My derailleur on a LOOK 586 hangs crooked!

Preamble: about this frame’s poor design
The dropout is excessively fragile on this design (= the tip of the rear triangle that takes the alloy hanger). Not only is it very thin, but it’s also drilled twice for those little hex screws, holes that only weaken it.
Background
When I built up my first 586 SL frame, early in 2011, the dropout literally crumbled while I was tightening the derailleur mounting bolt . . . At the time I thought it was my mistake, such as using a faulty torque wrench that didn’t click at the right torque. Or that I’d fallen for the classic beginner’s mistake of screwing the derailleur in while the B-tension adjuster isn’t at the right angle, touches the dropout, and ends up breaking it when tightened. But no, it really was a finishing defect from the moulding in Tunisia . . .
After quite some effort (I was abroad at the time, not easy to get a new frame quickly), LOOK’s warranty came through, and they sent me a new frame. This frame had another defect: the bore diameters of the cable guides were too small. No outer housing would go through. It wasn’t excess paint, but really the design diameters . . . Anyway.
The warranty kicked in again, the weeks went by. A new frame was sent free of charge. But another disappointment: they had pre-cut the seat tube to what they thought was “my” height, actually more than 30 mm short. So impossible to set up for my size.
Getting to the point
A fourth frame, a 2012 model this time, was then sent to me, still under warranty. Bad things come in series: yet another defect: the plane of the dropout wasn’t parallel to the plane of the frame. The whole derailleur, and so the cage in particular, leaned inwards towards the bike. As a result, the jockey wheels and the cassette caught the chain links at the edge, as circled in red in the photo. Shifting was poor, the chain jumped off the lower jockey wheel at the slightest bump, pothole, rail, etc . . . or even at the slightest slightly sharp shift. Also, the chain jammed completely in the cage on a quick big-to-small chainring shift. In short, a miserable life on the old bike. But this time, though, there was no way I could spend hours again stripping the groupset and housings, sending the frame back, and waiting for a hypothetical new frame, because I was on the road, riding Lyon to Nice via Savoie, in 5 stages.
Still, over the phone, LOOK’s customer service told me it was possible for the dropout not to be parallel, while noting they had “never heard of such a case”. Today, that last statement strikes me as kicking the ball into touch.
Anyway, I wanted other opinions than my own, to be sure it really was the dropout that was crooked, and not the alloy hanger, or even the rear derailleur (SRAM Force). That’s when I learnt that your average bike mechanic knows far less than I do about analysing a practical problem of geometric origin. That’s where basic theoretical knowledge can’t be skipped. I was very disappointed by their mathematical reasoning, or lack of it. I asked three people, with the bike or photos to back it up, but nobody questioned the parallelism of the planes. One suspected the parallelism of the derailleur, another the cage length, and a third the cable tension. In the end, since it was a purely geometric problem, it was me, the maths guy, who had the right answer. So, like a blogger who wrote on velo101.com, sanding was needed . . . Except there’s a dilemma: sand the carbon dropout, or sand the alloy hanger?
The fix
Inspecting the alloy hanger from every angle showed me it was flat. So in theory it was the dropout (i.e. the frame, really) that needed sanding . . . I wasn’t keen on that. So I preferred to sand the part that had no defect, but was easy to replace and cheap: the alloy hanger. So I filed it, with the help of a friendly mechanic at Decathlon in Gap. I filed it at an angle so as to make up for the lean. An hour of work and testing, because if I filed too far, that was it, no way back, and I was on the day of the fourth stage.
Obviously, you need to file the face of the hanger that actually sits against the carbon hanger seat (= the dropout), so that once bolted on, the apparent plane of the derailleur is parallel to the plane of the bike (= the frame plane). I know there’s a sophisticated tool, a gauge, to check the verticality and straightness of the plane, but in my case I judged that if, to the naked but critical eye, the cage was vertical and parallel to the frame, the result should be satisfactory. You need to know that, due to the pantograph effect, the cage magnifies the defect. That is, a 5° lean is visible to the naked eye on a cage even if it’s debatable on the alloy hanger, simply because 11 cm of cage produces a gap visible to the naked eye at its end, hardly visible over the 25 mm of the hanger.
I must have filed well . . . because in the end I made up exactly for the lean, and suddenly, as if by magic, the cage “came back into the plane”. After a quick readjustment of my gears, everything shifted perfectly. No more clicking, jamming, dropped chain, and the whole series of related bugs.
Several years later, I ran into a derailleur-plane parallelism problem again on another 586 frame, but this time I clear the frame, because it happened when fitting a new Shimano Ultegra derailleur, while the derailleurs previously fitted to this frame showed no fault. However, I solved this one-off problem with that derailleur by applying a patch, described below
Another way to fix the parallelism

See the picture alongside to see how I cut a standard washer into a smaller one that fits the hanger’s hole.
Conclusion
Once I reached my destination in Nice, I did the following test: I took the spare hanger supplied with the accessory kit, fitted it, and, no surprise! The derailleur was crooked again. So it really is the dropout that, to this day, is crooked.
I have another 586 frame, a 2011 one this time, bought second-hand, and this one has no parallelism or dropout fragility problem at all. However, in a few places the clear coat seems to bubble, but only over a square millimetre or two.
On the other two LOOK frames in my collection, an old 2003 KX and a mid-2000s KX Light, no problem at all; that was the good old days when the derailleur mounting pivot was taken directly into a metal-reinforced dropout near the wheel axle. I much prefer that design, even if it weighs a few dozen grams more.
If one day my hanger breaks in a crash, it’ll be back to the workbench to sand the new hanger to the same geometry.


