At 06:03 PM 11/16/99 -0500, Wagner wrote: >An oven, or whatever is under temperature control, has "own life" and >each one reacts differently for different temperatures. This is why, >when installing a temperature control unit, you should "calibrate" it. >The calibration process should understand *how* the oven works, reacts >to temperature increase, and very important as the oven cools down. > >If you only supply power to the heater when the temperature is below the >set point, and removes power when the temp is above it, you will have a >temp oscillation inside the oven, all the time, nothing will be stable. >This happen because it takes time to heat or cools the oven. I'm not sure I buy this argument. The issue is not whether a full temperature shift occurs instantly, but whether a change in temperature at all occurs approximately instantly. In order to use proportional control, don't you just need to be able to sense a reasonably immediate change in temperature due to heating? Assume the oven is at the exact temperature. It cools down 0.01 C. The sensor picks up the drop and turns on the heater until it senses +0.00 C error. The heater overshoots to +0.03 C. The oven droops in temperature until -0.01 C again. The effect of the integral in the time lag is to give a 'dither' in temperature that's 0.04 C large, with an overshoot offset of +0.02 C on the average. The offset can be calibrated out, and the dither is acceptable so long as it's small. So the whole issue is the degree of overshoot and drift present in the controller. If it's within requirements, it's ok. A lot of water bath controllers in labs work on this principle. If you need to fix the overshoot and lag, you just need to design an integral controller which has compensation for the physics involved. A simply method would be to simply damp the correction for the temperature drop. If you're using a PIC, you could make this adaptive ('fuzzy?') by dropping the damping when you consistently undershoot the correction, and increasing the damping when you overshoot. Asymptotically you'll have optimal control, so long as your algorithm is stable. ================================================================ Robert A. LaBudde, PhD, PAS, Dpl. ACAFS e-mail: ral@lcfltd.com Least Cost Formulations, Ltd. URL: http://lcfltd.com/ 824 Timberlake Drive Tel: 757-467-0954 Virginia Beach, VA 23464-3239 Fax: 757-467-2947 "Vere scire est per causae scire" ================================================================