universal RC-bandpass Filter for compressor sidechain

hello,

I'm thinking of a simple rc cirucit that I can fit to the sidechain of any compressor (ssl/1176/...) ...
there was a lot of discussion about filters for the ssl in the past.

what I need is:

a highpass filter to cut the low freq. from the detector and

a lowpass filter to cut the high and the mids, if I want to let the
detector react to the bass part only...


my first design approach is this:

http://www.cmaudio.de/rc_hp_lp_filter.pdf


regarding this I have a few questions:

can I connect the two passive filters in series or do I have to
seperate them with a buffer..

does the first buffer (unity gain) need a cap for decoupling??


thanks,
mat

[quote author="matthias"]hello,

I'm thinking of a simple rc cirucit that I can fit to the sidechain of any compressor (ssl/1176/...) ...
there was a lot of discussion about filters for the ssl in the past.

what I need is:

a highpass filter to cut the low freq. from the detector and

a lowpass filter to cut the high and the mids, if I want to let the
detector react to the bass part only...


my first design approach is this:

http://www.cmaudio.de/rc_hp_lp_filter.pdf


regarding this I have a few questions:

can I connect the two passive filters in series or do I have to
seperate them with a buffer..

does the first buffer (unity gain) need a cap for decoupling??


thanks,
mat[/quote]

The two passsive RC section should be buffered in between unless you make the impedance of the second one much higher than the first---otherwise loading effects will spoil your expected response.

Caps for decoupling, meaning the power supply rails? Or caps to block d.c. ahead of the circuit? If the former, it rarely hurts and usually helps. If the latter, it depends on whether your sources have d.c. on them or not.

the impedance of the second section (lp) is ~ 1/10 of the first hp section..

can I swap the order of the filters, so that the lp is the first to avoid the loading effects...


with "decoupling" I meant the feedback-path of the opamp to prevent oscillation... but I don't know if that's necessary if the opamp is running at unity gain..

"with "decoupling" I meant the feedback-path of the opamp to prevent oscillation... but I don't know if that's necessary if the opamp is running at unity gain.."

No, like you said. Its not necessary in a unity gain opamp, it cant really oscillate because there is maximum feedback at ALL frequencis.

I looked at your schematic and this plain does not work, sorry.

To start with, I assume you are wanting to use it as voltage follower (buffer)? In that case you need to feed the signal into the + input. The - input tied to the output as you have it drawn.

The highpass and lowpass filters need to be swapped around. As it is now drawn, C2 is loaded by R3 (10k) in paralell with R4 (1k) = 910R. This is because OP2 is an inverter and as such, - input is at virtual ground.

C3 does not form a highpass filter with R4 for the very same reason (it is sitting between virtual ground and actual signal ground.

What you need to do is put R4/C3 first, followed by C2/R3. OP2 needs to be a high impedance buffer to avoid loading the filters. Make it a follower or a non-inverting amp. Send the signal into the + input, put, say 10k trimmer pot between - input and output in paralell with 22pF. Add a 10k resistor between - and signal ground. Now you can vary the gain between unity and +6dB.

Unless the load is high impedance, I would upsize C5 to 100uF. Then you are OK for a 10k load.

ok, thank you...

I based the filter on the drawing of the ssl sidechain filter..
http://members.cox.net/capstanrecording/Sidechain_Sch.pdf

to calculate the cut frequencies I used this calculator:
http://www.muzique.com/schem/filter.htm

Thank you Cuelist! For now, I don't understand everything, but I will be re-reading this post a couple of more (a lot more that is) times till it starts to make sense....

[quote author="matthias"]ok, thank you...

I based the filter on the drawing of the ssl sidechain filter...[/quote]

Sure, but in the SSL sidechain filter, C4 thru C7 (68n ... 220n) form a highpass filter with R6 (12k). R6 is the input resistor of an inverting amp, and thus defines the input impedance of that stage as well.

The first opamp in the SSL sidechain is a polarity inverting buffer to allow the L and R channels to be summed together, to provide the insert to the sidechain with drive as well as give the HP filter a low source impedance.

thank you,

here's another rework... what do you think ??

Hello Matthias.

OP1 won´t pass audio as it is.

AC in should be tied to + input and output tied to - input.

I´m thinkin OP2 should be a inverting amp, C2 in series with R4 tied to
- input, + input grounded and the resistor of your choice (10k for unity gain) in the feedback loop to- input.

OP3 should be an non inverting high Z input amp.

Check out this link for basic theory on opamps:
http://sound.westhost.com/dwopa.htm

thanks,

but as I already wrote it is based on that filter:
http://members.cox.net/capstanrecording/Sidechain_Sch.pdf
that has the same opamp configuration..

I think the first OPAMP buffer is drawn wrong. Otherwise, you get the filter seperation mentioned above.