Sanity Check on PID Controllers

[QMESAR]

Hi All.

Please help me out here I am loosing all my faith in my maths ability

I have a SW that I easily can create Graphical user interfaces very similar to LABVIEW,(Graphical programming ) this SW has a PID component (controller block).
my problem is when I use a positive KI coeficient then the controller output is positive and when I use a negative KI then the controller output is negative.This to me is incorrect looking that the method of calculating the controller output,
a negative KI should help to reduce the intergal value and even when the complete term is negative the output should not change to negative as it is first term + Second term for example 2 + -1 = 1 not as in this case from the controller -1

The help file of the component shows this

1.JPG[/attachment Please help to clear my mind that I am not having error thinking in my head ,I fail to see how the second term can change the controller output to a negative when the Error is always positive ,and even when the error is negative a negative KI should not change sign of the output.This is also happening when the error is negative the output changes to a Positive with a negative KI . :oops: I am lost here please put some sanity in this for me [attachment=0]2.JPG

[Benj]

Hi Peter,

This is my simple PID formula I worked out at Uni and have implemented in a few places (including the Flowcode DSP Control component), seems to work well and be correct.

The equations they gave to us were horrible but boiling it down to the basics it's not actually that bad.

Error = (Setpoint - Feedback)
PrevError = Error from last cycle
PrevPrevError = Error from two cycles ago

Control Signal = P * (Error - Prev_Error + (Error / I) + D * (Error - (2 * Prev_Error) + Prev_Prev_Error))

Hopefully this should help.

[QMESAR]

Ben
Thank you very much this is very helpful