| ??? 02/19/05 03:39 Read: times Msg Score: +1 +1 Good Answer/Helpful |
#87937 - Some averaging is needed Responding to: ???'s previous message |
Prahlad said:
If the caliberation firing angles are fixed by the controller and are not modifiable by user does the risk you mentioned do exist? If my controller always fire three positions maximum, minimum and middle and measure currents to calculated the polynomial co-efficient? There must be a methode that eliminates big influence of (hopefully seldom) calibration errors. So, a statistical approach would be helpful, where the correction data is gained by some averaging. Would do you think about a fix table and a polynom only for the correction data? This would eliminate drastical misfits. I think the user should not be able to make any corrections by hand. This intelligent calibrating should be done hiddenly. In any case, if you plan polynom approximation this 'maximum', 'middle' and 'minimum' current calibration points should be equally spaced. But you can also make a 'statistical' parabola approximation, where much more than only three points are taken into consideration. This will minimize the influence of (hopefully seldom) deserters and provide a best fit to the whole curve. Kai |
