Hi Thomas,
You are right about the LabVIEW 7.0 optimization routines, I don't
think they are not suitable for your application. The formula
string interface definitely limits the objective functions that may be
expressed for minimization. I see a few options for your problem.
1. If you are interested in just equiripple FIR filter design, LabVIEW
7.0 does ship with a VI called Parks-McClellan.vi. Unfortunately
this VI only allows the specification of a magnitude response.
The phase response will be linear (constant group delay).
2. You could consider this to be a curve fitting kind of problem,
and use the Nonlinear Lev-Mar Fit.vi. This will give you a
least-squares approximation of the impulse response, but the quality of
the solution will be greatly dependent on the initial guess you
supply. It will be somewhat problematic to always get a stable
IIR filter using this method as constraints or bounds are not supported.
3. The Current Digital Filter Design toolkit allows linear and
non-linear phase FIR and IIR filter design. This would require
you to transform your impulse response to the frequency domain, but at
least the design VIs allow the specification of a complex response
(magnitude and phase), and offer an equiripple or Pth norm
criteria. This is probably your best option in terms of options
and amount of development effort on your part.
4. LabVIEW 8.0 will offer additional optimization VIs that will
be more suitable for your problem. The interface will be much
easier to use and more flexible, and we will have more algorithm
options. That is about all I can say until we release LabVIEW 8.0.
Hope this helps,
Jim
