Proper use of DSP toolbox (filter)

I would like to use properly filters from DSP toolbox. Unfortunatelly I have just used 'lowpass' and it seems incorrectly ...

How should I use it correctly? I should use zoh block before Lowpass? What would be sample rate for it?

Thank yout for your help.

Answer 1

[Walter Roberson]

https://www.mathworks.com/help/control/ug/continuous-discrete-conversion-methods.html

DSP toolbox is for Digital Signal Processing. DSP is inherently working with discrete samples: something that has been measured at regular intervals and the measurements have been recorded as electronic data. It is incompatible to use DSP methods with continuous signals.

You have three choices:

• switch your model to discrete instead of continuous; Or
• use one of the methods to sample the continuous signal; Or
• results design to use continuous filters instead

Answer 2

[Peter Balazovic]

• use one of the methods to sample the continuous signal;

What Simulink block to use for sampling the signal? What sampling rate to set for sampler and DSP Toolbox filter?

• results design to use continuous filters instead

Is there any Simulink blocks for continuous filters which should I use instead existing one?

Thank you!

[Walter Roberson]

https://www.mathworks.com/help/signal/analog-filters.html

[Peter Balazovic]

How to get those analog filters into Simulink though?

[Walter Roberson]

I linked above to a page that listed a number of possible ways to sample. Which one you use is going to depend on your knowledge of the expected behavior of the signal, and of which aspects of the signal are most important to you.

You want a low pass filter. If your requirements were for a high pass filter then in that situation, the higher the sampling rate, the higher your frequency cutoff could be. With a low pass filter, using a higher sampling rate does not give you access to additional frequency range, but it would permit you to be more precise about the cutoff frequency and to have a shaper cutoff.

Imagine for example that you could implement a digital low-pass filter with an acceptable frequency drop over 6 samples. It would then follow that increasing the sampling rate would not reduce the number of samples but it would reduce the frequency bandwidth that the 6 samples represented. If you wanted to be precise that you passed through 120000 Hz practically unchanged but that 120001 Hz should already be cut by 10 dB then you need a higher sampling rate than if you wanted to pass up to 12 Hz and didn't need 10 dB drop until 15 Hz