First how to select low-pass filter order?
If I thinking right I need to know guaranteed attenuation value(g) in dB and frequency(Fg) for this value to calculate low-pass filter order.
For example cut-off frequency Fc=20000 Hz, g=50 dB, Fg=6000 Hz. So As=20000/6000=3.33
(Butterworth filter fading)
At least 4 order is needed.
The question is how to know what values of Fg and g should to use?
If I thinking right I need to know guaranteed attenuation value(g) in dB and frequency(Fg) for this value to calculate low-pass filter order.
For example cut-off frequency Fc=20000 Hz, g=50 dB, Fg=6000 Hz. So As=20000/6000=3.33
An externally hosted image should be here but it was not working when we last tested it.
(Butterworth filter fading)
At least 4 order is needed.
The question is how to know what values of Fg and g should to use?
There are passband frequency Fpb, passband attenuation Abp, stopband frequency Fsb, stopband attenuation Asb need to know to calculate reconstruction filter (low-pass active filter which used after DAC output). Right?
Sound frequency range is 16-20 kHz so Fsb = 20000.
Fbp should to choose close to Fsb but not too hight: take into consideration that filter order must be about 2-5. Or not?
How to decide what Apb and Asb values is use for calculation?
Sound frequency range is 16-20 kHz so Fsb = 20000.
Fbp should to choose close to Fsb but not too hight: take into consideration that filter order must be about 2-5. Or not?
How to decide what Apb and Asb values is use for calculation?
Butterworth is probably not the best choice for a reconstruction filter, unless you value phase response over amplitude. But is this really a reconstruction filter with 6kHz for the passband edge?
For reconstruction, the first choice is how much aliasing (strictly speaking, imaging) can you tolerate? A 20kHz signal (the worst case) generates a 24.1kHz image when the sample frequency's 44.1kHz. The image moves much further out, to 68.2k though with 2X oversampling.
For reconstruction, the first choice is how much aliasing (strictly speaking, imaging) can you tolerate? A 20kHz signal (the worst case) generates a 24.1kHz image when the sample frequency's 44.1kHz. The image moves much further out, to 68.2k though with 2X oversampling.
Last edited:
Are any step by step example of low pass filter calculation when you know only frequency range?
And where is aliasing present in reconstruction filter calculation?- I didn't find anything about this in internet. Russian community told me that reconstruction filter is common low pass filter for cut off signal that not in sound range .
If you will help me to understand how to design LP filters I would be grateful a lot.
And where is aliasing present in reconstruction filter calculation?- I didn't find anything about this in internet. Russian community told me that reconstruction filter is common low pass filter for cut off signal that not in sound range .
If you will help me to understand how to design LP filters I would be grateful a lot.
Designing low pass filters isn't really something that can be explained in a posting on a DIY forum. You'd be better off getting hold of a book on filter design. One I recommend is by Arthur Williams - Electronic Filter Design Handbook: Arthur B. Williams, Fred J. Taylor, F. Taylor: 9780070704343: Amazon.com: Books
What's the application circuit you're using with the reconstruction filter, and is it a passive or active one?
What's the application circuit you're using with the reconstruction filter, and is it a passive or active one?
Designing high specification analogue filters is a very specialist job, although made easier nowadays with computers. I suspect the OP is still at the stage of setting the required specification? He needs to decide how much image he can tolerate, and how much HF signal he can afford to lose. Or is he asking that too?
- Status
- Not open for further replies.