Low pass RF filter and phase preservation if possible?

Thread Starter

DaniKowa

Joined Sep 23, 2020
218
Hello, I have a multistage amplifier stage to which I would need to apply a low pass filter with 100Mhz cut as sharp as possible but I see that when I apply the filter the phase is altered. Question: Is there any way to apply in cutoff frequency without losing the linearity of the phase? My hunch is that it's not possible but maybe I'm wrong. Thanks.
 

AnalogKid

Joined Aug 1, 2013
10,986
It is possible, but it takes a lot of parts. One option is to start with a filter topology that has a linear phase characteristic by design, such as Bessel, but this has the least steep cutoff performance compared to other filter types, so it takes more poles to hit a specific cutoff target. Another is to add an all-pass phase correction stage.

Linear Tech and Maxim (and others) make video-specific filter ICs.

ak
 

Papabravo

Joined Feb 24, 2006
21,157
Hello, I have a multistage amplifier stage to which I would need to apply a low pass filter with 100Mhz cut as sharp as possible but I see that when I apply the filter the phase is altered. Question: Is there any way to apply in cutoff frequency without losing the linearity of the phase? My hunch is that it's not possible but maybe I'm wrong. Thanks.
I'm sure you didn't mean 100 MHz. That's \( 100 \times 10^6 \) That's pretty far above the range of human hearing. You can use a Bessel filter to ensure linear phase at the expense of a less steep roll-off in the transition band. You can use a Chebyshev filter for a sharp cutoff in the transition and but some level of ripple in the passband. Pick your poison
 

Thread Starter

DaniKowa

Joined Sep 23, 2020
218
@AnalogKid

I looked at some filters but I always lost the linearity of the phase. Do you have any links for an online calculation for the phase stage?


@Papabravo

Yes i mean 100Mhz (100MHz). I never said is audio filter and sure is a RF Filter. I've tried the filter you suggested but the graphic response mismatch with you have said. I've simulated a filter from rf-tools.com - lc-filter and the results are these:

Chebyshev:

Chebyshev3°_2.png

Chebyshev3°_1.png

Bessel:


Bessel3°_2.png

Bessel3°_1.png

Did I do something wrong in the simulation?
 

Papabravo

Joined Feb 24, 2006
21,157
No you didn't do anything wrong. For the Bessel you have a constant group delay in the passband and the phase is nearly linear by design. For the Chebyshev you did not plot the magnitude so there is no way to judge the passband ripple or the transition band roll-off from what you have shown. As has already bee noted you will likely need a higher order filter to get the response you need.

BTW you didn't say it was an RF amplifier in your original post. We get lots of posts where numbers are off by multiple orders of magnitude,
 
Last edited:

Thread Starter

DaniKowa

Joined Sep 23, 2020
218
You're right, I didn't specify it in the title. In the post, however, it was implied that it was RF since we are talking about 100Mhz. However, I edited the title to be fair. ;)
 

Papabravo

Joined Feb 24, 2006
21,157
You're right, I didn't specify it in the title. In the post, however, it was implied that it was RF since we are talking about 100Mhz. However, I edited the title to be fair. ;)
No problem. Once you confirmed that 100 MHz was correct we were both on the same page.
I did check on inductors in the 75 to 90 nH range and they exist with SRFs above 300 MHz with still respectable values of Q in the single digits and DCR in the milliohm range.
 
Top