Getting Error while implementing Filter object.

Hello,

I am using the Digital Filter with properties. I have defined filt(1) as:

 digitalFilter with properties:

            Coefficients: [20x6 double]

   Specifications:
       FrequencyResponse: 'bandpass'
         ImpulseResponse: 'iir'
              SampleRate: 1000
      PassbandFrequency1: 59.9000
          PassbandRipple: 3
    StopbandAttenuation2: 80
    StopbandAttenuation1: 80
      PassbandFrequency2: 60.0800
             FilterOrder: 40
            DesignMethod: 'ellip'

Similarly, I have other filters under filt object as filt(2), filt(3) upto filt(8) each at different frequencies.

Now, for some reason I get the error, when my program tries to run the line:

Q = filter(filt(1),x);

where, x is a 600000x1 column vector with only positive and negative real values (resembling a long sine wave)

The error that I get is:

Subscript indices must either be real positive integers or logicals.
  Error in digitalFilter/filter (line 870)
      y = obj.FilterFunction(obj.Coefficients,x);

Could anyone please point out the problem? Thanks a lot.

[Walter Roberson]

Could you include the commands you use to create the filter? Also, which version are you using?

[Riyasat]

I have already created and saved the various filters using

designfilt

and saved them as .mat file. The program then loads the .mat files and assigns the filter to the variable filt. It assigns eight different filters to indices 1 to 8 in the filt.

This program, used to work perfectly an year ago. I need it again, and can't figure out where's the problem.

The version - MATLAB R2016a with DSP Toolbox.

[Walter Roberson]

I am not all that familiar with designfilt. It would be easier if you could give specific commands to create a filter that you have difficulty with, or if you were to attach the .mat that had the stored filters.

[Riyasat]

designfilt

command opens up an interactive interface where you can change and tune your filter requirements. I have optimized the filters, and saved the results as a .mat file. [This whole program used to work, was part of a significant project which we completed succesfully, but after a year doesn't seem to work somehow.]

I have attached the filter_8th.mat file. There are various filters here. I have already created these filters before hand because Matlab takes a lot of time to create these.

In this .mat file, the filter objects are named as D_n1_50, D_n1_60, D_sig50, D_sig60 etc. (easily recognizable from the Workspace.)

The line of code that assigns these filter to the variable filt is

filt = D_sig50

for example. (Depends on a conditional if-else structure which filter would be selected.)

I have also, attached the variable x as a .mat file.

Answer 1

[Walter Roberson]

When I design a filter matching the specifications of D_sig50(1), I see for it

                        FilterFunction: @filterIIR
                      FiltfiltFunction: @filtfiltIIR
    GetCoefficientsForAnalysisFunction: @getCoeffsForAnalysisIIR

whereas for the loaded objects

                        FilterFunction: []
                      FiltfiltFunction: []
    GetCoefficientsForAnalysisFunction: []

So either the filter functions were not saved or else the saved function reference could not be located at the time of reload of the objects and [] was substituted.

... Experimenting with a few back releases, I find that the filters work if you load them in R2015a, but not if you load them in R2015b.

[Riyasat]

How did you check this? Could you please share the commands/codes.

[Walter Roberson]

Load the file. Then

filt = D_sig50(1);
sf = struct(filt) ;
sf.FilterFunction

The field does not exist in R2015a because the implementation was different

[Riyasat]

I think this was the issue. I generated the filters again in MATLAB 2016a, and now it's working properly. Thanks.