Redesign of the damper and damper torque sensor

4.2.5 Redesign of the damper and damper torque sensor

4.2.1 developed a model of a mechanical system (the damper cup) and showed how the torque we wish to measure is generated. 4.2.2 showed how our sensor generated a signal proportional to that torque. 4.2.3 showed with the same model as 4.2.1 that in addition to the torque, the signal which we want to measure, there are radial forces generated, which represent noise we don't want to measure. Finally, 4.2.4 showed how the sensor generated a noise signal which was proportional to the radial forces in the damper, resulting in an unacceptable signal/noise ratio. There are two ways of improving a signal to noise ratio in this situation : 1) Reduce the magnitude of the noise, and 2) Reduce our sensitivity to the noise.

The first effect is accomplished by increasing the gap size in the damper, and then increasing the viscosity of the damper fluid to maintain the damping torque. We increased the gap size by a factor of three, which, according to our model, reduced the radial forces by a factor of nine. While helpful, this improvement is not enough - we still have a signal to noise ratio of approximately 40. By reducing the circuit's sensitivity to radial loads, we can reduce the noise to less than a percent of the signal we are trying to measure.

This reduction of sensitivity to radial forces is accomplished by changing the configuration of the strain gauges. Our circuit was sensitive to radial loads because when was , the radial load behaved like a damping torque as far as the strain gauges were concerned. Figure 29 shows how we rearranged the strain gauges so that the sensor is no longer sensitive to radial loads.

Figure 29. Cross-section of the damper cup and flexure elements, with strain gauges on two flexures.

For the case of measuring the damping torque, we see that the bridge will measure the signal, as and will experience the same strain and and experience the negative of the strain experienced at and . The result is the same as that found in 4.2.2. For the case of the radial load, we see that things are quite different than they were before. Now, regardless of , and , so that no output signal is measured. The benefit of this measure is difficult to quantify, as Wheatstone Bridge theory predicts that the sensor will now be immune to radial forces. However, imperfections in the strain gauges will result in some of the signal getting through. The improvement should be at least an order of magnitude, though, since the gauge resistances within a matched set are specified to be within 2% of each other.

This section has dealt with the practical difficulties encountered in emulating low impedances with a damped haptic display. Through the use of models, we've shown that velocity-based damping cancellation is not an attractive option, as it relies on a model of the damper to ensure system performance. We've shown that torque-based damping cancellation, while presenting challenges in terms of electro-mechanical design, is a much better approach to achieving a wide range of impedances in a damped haptic display.

-->