Is this probably a bent derailleur hanger?

[Manc33]

I have a new frame and a 12-36t cassette + wheel from a setup that was working perfectly.

When setting up the rear shifting, if I set it so it shifts from 12t-14t, then up at the 32t sprocket it's almost trying to shift to the 36t.

If I set it up so the it is not trying to shift to the 36t from the 32t sprocket, then I get only about 1/4 of a shifts worth of movement from the mech at the 12t-14t shift.

Giving the shifter a little push does get the mech on the 14t and it doesn't want to shift back to the 12t.

If I do that 1/4 shift it's like the very last bit of travel in the mech at the smallest sprocket is not there, I mean the mech can be moved up slightly and not compress back again which I would normally assume is bad cable housing.

On the 12t the cable is loose so I started winding out the high limit screw on the RD but the amount that would need to be wound out to alleviate the slack in the cable puts the mech about half a sprocket width too far past the 12t, I am surprised the chain didn't come off the smallest sprocket, but I can tell by lining it up by eyesight it's miles too far over, if the slack is taken out of the cable.

It seems like housing but I think it's a bent hanger. I know the tools to check that are silly prices and you can make DIY tools but that's faffing I just wondered how common a bent hanger is on a new frame. The frame was cheap, but brand new.

I saw someone say if you have an old wheel with an axle, that has a M10 axle and you can screw it into a derailleur hanger. I have no idea if the thread sizing is the same but if so, this means you can screw an old wheel into your hanger and check alignment that way, where the screwed in wheel becomes your alignment tool. Then I would just measure at the valve on the problematic frame and so on the way it's normally checked.

One website said up to 1 in 4 new frames do have a misaligned hanger.

I happen to have an old wheel and might try to align it it that way, since I don't have access to one of those drills that drills perfectly downwards (which would be imperative for making this sort of tool) nor do I have 16" x 2cm x 2cm steel tubing to make the tool out of, or even a M10 bolt with nuts, although it's a clever bodge since those tools are daft prices.

[rogerzilla]

You can do it with a long steel rule, using the seat tube and BB shell as a reference. Screw an M10 x 1 rear axle into the hanger first to protect threads from distortion and to use as a lever. Check for verticality and skew - often you find both. I realigned a very bent hanger on a Kona Cinder Cone and shifting is now impeccable.

[Brucey]

photos?

cheers

[NetworkMan]

On the 12t the cable is loose so I started winding out the high limit screw on the RD but the amount that would need to be wound out to alleviate the slack in the cable puts the mech about half a sprocket width too far past the 12t, I am surprised the chain didn't come off the smallest sprocket, but I can tell by lining it up by eyesight it's miles too far over, if the slack is taken out of the cable.

Sorry, I don't understand; on a 12-36 cassette the cable is intended to be slack on the 12T isn't it since it's the upper limit screw that determines the position of the RD on the top sprocket, not the shifter and cable tension.

[Manc33]

Sorry I can't do photos. I screwed an old wheel into the hangar last night and it seemed a little out by eyesight so I got the wheels parallel then tried the shifting again, it's the same as it was, needing to have a loose cable on the smallest sprocket to be loose enough to not have it making that tinkling noise (wanting to shift to a lower sprocket) when on the 32t (the lowest sprocket is a 36t).

I reckon it's just the cable routing, which is a joke on this bike (has an S bend in the middle of the housing run, to avoid the rear shock).

Although it's not got internal routing, it's behaving like it has.

I might end up doing what I did on my last bike that had that, and just re-route it completely, with cable ties. That sounds like it wouldn't work but it gave me proper shifting back last time, it's just a pain avoiding the shock spring.

I noticed if I put the rear mech on the smallest sprocket and move the shifter a 1/4 of a shift to move the mech, the mech does not collapse again like it should, so that's definitely bad/squishy housing. Another thing is I am not using SP-41 housing, it's some cheap no name stuff, but then it was on my last bike and that worked. I should just get some SP-41 I guess. Maybe it's those little rubber bungs inside the ferrules (I forgot to check might be there) that are said to keep out gunk, but to me are a pain because they cause compression in the housing run - which doesn't matter on a front shifter - but on a rear it throws it out in my opinion, I'd rather have gunk possibly getting in and have decent shifting. I thought I double checked all my ferrules had those taken out but I might have missed a few.

I posted about that a few years back: viewtopic.php?t=106552#p1019336

The fettling continues...

[Chris Jeggo]

Some time last year I took a Shimano 9-sp 12-36 cassette out of my spares box to replace a worn out one, and after fitting it I found I had similar symptoms to those you describe. If the wire was adjusted to give good alignment on the 14 sprocket, the alignment was wrong by the time I had clicked across to the 32-36 region, and vice versa. After some investigation I found that the total width of the cassette was too small - see https://www.sheldonbrown.com/cribsheet-spacing.html. How could that happen? Wrong sprockets? Wrong spacers? I did a quick fix by cutting some shims from an aluminium beer can (approx 0.1mm thick) and re-spacing the cassette so that it worked.
So, check your sprocket pitch.

Edit: The cassette I took out of my spares box was unused.

