`pid` with zero proportional gain

[mzaffalon]

pid fails when the proportional gain is 0. This is because the coefficients are converted to the standard form which assumes a non-zero proportional gain.

julia> pid(0, 1; form=:parallel) # expected 1/s
TransferFunction{Continuous, ControlSystemsBase.SisoRational{Float64}}
NaNs
----
1.0

[albheim]

Hmm, that's not good.

I think using the standard form as the "default" that was used in conversion was something I just picked, didn't really think it through. Probably would have been better to have the parallel form for this since the parameters are not tied together and thus won't (?) lead to these problems. On the whole that code is not too pretty, and maybe adding another special case would be the simplest for now.

[baggepinnen]

Yeah, using parallel form by default is probably better in general

[albheim]

So maybe just replace the convert_pidparams_(to/from)_standard with convert_pidparams_(to/from)_parallel and change surrounding code accordingly. Wasn't too many places that called them directly so should be relatively easy I think.

Anyone want to have a look? Otherwise I could probably have a go during the week.

[mzaffalon]

Anyone want to have a look? Otherwise I could probably have a go during the week.

Let me have a go.