So I've just gone from 8spd to 9spd, and had to use beefier shift housing because the skinny stuff that was fine for 8spd was too mushy to let the 9spd system shift accurately...

And it worked beautifully for a while, but now I'm getting intermittent auto-shifting and bad downshifts (to a smaller cog), despite the facts that: I'm using an older derailleur with a stronger torsion spring with the preload adjustment hacked for even more tension, and the derailleur is slop-free and aligned, and the housings are now bedded in, and I know what I'm doing when I adjust it.

So I'm looking at the total cable travel of 20mm or so and thinking that derailleur geometry may have been fine for 6 and 7spd, but 8spd was pushing it and 9spd is stupid. It's just crazy to pull on (or particularly, release) the cable by 2.2mm or whatever and cross your fingers that signal will make it +/-0.1mm to the end of it under all normal conditions. Hence the big-dollar aluminium or ceramic segmented housing you can get these days... and the growing size of derailleur loops on pro bikes.

I have a feeling Shimano have known it'd be a good thing to revise their cable pull for quite a while, but are just trying to push folks to electric... it sounds overly cynical, but I can't imagine backwards compatibility of derailleurs is considered so important we have to suffer unreliable shifting. As for Campy and SRAM, I'm not sure how much this relates to them, hence the poll.

I have an idea for a partial solution, but my first stab at a prototype isn't quite workable enough to test, and the refinement I need to get around it pretty much demands CAD and some time on a 3D printer... so I want to know if there's much demand for a solution to this issue.

So basicly ...
You, with your old "hacked" derailleur with extra torsion ... are having a difficult time to shift 9-speed systems ...
While everyone else in the world, including me, is shifting perfectly fine with all kinds of 9-speed or 10-speed systems ...

Am I getting this correctly?

My old 6400 derailleur is perfectly compatible with my 6500 levers and in good condition. In fact, it's better-qualified to perform well with the extra spring tension trading off teh ghey lighting shifter effort for greater precision.

Instead of just pooh-poohing, how about responding to my poll, or at least tell me which system you're not having any trouble with?

Also you might care to inform Nokon and Vertebrae they have no business model.

shimano 9-sp is highly regarded as one of the most reliable systems out there. cyclocross guys love it because of how durable it is. i'm a diehard campy guy, and have three bike with 11-speed, which shift perfectly, i mean perfectly, with thousands of miles on each now. i also bought a used CX bike that came stock with ultegra-9. stuff is beat to hell but works flawlessly. i've done multiple dirt road rides, complete with rocks, mud, rain and even snow, and that 9-sp stuff is dead on, no adjustments necessary.

just sayin...

Hm... quite a few 'not happy' respondents already though, notably for all systems. And the cheap housing that was fine for my 6400 setup was pure fail on 6500.

The brutal, warranty-voiding hack:



Replaced the adjustment cam with a bit of spring from a cheap allen key set.

I have no problems with mine

Well, since it's limited to ten options and it matters less which system you're on if you're happy, I'm hard pressed to imagine a better way of breaking it up.

Actually come to think of it, you could totally flip that and say that given an even spread you'll see much longer bars in the 'happy' category.

I've had Shimano 9sp (6500 and 7700) and 10sp (7800), and Sram 10sp all set up and shift properly on a number of bikes. Shimano 10sp (7800) is the best of the lot but they all work. I use Shimano cables and housing except on the Sram where I use Sram (Gore) housing and cables.

I think you're headed in the wrong direction if you want reliable shifting with minimal effort. If you want to screw around with your own ideas, that's different.

Nope.

Try a stock 9 or 10 speed Shimano RD and report back

I assume that you're working on the flawed theory that too little cable pull results in problems. FWIW, Shimano 8 and Campy 10 have the same 2.8mm average cable pull per shift. Campy 11 averages 2.6mm. Note that I say average, because both brands use nonuniform pulls. For example, Campy 10shifter pull 2.5mm 5 times, 3mm twice and 3.5mm twice. Only SRAM shifters pull equal amounts of cable for each shift, with the 10 speed road shifters pulling 3mm for each shift. Shimano 10 pulls the smallest amount of cable with an average of 2.3mm.

Cable housing has not changed much, for a very long time. Any decent brand uses a low friction plastic liner and the liner is surrounded by wires that run parallel to the cable, in order to eliminate compression. Stock cables from all brands usually work quite well.

I've used Shimano 4mm housing with Campy 11 drivetrains and had no problems, but I always use genuine Campy cables. I find that the shift cables housing needs replacement more often than brake cables. The cheap way to refresh your shifting is new cables, with Shimano housing that can be had very cheaply, if bought in bulk rolls, or by the foot, from those same bulk rolls.

Buy a new rear derailluer and be done with this.

I'm no expert, but I'd have to agree with this, and IIRC, with indexed systems, shift housing is shift housing. I'm running on a 9spd right now and I love it because of its dependability and ease-of-use and installation.

Come on, Kimmo, I know you love experimenting around with this stuff (thank you for the STI brifter guide, BTW), but seriously, old 9spd RDs aren't THAT expensive. Why hack around with an 8spd RD to perform the same tasks as a 9? What exactly are you trying to prove?

Exactly.
Even the poll is agreeing that just about everyone is happy

Strange, all the unhappy ones responded straight away. Then not a peep more while all the happy folks came along...

Look, you guys - I could use a 6spd derailleur on a 10spd bike, or vice versa. Shimano has not changed the geometry (8spd DA and Dynasis aside). All they've done is made the return springs weaker for marketing reasons, which I'm having a hard time imagining helping.

But hey, just to humour you and rule it out, I'll have a dig around today at my co-op for a 9spd RD.

Oh, I can overhaul STIs but I can't manage a bowden cable? This is the first time in 25+ years of wrenching I haven't been able to sort such an issue.

Evidence suggests...

Look, most people with 9+ speed setups have their rear indexing set up perfectly or very close to it. You don't. If it's not an issue with the modifications to your RD, which might not be as straightforward as you think, then your mechanic skills are inadequate to set up 9+ speed systems. There really aren't many options here.

The first indexing shifter/derailleur installation I ever made happened to be a Shimano 9 speed setup. It's still shifting well today with the original cable and housing. And I don't claim to be a great mechanic.

Is your premise really that since your first attempt at 9 speed doesn't work well then the whole bike world currently has bad shifting? From mountain bikers to pro riders? And has since what, 1997?

My premise is that I can't see the flaw in this theory. What if you reduced cable pull to 1mm per shift? Think that'd work? Okay, what if it was 5mm? 'Compressionless' housing isn't truly compressionless, hence the fact you don't use it for brakes, and the existence of segmented housing.

This compression, along with cable friction, constitutes noise in the signal of the shift, does it not? And it remains relatively constant no matter what the cable pull is, amirite?

So a larger signal would have a higher SNR. Where can I be wrong with this?

So most people don't have an issue with this. I never had the slightest issue with 7spd, myself. But I wasn't too happy about the SNR on 8spd, since the adjustment can be a bit fiddly once normal housing degradation sets in...

Folks can cast as many aspersions on my skills as they like, but in addition to my practical experience, I've spent pretty much my whole time on this forum in the Mechanics section chatting with far more experienced guys than myself, not to mention the fact IQ tests put me in the 98th percentile, so it's a bit of a stretch to suggest I'm missing something basic.

The stuff I'm using isn't new. But the DA chain and Ultegra cassette I have show minimal wear, and the chain has no twisted links. This gear would certainly pass muster on an 8spd system.

But the problem I seem to be running into is that the finer tolerances involved here mean a close visual inspection of everything just doesn't tell you enough...

Sure, new everything would certainly fix it. But I'm saying that if I'm using new cables and everything else appears to be up to snuff and it still doesn't work reliably, what we have here is a pretty damn finnicky system.

Although I have a feeling my issue might be something a bit out of left field, since the malfunction I'm getting seems a little different than usual: rather than it being consistently crap, it's mostly sweet, but with sporadic moments of extreme woefulness. So there's that... I'll throw it up on the stand at my co-op today and see if some of the old hands can figure it out.

When I was at a loss getting my 10sp Shimano RD to shift nicely (7800 shifter, 6600 RD, stock Shimano cables), I wondered if it was the derailleur hanger.

Rode it to a shop, had the guy check it, and yup, that's what it was. Fixed it, rode home and it's been sweet ever since (more than a year now).

Nobody's said to check the hanger yet, so just throwing that in here.