[Brucey]

cassettes which have plastic spacers can change spacing in use; essentially the plastic 'creeps' under the sustained load of the lockring and the sprocket pitch varies. Occasionally plastic spacers will shrink because of their own internal stresses (they will try to revert back to some other shape). If the lockring isn't tight then the spacers can wear in use, too. Normally all this happens at the same time as the chain wears etc and everything keeps working. But if you (say) fit used/old cassette to a wheel then you will occasionally find that the cassette pitch is obviously slightly wrong.

Back in UG days the top sprocket got tighter every time you sprinted in top gear. Quite often the sprocket would move, the (plastic) spacers would creep, and then (days or weeks later) the sprocket would move again. Each time it happened this would throw the indexing out of course. I got so cheesed off with this that I machined up aluminium spacers for all my UG cassettes; much better. However the spacer width needs to be absolutely correct if cumulative errors are to be avoided; IIRC my spacers needed to be 3.05mm wide and if a whole set was (say) 3.00mm instead the result was less than satisfactory.

These days HG cassettes impose a lower load on the spacers and they make them from different (more creep resistant) plastic. But tolerances are tighter. I still prefer cassettes with metal spacers, because I have fewer problems with them.

cheers

[Manc33]

Brucey I know what you mean. They should just be making the cassette spacers out of metal at 600/Ultegra level, but don't.

Here's what I can rule out:

- Cheap cable housing - because that same £1 a meter stuff was used on 2 other bikes, one a carbon frame bike where I had an entire run of housing from the shifter all the way to the rear mech in 1 piece (cable tied to the outside of the frame to get around the silly internal cable housing it had). Shifting worked on that bike and the next bike with a more conventional housing setup.

- Cassette - same cassette and wheel is being used, the cassette was never taken off the wheel. It was working perfectly on the last bike with a conventional housing setup and cheap housing.

- Rear mech - same one.

- Shifter - same one.

- Ferrules - none of the six ferrules involved (because I am using inline cable adjusters) have those rubber bung things inside.

I did notice though, my chainstay must be longer now because the chain seems a bit short on middle chainring and lowest sprocket, the swingarm should be vertical but it's probably 1 inch forwards, but I cannot imagine that would be causing the tension issues at the smallest 2 sprockets.

I am going to again screw in an old Maillard rear hub I found lying around and see what's what, but it's tough checking it by eye.

Wiggle used to sell something for aligning a hanger and it was only about £25, I would buy it in a heartbeat if they still sold them but they aren't available. Maybe Park Tools bought the company (half joking).

I am 95% sure it must be the hanger out of alignment.

rogerzilla cheers I will try to do it that way now I have got a hub on its own to screw in as opposed to having a 2KG wheel hanging off the hanger and trying to hold it up.

[Brucey]

maillard rear hubs with solid axles normally have 3/8" x 26tpi threads on them. Gear hanger threads are more usually M10 x 1mm, so a 3/8" thread can strip the gear hanger if you are not careful.

cheers

[Manc33]

maillard rear hubs with solid axles normally have 3/8" x 26tpi threads on them. Gear hanger threads are more usually M10 x 1mm, so a 3/8" thread can strip the gear hanger if you are not careful.

cheers

It's been in and out before I read your comment, it threaded in loosely.

After resting a spirit level on the flanges holding it there, while trying to balance the bike perfectly upright, the spirit level was telling me the same thing I initially suspected when eyeing up the old hub - it needed bending up a bit, so I did that, then checked again with the spirit level and it was level.

I put the rear mech back on and it's just shifting exactly the same.

I keep trying to think if it's not shifting on 12t-14t while wanting to shift from 32t-36t which way does that mean it's bent. None of it makes any sense. If the mech is bent outwards that angles the upper jockey wheel too far away from the 2nd to smallest sprocket but the same should apply up near the low sprockets, yet it's not doing that. So like someone said above it sounds like a twist, rather than a bend in or out.

If the hanger is twisted, I can see how the chain would catch on the low sprocket when it shouldn't, the cable tension wouldn't really come into it, except that it would need to be unusually loose, which it does need to be.

What's funny is when the bike is the right way up (I always setup gears with bikes upside down) then it alleviates the problem a bit - it manages the shift from smallest sprocket to the next smallest - so whichever way the frame is flexing slightly because the bike is the right way up is helping it.

I'll just take it to my LBS to get the alignment checked properly since it's free if I do (part of buying the bike).

Really though I think the best bet is to make one of these:

https://www.youtube.com/watch?v=sWdO4dnu18g

That guy is great.

I worked out making the tool will cost about £10 and that includes the drill bit.

I spent a while whittling it down to the cheapest stuff on *B*y but here's the item numbers if anyone wants to make this tool:

400801773676
£1.25 (M10 washers | pack of 15)

361079755654
£1.29 (M10 nuts | 1mm pitch | 2 pack)

151820070599
£2.00 (10mm HSS drill bit)

183297283819
£2.12 (M10 X 50mm bolt | 1mm pitch)

302206680000
£3.60 (Steel tubing | 19.01mm square x 500mm long x 1.5mm thick)

TOTAL: £10.26

The OD of the washers is 20mm so you'd have 0.5mm of washer sticking out but the 20mm tubing I could find was all more expensive and it doesn't matter if it's 1mm smaller.

