MX, thanks for coming to the rescue. I had read p 47 and my earlier post was an apparently faulty attempt to understand what was being described. I hope you'll be willing to shed a little more light on the subject.
Part of the reason I read it the way I did was this bit:
The cross shaft moves within a torque
tube which is attached to the cross shaft
near the bellows lever. The tube is
secured in the body casting at the
opposite end by means of an
adjustment bushing.
That sounds to me like one end of the torque tube is fixed to the cross shaft and the other end is fixed to the case (at least that is what I figured "secured in the body casting" meant). If one end twists with the cross shaft but the other end doesn't move, that sounds a lot like a torsion spring.
You say the torque tube "simply serves to transmit the forces", but transmit them where? To the body casting? The diagram I linked to shows the metering valve being actuated by the far end of the cross shaft, not by anything the torque tube drives, so I'm scratching my head over this too.
The reason I was looking to find a spring somewhere was that without one I couldn't see how the bellows/metering valve would avoid travelling all the way to a stop whenever the Px/Py ratio was off. I think there are two ways to prevent that result in a control system such as this:
1. Rate limit the control (how quickly the metering valve moves) so that the system under control can respond before it reaches the stop. In this application I think that rate would have to be pretty slow because the response times (fuel flow affects N1, then N1 adjusts bleeds that modify Px/Py) are slow, but I didn't see anything that would accomplish this.
2. Add a spring that supplies a proportional counterforce to the movement. Now a small pressure imbalance can cause a small valve motion, and a larger imbalance a larger motion.
The second seemed a more likely candidate, but if the torque tube doesn't do the job is there another answer?