Talkin' 'bout filters

Discussions about anything analog, digital, MIDI, synth technology, techniques, theories and more.
Post Reply
Logan
Newbie
Newbie
Posts: 2
Joined: Sat Sep 17, 2005 4:49 pm

Talkin' 'bout filters

Post by Logan » Wed Sep 12, 2007 1:11 am

I am working on some analog filters right now, and was wondering if anyone knew anything about the filters used in old analog synths.

Passive or active?

What type of active filter? I don't think it would be Sallen-Key, as then Q (resonance), and cutoff frequency are intertwined.

Does anyone know anything about this?

User avatar
micahjonhughes
Active Member
Active Member
Posts: 702
Joined: Fri Mar 09, 2007 4:27 pm
Location: Chicago

Post by micahjonhughes » Wed Sep 12, 2007 1:43 am

Many are based on Bob Moog's ladder filter. It is an active filter.

User avatar
piRoN
Active Member
Active Member
Posts: 541
Joined: Sun Dec 19, 2004 7:04 am
Gear: A fair bit.
Location: The crotch of The Lucky Country

Re: Talkin' 'bout filters

Post by piRoN » Wed Sep 12, 2007 1:46 am

Logan wrote:I am working on some analog filters right now, and was wondering if anyone knew anything about the filters used in old analog synths.

Passive or active?

What type of active filter? I don't think it would be Sallen-Key, as then Q (resonance), and cutoff frequency are intertwined.

Does anyone know anything about this?
The Korg MS-20 used a Sallen-Key, as far as I know. I don't know what you mean about the Q and cutoff frequency being intertwined... where did you get that?
Ensoniq EPS-m | Elby Designs ASM-2 | Yamaha TQ5 | Akai S3000XL | Akai X7000 | Novation KS-4 | SCI Prophet 2000 | Kawai K1m | Korg Monotribe | Korg Monotron | Roland S-50 | Roland D-10 | Roland MKS-7 | Yamaha RX11

Logan
Newbie
Newbie
Posts: 2
Joined: Sat Sep 17, 2005 4:49 pm

Post by Logan » Wed Sep 12, 2007 2:24 am

Thanks for the tip about the ladder filter. I'll look into that.

As for Sallen-key: here are the equations based on component values:

Image
Image

What I am looking for is one potentiometer for Q, one for Fc. But, adjusting R1 changes both Q and Fc, unless other values are also changed. Am I going about this wrong?

User avatar
piRoN
Active Member
Active Member
Posts: 541
Joined: Sun Dec 19, 2004 7:04 am
Gear: A fair bit.
Location: The crotch of The Lucky Country

Post by piRoN » Wed Sep 12, 2007 2:39 am

I have to admit, I'm rubbish at maths. If I only had my graphing calculator at work...

For an example of a practical Sallen-Key based VCF, look at Jurgen Haible's Korg MS20 clone here.
Ensoniq EPS-m | Elby Designs ASM-2 | Yamaha TQ5 | Akai S3000XL | Akai X7000 | Novation KS-4 | SCI Prophet 2000 | Kawai K1m | Korg Monotribe | Korg Monotron | Roland S-50 | Roland D-10 | Roland MKS-7 | Yamaha RX11

User avatar
micahjonhughes
Active Member
Active Member
Posts: 702
Joined: Fri Mar 09, 2007 4:27 pm
Location: Chicago

Post by micahjonhughes » Wed Sep 12, 2007 2:46 am

I can across this today. I'm sure the answer is there.

mysite.wanadoo-members.co.uk/tstinchcombe/synth/MS20_study.pdf

User avatar
piRoN
Active Member
Active Member
Posts: 541
Joined: Sun Dec 19, 2004 7:04 am
Gear: A fair bit.
Location: The crotch of The Lucky Country

Post by piRoN » Wed Sep 12, 2007 3:57 am

Excellent! So the MS20 did use OTAs... that's answered a question of mine as well!
Ensoniq EPS-m | Elby Designs ASM-2 | Yamaha TQ5 | Akai S3000XL | Akai X7000 | Novation KS-4 | SCI Prophet 2000 | Kawai K1m | Korg Monotribe | Korg Monotron | Roland S-50 | Roland D-10 | Roland MKS-7 | Yamaha RX11

User avatar
JackAstro
Junior Member
Junior Member
Posts: 129
Joined: Fri Jan 06, 2006 4:21 am
Contact:

Post by JackAstro » Thu Sep 13, 2007 7:10 pm

All synth filters are active. They have to be to buffer the stages. Typically they are either 2 or 4 pole.

The big exception is the TB-303. It is 3 pole and the stages are not really buffered. Also, the poles are in different places.

The MS-20 is perhaps my favorite filter. piRoN is correct about is being a Sallen-Key design. It has a buffered feedback stage with a diode pair that shorts the path when the voltage gets high enough. This means the filter distorts the signal at the resonance peak but no where else (assuming you don't overdrive the input).

The resonance control changes the amplitude of the feedback. I am not sure how that decouples it from the cut off frequency though, but it seems to.

I think the Oberheim filters are Sallen-Key also. Most multimode filters are.

chamomileshark
Active Member
Active Member
Posts: 357
Joined: Wed Feb 15, 2006 6:00 pm
Location: UK
Contact:

Post by chamomileshark » Thu Sep 13, 2007 8:34 pm


User avatar
nathanscribe
VSE Review Contributor
VSE Review Contributor
Posts: 2889
Joined: Fri Aug 11, 2006 1:03 pm
Location: The right side of the Pennines
Contact:

Post by nathanscribe » Thu Sep 13, 2007 11:32 pm

There were 2 versions of the MS-20, the first using custom ICs, and the second using LM13600 OTAs. The first version filters are quite different in structure to the second. I've not compared the sound though, not being so lucky... :D

Also, no filter with more than 1 pole will be frequency-controllable with just one pot, unless you use a dual gang for a pair of poles or use one pot to provide control of OTAs or similar. Switched caps are used in some filters (see Maplin or ETI). Resonance can be controlled with a single pot, no problem.

Most active filters in synths use OTAs or similar to act as the variable resistors in the frequency network. There are many articles on this on the web and in books.

The basic types of filter most often found in synths are the 24dB (4-pole) active low pass (for example, most Rolands), the ladder filter (either transistors or diodes, as used in the Minimoog) and the state-variable or multimode (for LP, BP and HP in one unit) as on the Yamaha CS series. Some of these were using off-the-shelf parts (eg, the mini), while some were custom ICs (such as the SSM, CEM found in SCI gear or the Yamaha custom ICs). However, a multimode filter can be designed easily using just a handful of common components.

Hope that helps.

Post Reply