Positive Position Feedback

I would guess that the feedback pushes in the direction of deflection with a little less force than the vibrations. The effect is to slow the vibration down as it moves toward the centerline. Adding a little integral action would allow the force to be applied only as the system returns to the centerline.

If the system had positive feedback based on velocity, then it would most likely be unstable.

Michael

I would suggest that you do an experiment with a second order model with a set of complex conjugated poles, e.g. G(s) = 1 / (s^2 + 2*zeta*w0 + w0^2)

For this system you can calculate how the poles move for respectively negative and positive feedback. You will observe that it is better to use positive feedback when comparing to negative feedback if you want to dampen the oscillations in the system (getting the poles closer to the real axis). You can also see that the poles remain in the left half plane as long as the gain is not too large.

To get a physical insight you do the calculations for e.g. a mass- spring-damper system. The oscillations will be sinusoidal movements and you can see the a reason for the improvement by the relation between position and acceleration.

/Kasper

...

I have no idea about what the OP encountered, but I wouldn't dismiss positive feedback out of hand.

Positive feedback with a loop gain less than one can be useful. For one example, positive current feedback lowers a device's output impedance, even making it negative.

Compounded generators can have a voltage output that rises with load, thereby largely compensating for the IR drop in the transmission line. Instability occurs only when the total resistance -- generator, transmission line, and load -- is negative. Some audio amplifiers have used positive current feedback to offset the effects of voice-coil resistance, and even a tiny bit of it does more than any Monster cable.

Jerry

Jerry Avins wrote in news:rbSdnSfZqqTh5SXbnZ2dnUVZ snipped-for-privacy@rcn.net:

One situation where positve feedback is regularly use is as a "negative capacitance" amplifier to null out a high capacitance glass microelectrode.

The guard terminal on an instrumentation amplifier used the same principle.

Jerry

Jerry Avins wrote in news:7tadnZ-BlqacPiXbnZ2dnUVZ snipped-for-privacy@rcn.net:

Do you have any reference for that, Jerry? I'm gearing up to teach an instrumentation course, and that's pretty new to me. I'm very aware of driven grounds, but I'm pretty sure that doesn't have a negative input impedence. The glass microelectrode is a pretty specific problem. You have a very thin-walled glass tube, filled with conductive medium, trying to record in body tissue, also filled with conductive medium-- a pretty big capacitor that needs to be nulled out to achieve any sort of realistic bandwidth.

I hope I didn't mislead you, Scott.

A considerable capacitance is subtracted, just not enough to make the sum negative. Consider the simplest case, not an instrumentation amplifier, but an emitter follower. We use double shielded cable, with the outer shield grounded. For our length, the capacitance between conductor and inner shield is 100 pF. The "gain" of the follower is .99. (for an IA, the gain to the guard terminal can be .999.) Connecting the inner shield to the emitter makes the effective capacitance C*(1-gain), in this case 100pF*.01 = very small. You can think of it as connecting an effective capacitance of -99pF to the shield end. With a modest number of extra components, one could make that -110pF and neutralize some of the capacitance of the source itself.

Jerry