are unless one has a caliper tab or something) you can hold your two CS next to each other before you've attached them to anything and check them like that. If the dropouts slide in and out of the CS you have a bit of wiggle room there. But not for breezer dropouts. You also want to fixture it without any roll errors obviously but these are much easier to correct before the SS are attached, which should always be the last tubes added to the frame.
It's also a good idea to leave a dummy axle (I just use a threaded rod with four nuts inside the dropouts-- one to set the position each side and a locknut) and two outside securely bolted between the dropouts at all times while welding, brazing, or dropping the frame on the floor by mistake. Just leave it in there all the time. This will help to minimize distortion.
If it was all correct in the fixture and the CS were the same length then after welding you will usually find that the dropouts have closed up, usually symmetrically. But whatever the case, if you rig up the string and correct any translation error you will find the yaw is correct, or if you put a wheel in and correct the yaw the translation will be correct.
The way I make a frame is: BB shell to ST, fully welded. CS to BB and dropouts, fully welded both ends. Correct the spacing symmetrically and put a rear wheel in to check the yaw. It's always been perfect. Then I line up the centre of the rear axle (the dummy axle is marked in the centre) with the ST and the HT using a laser level (and also use this to get the HT and ST parallel which is very important). Then I double-check with string. Then I tack the front triangle and then weld it. There is a possibility that during this stage a translation error might be introduced due to distortion. If that happens there are only two fixes: try to bend the front triangle; or correct the translation by pulling the CS in or out, but you will now have a residual yaw error, because if it was correct when you tacked the HT, and things moved, it won't be now, and you will now have to file the dropouts. Not good. Fortunately this hasn't happened on my last three frames where I have been using this method. Errors I ended up on two of my earlier frames that resulted in filing were more because of poor fixturing than welding distortion. Then it's CS bridge, then SS then SS bridge. Installing "seatstays" made of string wrapped around the TT and then fitting a wheel is a good way to check for roll errors before attaching the actual SS.
Most people make the front triangle first and then attach the CS. The only reason I do the CS before the front triangle is because it's easier for me to fixture without the front triangle in the way. It also makes access to the tricky CS to BB welds a bit easier because you can turn the whole thing on its back on the table.
These are my thoughts on it in case anyone is interested. Obviously I'm a total amateur but I do have a method that now seems to work pretty reliably for me (after a good deal of angst and/or luck on the first two or three frames) without having a high precision jig.
The questions are: what can be wrong in theory, what matters, what can you measure, and what can you fix and how?
There are 3 possible translational errors. The axle can be too far up or down, fore or aft, or left or right.
The first two are things you get as close as you can with a ruler and angle finder while fixturing but they won't need fine-tuning and don't matter all that much. They will alter the BB drop and seat and head angles a small amount but these won't have any dramatic effect. Left and right translation however is important-- it will put the two wheels in different planes and mean you're crabbing down the road.
There are a further 2 possible rotational errors: roll and yaw. Whether the rear wheel is tilted (roll) or angled as though it were steering left or right (yaw). Pitch is not an issue because the axle is symmetrical about that axis. Roll and yaw are very important.
So 3 errors that matter. What can you measure? The string method measures the translation error directly. Putting a known true and correctly dished rear wheel in will, assuming no translation error, reveal roll errors (wheel is closer to one SS than the other) and yaw errors (it's closer to one CS than the other). The wheel magnifies these errors so it's a pretty good way to check for them. When checking with a wheel always try it backwards as well as forwards in case it's not as correctly dished as you thought.
What you can fix? The easiest thing you can do is bend the stays open or shut a little bit more on one side than the other. This changes translation and yaw at the same time. If you bend the TT and DT a little you can translate the HT left or right instead and this will correct translation without affecting yaw. But this is much harder to do.
You can change roll very easily if you haven't attached the SS yet, just by bending one CS up a little and the other down. If the SS are on there your only option is to file the top of one of the dropouts a little. Likewise you can change yaw without changing translation by filing the front and back of the dropouts a little.
If the dropouts are horizontal track-ends, you never need to file the front and back-- you are adjusting the yaw anyway every time you fit the wheel. If they're those old-school slots that go up and backwards like all bikes used to have you can adjust roll and yaw every time you fit the back wheel (with small changes in fore-and-aft and up-and-down translation which don't matter). Vertical dropouts can be filed a bit. Thru-axle dropouts cannot be touched.
It should not be necessary to file dropouts on a new frame you're making but can get you out of trouble. Don't attempt a thru-axle for your first frame.
Because bending the CS changes yaw and translation at the same time, don't try to measure or fix yaw until you have set the translation right. Rig up string (or use a straight-edge) and get the two dropouts equidistant from the centre-line first. Now if the wheel is too close to one CS than the other you're going to have to get that file out.
On a new frame it's critical to fixture up the rear triangle with high precision. The BB and rear axle must be parallel, the dropouts centred and the CS must be exactly the same length. And I mean exactly. If the two dropouts are the same length (which they usually are unless one has a caliper tab or something) you can hold your two CS next to each other before you've attached them to anything and check them like that. If the dropouts slide in and out of the CS you have a bit of wiggle room there. But not for breezer dropouts. You also want to fixture it without any roll errors obviously but these are much easier to correct before the SS are attached, which should always be the last tubes added to the frame.
It's also a good idea to leave a dummy axle (I just use a threaded rod with four nuts inside the dropouts-- one to set the position each side and a locknut) and two outside securely bolted between the dropouts at all times while welding, brazing, or dropping the frame on the floor by mistake. Just leave it in there all the time. This will help to minimize distortion.
If it was all correct in the fixture and the CS were the same length then after welding you will usually find that the dropouts have closed up, usually symmetrically. But whatever the case, if you rig up the string and correct any translation error you will find the yaw is correct, or if you put a wheel in and correct the yaw the translation will be correct.
The way I make a frame is: BB shell to ST, fully welded. CS to BB and dropouts, fully welded both ends. Correct the spacing symmetrically and put a rear wheel in to check the yaw. It's always been perfect. Then I line up the centre of the rear axle (the dummy axle is marked in the centre) with the ST and the HT using a laser level (and also use this to get the HT and ST parallel which is very important). Then I double-check with string. Then I tack the front triangle and then weld it. There is a possibility that during this stage a translation error might be introduced due to distortion. If that happens there are only two fixes: try to bend the front triangle; or correct the translation by pulling the CS in or out, but you will now have a residual yaw error, because if it was correct when you tacked the HT, and things moved, it won't be now, and you will now have to file the dropouts. Not good. Fortunately this hasn't happened on my last three frames where I have been using this method. Errors I ended up on two of my earlier frames that resulted in filing were more because of poor fixturing than welding distortion. Then it's CS bridge, then SS then SS bridge. Installing "seatstays" made of string wrapped around the TT and then fitting a wheel is a good way to check for roll errors before attaching the actual SS.
Most people make the front triangle first and then attach the CS. The only reason I do the CS before the front triangle is because it's easier for me to fixture without the front triangle in the way. It also makes access to the tricky CS to BB welds a bit easier because you can turn the whole thing on its back on the table.
Last edited by guy153; 02-01-21 at 08:33 AM.
Likes For guy153:
