Nonlinear PWM

eslavko wrote:

Now I hit with some other problem?
How to connect signal??? The bias and FF0 are RW!
'spindle-lps' can not add I/O pin 'pid.0.FF0', it already has OUT pin 'lowpass.0.out'

You don't normally "net" FF0. It is a gain (like PGain) and so you would normally just setp the value to 1 (as you already have good open loop control, that will work fine).

If you _did_ want to make it variable, then the lincurve component has an RW pin specifically to be connected to the PID gains.

(I may have been confused and misleading in a previous message)

eslavko wrote:

But if I speed it down I got unwanted behaviour. As control signal goes down more quickly as router slows down the I error goes high. When spindle decelerate to commanded value then due I term goes far under commanded speed and then return to correct one.

We call this "I-term windup" in the trade.

You probably need to lower the I-Gain until the problem goes away, then see how much of the speed tracking you can get back with P-gain.

If that doesn't work then you can make the I-Gain come out of a seconf "lincurve", one that takes PID error as X and outputs I-Gain as Y, and which has a lower I-Gain when the error is positive than when it is negative.

There is a PID controller that I look after at work where the P-Gain is the product of 2 16x16 maps, the D term is the same, and the I-Gain is the product of 3 16x16 maps.
Then the precontrol (like your FF0) is another 4 16x16 maps. I have about 3000 numbers to adjust to get it all to work right.