ΣSigmaStudio Ask AI Русский Log in

Very low frequency/slow response filters

SigmaDSP & SigmaStudio J3627 Jun 27, 2014 00:00 2 replies Original on forum ↗
Thread Summary

Thread Summary

The user wants to create a system with a very slow RMS response, rising to full value over several minutes. The final answer suggests using a Second Order Double Precision General filter with custom coefficients for a 0.01 Hz cutoff frequency to achieve the desired low-pass filtering with no overshoot. The accompanying answer provides a method to calculate the coefficients and confirms the filter's stability and response characteristics.
AI Generated Content
Subscribe
J3627 Question
Jun 27, 2014 00:00

I would like to make a system that responds very slowly to an RMS measurement, like rising to full value over several minutes. Setting the RMSTC to 1 isn't slow enough, so I thought I could follow it by a LPF (also the envelope block has a linear decay which I don't want, while the LPF would have the correct log decay), but none of those can be set to a low enough frequency either. State-variable filters are supposed to be numerically stable at low frequencies? But the lowest SigmaStudio lets me set it is 0.1 Hz with 0.5 Q, and this still rises to full value within 20 seconds (and the 0.5 Q still allows a little overshoot, which I don't want). Any advice on making a lower frequency filter with no overshoot?

low_pass_filter

KJBob Reply
Jun 27, 2014 00:00

Hello,

You can type your own coefficients into a Second Order Double Precision General filter to bypass the frequency limitation. Since you desire no overshoot, I recommend a set of first-order coefficients (and they are easier to calculate anyway). The remaining coefficients are zero. The picture below shows how to calculate the two needed coefficients according to your desired lowpass cutoff frequency. These are taken from the first-order coefficients in the tutorial,

Calculating Filter Coefficients to Store in a Microcontroller

except that the A1 coefficient's sign is negative due to a sign reversal as stored in the DSP. The numbers shown are for a 0.01 Hz cutoff frequency, thus the plotted frequency response is -50dB down at 3 Hz (as low as this window goes). It's also easy to build this filter from an adder and feedback delay, but the ready-made filter is double precision, which you'll need at this frequency.

Best regards,

Bob

5466.82f51e0ee35a3f8510a09e11276904b9.png
5466.82f51e0ee35a3f8510a09e11276904b9.png
5466.82f51e0ee35a3f8510a09e11276904b9.png-620x306.png
5466.82f51e0ee35a3f8510a09e11276904b9.png-620x306.png
EZAdministrator Reply
Aug 2, 2018 00:00

This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

Thank you,
EZ Admin