I've been fussing over this constant-torque clutch solution for my fuse spinning machine for a few weeks.
The sprag clutch on the takeup reel solved a lot of problems -- primarily, it removed the "letoff" that was occurring when the mechanical friction clutch slipped.
However, it still didn't solve the problem of the clutch sticking and slipping, thus varying tension on the roving as it passed through the machine. It was a minor thing, but resulted in cosmetic problems where the final over-wrap on the fuse was not uniform.
Mag-fluid clutches were too expensive, and multi-plate clutches still had the stiction problem. I even enquired of one multi-plate outfit that claimed "static friction is less than dynamic friction", which not only defies the rules of physics, but when they were challenged to produce torque curves, the could (would) not.
So, what to do? Hydraulics. Constant pressure on a pump (disregarding bypass losses) equals constant torque.
So I bought a re-furbished power steering pump, and set to making it into a constant pressure pump -- thus constant torque.
I modified the pump by locking the internal relief valve. It could have been made to relieve at the correct pressure, but had two drawbacks -- it had a large throw, and caused oil to recirculate IN the pump, rather than drawing from the reservoir during bypass. I also modified the pump by adding a port to the top of the relief valve chamber to add my own custom poppet-style relief valve.
This was a bit tricky, not for the easy machining of the cast iron, but for keeping chips out of the work. To do that, I filled the chamber with beeswax, draped a cloth over the entire pump, with a hole through which to do the boring and tapping, then surrounded the work area with lots of little rare-earth magnets. That worked. Kept ever trace of swarf out of the body of the pump.
I didn't know the displacement of the pump, and couldn't find it in any specs, so the selection of the relief valve spring had to be empirical.
I finally got the thing producing a smooth, jitter-free torque of 15 ft.lb. by cutting a miniature die spring down 1/2 coil at a time until the torque fit the spec.
Now the only things left to do are to build a driving boss for the pump housing, and a takeoff coupling for the shaft -- and we'll see how it works on the fuse machine.
It's been a three week "off-the-clock" learning exercise for me, but worth the time. I can see all sorts of uses for such a mechanism. The torque varies only about 15% from 2rpm to 7rpm, and that's well within my needs.
LLoyd