[sdiy] 4-pole SVF theory

[Tom Wiltshire]

On 5 Aug 2011, at 01:02, David G Dixon wrote:

>> If you look at current single Chip SVF's implemented with 4 
>> pole capability, late CEM, late SSM and Roland, they are ALL 
>> 2pole x 2 serial. And those of more discrete design JP6 and 
>> others that i cant remember the name of are also 2 pole x2 
>> serial. I bet the designer of these had a good reason for 
>> this, particularly the CEM since they where designed for CD 
>> applications.
> 
> Most of the filter textbooks I've looked at always generate multipole
> filters by putting 2-pole filters in series, and tacking on a 1-pole if an
> odd number is desired.  This would certainly be the easiest way to go about
> it.  Of course, if you were using, say, a 2164 as the VC resistor, then you
> could control both filters with the same control voltage.
> 
> What I'm not sure about is what you'd actually get from all of the output
> taps.  For that I'd need to crunch the transfer functions (based on current
> balances).  There might also be some clever things you could do with
> feedback across the two stages.
> 

I agree with Karl, I bet those designers did have a good reason. But what I want to know is "what is it?"!

It's easy to run one 2-pole SVF after another and get 24dB HP/LP/BP responses, but the simplest way is to change the tap that you input into the second filter (e.g. feed the HP output from the first into the second, then take the HP output from the second for 24dB HP, for instance). But like this, you don't get simultaneous 4-pole LP/HP/BP outputs (like you do on the 2-pole version), since you have to switch things around to change the response.

Perhaps the reason is the extra versatility you get from having the two filters separate. You get 12/24dB switchable responses for free, I suppose.

T.