If you have a drill bit and that tubing (like from an old office desk) you'd save half the cost of it. You don't need 15 washers but the cheapest I could find was a listing with 15 of them.

The hard part is drilling a hole exactly vertically, without one of those drill press things, hmmmmmm....

[NetworkMan]

Do you have another RD you could try? I wonder if you have a combination of long squishy housing and a worn RD. When you move from top sprocket (cable slack, spring pushing against limit screw) to next-to-top you are taking out play in the RD and taking out any initial squish in the cable housing.

[Chris Jeggo]

I have a new frame and a 12-36t cassette + wheel from a setup that was working perfectly.

When setting up the rear shifting, if I set it so it shifts from 12t-14t, then up at the 32t sprocket it's almost trying to shift to the 36t.

If I set it up so the it is not trying to shift to the 36t from the 32t sprocket, then I get only about 1/4 of a shifts worth of movement from the mech at the 12t-14t shift.

This is certainly puzzling.

To be precise, are you saying that when you adjust the cable so that the jockey sprocket is well aligned with the 14t sprocket after you have shifted from the 12t to the 14t sprocket, and then you shift down to the 32t sprocket, the result is that the jockey sprocket is not well aligned with the 32t sprocket, but has moved too far by maybe 1/4 of the cassette sprocket pitch? This test seems to me to be the best measure of how well indexing is working, being independent of limit screw settings and unlikely to be affected by cabling problems. In doing this test it is important to change down. Changing up is more likely to be affected by cabling problems because the RD return spring is weaker than a hand on a gear lever. It is also important to avoid overshifting in changing down to the 14t and 32t sprockets because overshifting likewise requires the RD return spring to come into action when the hand is removed from the shifter.

Here's what I can rule out:

- Cassette - same cassette and wheel is being used, the cassette was never taken off the wheel. It was working perfectly on the last bike with a conventional housing setup and cheap housing.

- Rear mech - same one.

- Shifter - same one.

On the face of it your problem cannot be due to shifter, RD or cassette because these have worked together satisfactorily previously, but on the other hand you have paid enough attention to hanger alignment already that it is difficult to see how any residual error can give rise to an error of at least 1mm in indexing. You also consider your cabling to be OK. So the mystery remains.

Shimano (and SRAM) 9-speed sprocket pitch is 4.34mm according to Sheldon Brown, so outer face to outer face distance between 14t and 32t sprockets is six times this - 26.0mm - or 34.7mm between 12t and 36t. (Similar calculations can be done for a 10-speed cassette - you don't say which you have.) These distances can be measured quite easily, and accurately enough (a quarter mm maybe), using a straight edge and a steel rule, so I suggest it's worth doing. You will either positively eliminate the cassette from consideration or discover something interesting.

I hope this helps.

[Manc33]

Cheers folks, after making my DIY rear mech hanger alignment tool, it was the mech hanger that was bent.

At the bottom, the bar of the tool was 75mm from the rim. 180 degrees around it was 95mm from the rim. I bent it and got both to 85mm. Then I checked the left and right, same again, towards the back of the bike the tool was 95mm from the rim, but towards the cranks it was 75mm from the rim, I tweaked the hanger tool and thus the hanger again to make both 85mm. Doing that again bent the top to bottom slightly out, so I again bent the hanger so they were both 85mm.

I put the chain back on and what a difference that has made!

Now I have no cable slack on the smallest sprocket and it's shifting well.

I don't know if +10mm/-10mm is a lot (on a 16" tool/bar) but it was ruining my shifting. Now the hanger is straight, I have never had shifting so good and I am starting to think every bike I have ever had possibly had a bent hanger (in some capacity) because of how well this shifts now it's put straight.

[Chris Jeggo]

Congratulations! And I've learnt something from your experience.

The radius of the rim will be somewhere in the region of 320mm (less than 16"), so a 10mm error turns out to be around 1.8 degrees, and that's in both the vertical and horizontal directions. In one of the 45 degree positions (half past one to half past seven, looking from the RHS) these cancel, but in the other 45 degree position (half past four to half past ten) they will combine (by Pythagoras) to give a total error of 2.5 degrees. So now we know you've got to get much closer than that.

Enjoy your sorted bike!

[NetworkMan]

Yes indeed, congrats!
Also congrats to the guy on Youtube for his good design of tool. By using the bolt and two flat washers he ensures that the bolt is perpendicular to the steel bar (subject to the thread on the bolt being cut right). That bit of design is crucial.

I get it now. The RD uses a slant parallelogram so if the plane of the hanger shifts towards a plane at right angles to the slant angle then it will overshift which is what must have been happening. Over a distance of 35 mm, an angular shift of 1 degree will cause an error of 35*sin 1 or 0.6 mm or 1.8 mm for 3 degrees so the error is in the right ballpark.

That convinces me to keep away from 10, 11 and 12 speed systems - just too sensitive! I'll stick with 7, 8 and 9 as at present!