From Jurgen Haible:
In 2004, I built a String Filter - I even made a PCB layout for this, which was untypical for me at the time. (Built most of my stuff on veroboard back then.)
I liked it a lot, and I got a lot of positive feedback for the sound demos.
Unfortunately, the filter topology, a variant of the GIC filter and probably my own invention turned out to be a failure: It worked, but I only got it to work stable with NE5532 chips of a certain date code. Hardly something I could publish for other DIY'ers to build.
Now, I've started that project all over again, with a different filter topology.
Also very different from the classic Electronotes single-opamp filters (and as far as I can tell, also from Moog's string filter), I'm now going for a two-opamp biquad topology. This generally not so well-known filter uses two integrators like the ubiquitous 3-opamp state variable and biquad filters, but unlike these, one of the integrators is non-inverting: The so-called Deboo-Integrator.
But before I describe the new project, which will also be awailable as PCBs (if everything work as intended, this time!), I insert the information of my old string filter project here:
This was inspired by
Ian Fritz, "Filter Bank Design", Electronotes EN#107,
Bernie Hutchins, "A 39-Channel Variable Q Filter Bank", EN#115,
Arthur Benade, "Fundamentals of Musical Acoustics" 2nd edition 1990, and
Kenneth Elhardts fascinating string sounds
A string filter has been on my personal wish list for a long time. The above mentioned Electronotes articles give a good description what is possible with multiple resonances, and also describe a feedback method to control depth of the peaks and notches, starting with filters of fixed Q. And there is the legendary (and rare) Moog CE String Filter that is based on a similar principle.
The Moog filter uses Norton Amplifiers, Ian has used discrete transistor circuits, Bernie has used Deliyannis filters with opamps.
My version is based on LC resonators, where the inductors are replaced with 2 opamp, 4 resistor General Impedance Converters. The advantage of this is that there are only 4 equal resistors, two equal capacitors, and one dual opamp needed per partial filter, and that there are no trimpots required - the downside is that the capacitors must be selected.
I've made a stereo version, where even filter numbers go to one output, and odd filter outputs go to a second output. A "separation" potentiometer allows everything from Mono to extreme Stereo panning. It turned out that just a _slight_ amount of "separation" gives some pleasant room information without messing up the impression of a single sound source.
In addition to this 40-band Filter Bank, I'm using some parametric filters to emulate the Main Wood Resonance, Main Air Resonance, and Bridge Resonance of various sized string instruments. I plan to make an individual filter for Cello, Viola (Bratsche) and Violin, based on the plots in Benade's book. Unfortunately, Benade doesn't give detailed dB levels for his courves, so I have to try different gain and Q factors. So far, I have finished the Cello Filter. (which still needs some fine tuning of the Q's and gain factors, no doubt! Also have to find the right width of the Notch around 1.5 kHz for the Cello.) Viola and Violin will be next.
I don't expect to get the sound of real string instruments (this needs some physical modeling of the bow/string system at least!) , but I want to gain some "organic" quality for otherwise electronic sounds.
CS-50 with typical electronic string sound: saw wave, filter quite open, some vibrato and level boost via aftertouch. First dry, then run thru a set of four parametric filters (Wood resonance, air resonance, bridge resonance, bridge notch) and thru the 40-band GIC filter bank. (Sorry for the hum - this was a quick and dirty recording and I have caught a bad GND loop.)
Not related to string instruments. A percussive arpeggio from the OB-8, first dry, then after 11 seconds switched to the 40-band GIC filter. Resonance gradually increased. First just some warmth added, then quite unnatural ringing of the filters.
To bee continued. Hints for improoving Q and gain factors welcome. Circuit to be disclosed later - not quite finished at the moment.
Specs & Downloads
Specs & Downloads
|Additional Resources||Build Guide & Bill of Materials|