FurryOnTheInside
In Memoriam
I've used both in industry.
Spring idlers are for single direction steady loads. No reversing, power in one direction only, no shock loading.
The best were rubber cored tensioners like pictured below.
The rubber adds some damping to the arm. Spring tensioners are typically quieter.
On a reversing, shock loaded, variable load system (like most bikes are) a fixed tensioner would be a better choice "by the book".
A light/soft spring tensioner with travel limited to minimum would be an ideal solution, I agree.
Have not seen one yet. Maybe something like this could be made into ideal?
Steve
My thoughts are that a spring being constantly compressed wastes (some) power, but a fixed idler makes no allowance for the eccentric HT drive sprocket and/or wheel sprocket. That is why I thought a light spring is best. When the chain is at it's tightest spot in the cycle (once every 220 revolutions of the rear wheel with 44t rear sprocket) and when it's function is reversed, the soft spring would be compressed as far as it's (adjustable screw?) limiter, but when the chain is slackest it should be only compressed by the weight of the returning chain, so applying almost no load to the chain.
The picture appears to show an adjustable preload, great for switching from bumpy roads to smooth racetrack; but no limit to the spring's movement other than it's own. The part it would like to see when I look at the picture you posted is a bolt poking down next to or inside the spring.
I thought maybe there is something for "fixies" and single speed cycles built on bike frames with dropouts (lacking the proper rear facing forked wheel mount that you generally see on BMX bikes) but the idlers that I have seen so far appear to favour quick wheel installation over maximum efficiency so they are just sprung with lots of preload and really no limit until the chain pulls tight (so it does allow the top to go slack during pedal braking, eek!) just like an ordinary derailleur. Which does lead me to question the actual amount of power loss and whether it really is worth worrying about, whether it is fixed by the spring strength and preload or varies with the power put in.. and I really dont know.