You should get the compression less housing a bundle of wires on end

Brake housing is a coil of square wire.





my indexing is in the Hub.. Rohloff.. 2 pull-pull cables..

This proves again that IQ means little ... and I too know what I'm talking about, mind you.

Fact: thousands of road bike racers and enthousiasts have been racing and more casually riding 9-speed and 10-speed sytems of all manufacturers for years ... that's millions if not billions of shifts with an extremely small error count and I'm guessing 99% of those errors were user errors either by not shifting correctly or not setting it up correctly.

Fact: many people on this forum, including me, have reported completely flawless shifting with both 9-speed and 10-speed systems ... reported over a period of multiple years. For myself ... I have been shifting a mix of Shimano Ultegra 6600 & 6700 for 2 years now and have yet to encounter even one single little problem ... there have been none whatsoever.

So ... you, on the other hand, seem to have a little issue ... and you point the finger at the system itself.
Don't you agree that, seen from our perspective of people having no issues at all, that that sort of behaviour seems a little silly?

Well ... you are wrong in the sense that with indexed shifting the system has, so to speak, moved from analog to digital.
In the digital world the noise ratio isn't that important and we start speaking of "noise margin" instead of ratio because in the digital world any noise improvement above the noise margin does not contribute to a cleaner signal anymore.
As long as the losses in the system are lower than what is calculated to be a workable noise margin ... it'll be fine ... and it is fine ... for millions of people except you apparently.

Not even a little bit. Shift housing is compressionless, brake housing is not - there is nothing compressionless about brake housing, and it is an easily demonstrated empirical fact that brake housing does compress much more than compressionless housing (ie shift housing). The reason that one cannot use compressionless housing for brakes has to do with the forces involved - shifting generates far far less force than braking - compression is not a worry for braking (just squeeze harder), but the force involved in braking can easily break compressionless housing.

Brake housing contains metal cable tightly wound in a coil pattern (think of a slinky), and the effective length of this housing changes when it is bent - try to use it with an indexed system, and you would see ghost shifts when you turned the handlebars. Compressionless/ shift/ indexed housing is made of metal cables oriented longitudinally - try to compress metal cable. You can flex it all you want, it keeps the same effective length.

Hilarious. Reading this, I have trouble believing that this is not a joke. You cannot set up indexing properly, therefore the whole system must be deeply flawed. Millions of people have been riding these systems without problems for over a decade, but no, if it does not work for you, it could not possibly work for anyone else. And I am sure you can find lots of people in the 99th percentile who cannot set up an indexed shifting system, or build a true and balanced wheel, or change the oil in a car, or do any number of moderately skilled tasks. Thinking that you are omnipotent (ie able to do anything) because you scored high on an IQ test demonstrates a total lack of sense.

Often overlooked is the fact that bike parts wear out. I've checked the main pivot on a Campy 10 RD with only a few years of use and found a huge increase in the axial play on the main pivot. When new, a Campy RD may have .2mm of axial play, but with a few years of use, the play can increase to .8mm, which is really sloppy. I've fixed that by installing a shim washer to reduce the play down to a tighter than new .05mm.

Shifters wear out too. If you took the time to rig up a precision setup for measuring cable pull, you'd find old shifters don't pull exactly the same amount as new ones and may not have good repeatability.

Of course, it's obvious that if a certain amount of error is in a shifting system, reducing the amount of cable pull would make that error a larger percentage. In that case, Shimano is the biggest offender, and might have a 23% greater RD positioning error than SRAM or 18% greater than Campy. To know whether there is really a problem, you'd have to compare all three, using highly accurate equipment. If the error is very small, like 1-2%, then a 23% greater error would still be small, like 1.23-2.46%.

I voted no problems. Have 105 10 speed. However, it's a road bike. Used in fine weather, on roads. In spite of that, it takes new cables, housing, setting it up carefully.

My 8 speed Acera 29er is much more bombproof. Sand, water, salt, old cables, housings - it works much more crispy.

So yes, I'm happy with my 10 speed. Would I rather have a new 105 in 8 speed variant? Definitely. I ride with triple up front, so more than enough gear ratios with a tight 8 speed cassette would be perfect. Unfortunately, only 2300 (or maybe Sora) can now be bought in 8 speed. And they have those thumb shifters. :/