Tom, This sounds more like a Sunday job to sit down and characterize what is really going on with the control curve. Fixed input, step the control, measure the output and go from there. Neil -- http://www.njohnson.co.uk