Op Amp Noise in State Variable FIlter

[Nescafe]

Happy New Year Everyone,

I'm playing with some calculation for state variable filter and try to stretch the bandwidth of the filter to 6 octave, the calculation result have scary high R value (150k-250k) to get the desire bandwidth, so is there any suggestion on what op amp is better suite to take high input resistance (noise wise)?


Thank You.

 

[JohnRoberts]

Happy New Year Everyone,

I'm playing with some calculation for state variable filter and try to stretch the bandwidth of the filter to 6 octave,

Are you talking about the bandwidth/Q of the BP output or the adjustment range?

the calculation result have scary high R value (150k-250k) to get the desire bandwidth, so is there any suggestion on what op amp is better suite to take high input resistance (noise wise)?

Bifet/JFET input opamps are generally better for dealing with high value resistances.

Thank You.

A few quick observations. Using frequency pots as rheostats (variable series resistances) is not optimal for precision. Pot bulk resistance is generally +/-20%. Also in your schematic there is no end limit resistor so the behavior with pot shorted will not be pretty. For SVF I prefer using the frequency pots as potentiometer (voltage divider) to ground, with a fixed resistor in parallel to set LF when pot full down. This also gives you some more flexibility with (pot) values. 

Regarding noise in a SVF the good news is the bandpass filter actually filters the noise too, the bad news is that the noise gain of the first stage matters so high noise gain at that stage can increase noise. If you are going for a very broad Q (wide bandwidth bandpass), a lower noise way to accomplish this is to stagger or space the two integrator sections further apart. For example making the 4.7nF on the BP section smaller, while making the 4.7nF larger on the LP section whill shift the Q wider, without the noise gain penalty.

Have fun.. SVF are very flexible, with several ways to skin the same cat....

JR

 

[Nescafe]

Hi John,
Thank You for the reply,

Are you talking about the bandwidth/Q of the BP output or the adjustment range?

I mean the range / adjustment range for each band

Bifet/JFET input opamps are generally better for dealing with high value resistances.

I will try, but do You think it will be OK with 250k input?

The Q is OK, with the value in the schematic I can get 0.77 - 2.3 like what I expect.

I have try with cap value too, and maybe will play with that more tonight.

Thank You John

 

[JohnRoberts]

Keep in mind the noise of the R is not wide band but just in the band pass of the filter. Try it and see how it sounds.

I used an alternate approach for varying the Q in one of my SVF designs, where the resistance to ground panned between the + and - input of the HF section, that way it took less of either to drive the Q wide or narrow.

JR

 

[Nescafe]

Keep in mind the noise of the R is not wide band but just in the band pass of the filter. Try it and see how it sounds.

Got it Sir.

I used an alternate approach for varying the Q in one of my SVF designs, where the resistance to ground panned between the + and - input of the HF section, that way it took less of either to drive the Q wide or narrow.

I don't know about this method, would You like to give me more information, a link or schematic maybe?

The Q method really interesting since in my calculation from the schematic above, to get the desired Q the avcl of the op amp will be around 5, I think if that can be lower it will be nicer noise wise.

Yesterday I try to alter the cap value as You suggested and yes it can make my r value down to 30k area  ;D, Thank You.

 

[JohnRoberts]

It has probably been 20 years since I designed a commercial parametric, while I did several of them, all different, back in the '70s/80s.

I don't recall all the design equations from memory but IIRC no resistance to ground from either of first opamp (HPF) input pins, with unity gain from all bandpasses, delivers a nominal mid range Q or bandwidth. Adding R to ground from + input IIRC shifts the Q narrow, R to ground a - input shifts the Q wide.

If you want symmetrical adjustment range wide and narrow, a pan between + and - inputs with one R to ground from the wiper will give similar adjustment range and best noise gain for all setting. Biasing the Q range wider if you want asymmetrical adjustment range is perhaps best done noiselessly by tweaking the cap values apart.  While there are several moving parts in the Q equation as I recall, so several different values to tweak.

I had to do a lot of bench work, since some of these tweaks are not obvious, and you want to look out for unintended consequences. If one of the bandpass gains gets too high, that section could clip prematurely, and even though the BP section isn't clipped, the distortion will leak into the signal path.

Melt some solder and try different things..

JR