How much bend can I have on a gear cable before shifting performance is affected?

Question

I have a Sunrace m2t derailleur on an old Batavus holiday dutch bike and I am running the gear cable from the derailleur down the chain stay under the bb and up the down tube. The bend on the gear cable from the derailleur isn't too extreme, but it made me think about if there is a limit before shifting or long term reliability is affected.

I have added images below showing the current setup (UPDATE: and after). The issue is, because it is a step through bike, the alternative to running the cable along the chainstay is going up the seat stay and then down the seat tube, then back along the down tube.

Interestingly looking at http://www.killasgarage.bike/uncategorized/cable-routing-experimental-results/ perhaps shortening or repositioning the cable to minimum the distance out and then in might help, so I've done that and learnt something new (added photo)!

Answer 1

There are two different contexts this kind of question can come up in:

• When one is deciding between completely different routing schemes, i.e. a bike that gives you the ability to put the RD cable along either the top tube or the down tube.
• When all you're deciding on is how long to make a specific piece of housing.

For the latter situation, it's easy. You always choose the shortest housing length that allows the line of the housing to be straight into the housing stop, because the bends being overly sharp in those areas (from a too-short housing) cause severe cable friction that will readily affect performance. For housing pieces that move with the bars/suspension, they also have to be long enough to avoid kinking at the extremes. Once you've reached that point in length, any extra length causes extra friction at a very marginal rate. There's no reason to have it, but there will not be a singular threshold where before the performance was fine and now it's ruined. In other words, if you were to graph the total friction in the system, the too-short lengths result in highly elevated levels, and the too-long levels steadily taper up at a low rate.

If you are choosing between two different routing options completely, there are usually functional concerns that have to be weighed against performance impacts from friction, but as far as the friction part is concerned you will typically find the better performer is the one that has the most gradual bends in the housing, and avoids sharp bends as much as possible.

Answer 2

The rear loop of the cable housing is the worst. The cable runs rearwards but the rear derailleur wants it to run forwards so you need a 180 degree tight loop at the rear. I can't imagine that for example the bend under the bottom bracket would be at all as bad as the rear loop.

It's true that the rear loop puts a limit to shifting performance. Many recent rear derailleurs are made so that the rear loop wouldn't be 180 degrees but somewhat less. This improves shifting performance.

Make the rear loop as long as you reasonably can to reduce the curvature (edit: some experiments in a comment negate this claim). Also use ferrules with O ring seals to prevent dirt from getting inside the loop.