Scott Stites Synth DIY


Birth of a Synth | Modules | Buchla 291 BPF Adaptation | VTL5C2 Phase Shifter | Soft Sync | TriPhase/QFG, Sample and Hold,DSC2000 | 291 Filter Meets Dark Star Chaos 2000 | Patches | Schematics | Synth DIY Links | Kits, Modules and Systems | Synth DIY Music | Contact Me
Buchla 291 BPF Adaptation

The Buchla 291 BPF+ Project

Jeff Pontius and I have been working with the Mark Verbos treatment of the Buchla 291 Band Pass Filter. We've experimented with using an extra Vactrol for voltage controlled resonance, adding a signal mixer to the front end, and providing more CV inputs than those originally on the 291 BPF for Center Frequency, Bandwidth and FM input. One of the more exciting discoveries is that HP and LP responses can be tapped from the filter. The HP output is tapped from the output of the first buffer, and the LP output is tapped from the junction of the resonance pot and the 22nF cap. The LP output is inverted, but can easily restored to the original phase with an inverting buffer. These samples, however, were taken with no buffering - IE a wire was just placed at the points mentioned above (easy when breadboarding).

Mark Verbos has a PCB layout of his adaptation of the original 291 Dual Bandpass Filter here:

Mark Verbos, Simple-Answer


Jeff Pontius designed this Schaeffer panel of the 291 Core Filter, incorporating all of our modifications. For an explanation of the controls, see the bottom of this page.


The design in my adopted format.

Out of laziness, the same patch was kept throughout the samples. A sawtooth and a pulse wave are mixed together into my Vactrol phaser (with resonance turned all the way down), fed to the 291, and then on out through the VCA. A simple AR EG is used to control the VCA, and occasionally the frequency of the 291. A triangle wave is used for vibrato on the oscillators, and a square copy of it is used occasionally to modify the FM input of the 291 BPF. The content of the samples really doesn't do justice to the filter, (I think it lends itself nicely to more 'Subotnick-like' work), but I was in a hurry, and a thousand samples wouldn't cover the range of this wonderful filter anyway.....

Samples are now fixed (were left channel mono before)!!

BP, LP and HP Sweeps

The following sample first sweeps the filter frequency in BP mode, then LP mode, and finally HP mode. Bandwidth is set to narrowest.

BP LP HP Sweep (542 KB)

HP, LP and BP Note Sequences

The following sample starts out with a rather unartistic sequence of notes in HP mode, then another sequence in LP mode, then finishes in BP mode. Bandwidth is a bit wider than the first sample, and the FM input is modulated by a square wave, while the frequency is modulated by the EG.

HP LP BP Notes (453 KB)

HP, LP and BP Variable Bandwidth Samples

The next three samples use notes in the higher range. Each sample starts at a narrow bandwidth with no FM modulation, then the bandwidth is widened and the square wave modulation is introduced to the FM input. The EG is controlling the frequency of the filter in each sample.

High Pass Sample (398KB)

Low Pass Sample (394 KB)

Band Pass Sample (389 KB)

Sample and Hold

Here are a couple of samples of the 291 BP output in action, being knocked around by a sample and hold. The sample and hold is set for a fast series of pulses.

The 291 is being fed two pulse waves. The center frequency input is connected to the output of a sample and hold. The FM input is fed the envelope out of the AD EG. The AD EG is being triggered by the sample and hold clock. The VCO's are being controlled by the sample and hold output.

The sample and hold is set up in a 'correlation patch', which mimics a Buchla technique - a noise generator is going to a mixer, mixed with the output of the sample and hold and the output of the mixer is feeding the sample and hold signal input. The sample and hold is triggered by an internal clock.

Sample number 3 is my favorite of these three - it is just a clip of a patch that generated ambient music - it *really* shows the 'pluckiness' of this filter. In this example, the LP output is used. Only one pulse wave is being processed by the filter. The sample and hold is being triggered by a VCO pulse wave, and that VCO is in turn being modulated by the sample and hold. A detail of the patch (plus an extra sample of it) can be found on my Patches page.

Sample and Hold 1 (409 KB)

Sample and Hold 2 (508 KB)

Sample and Hold 3 (522 KB)

Here's another sample to show the acoustic twanginess of the Vactrol response when hit with varying levels of CV.

Sample and Hold 4 (625 KB)

The 291 - IMHO a Filter Built for FM!

I've read that Don Buchla didn't care much for resonant filters, and it seems that a lot of his design was intended for frequency modulated systems. Moreover, as I understand it, the System 200 was geared for spatial modulation - IE, to give the impression of advancing and receding sound sources. The following samples were created using two of Rene's VCO3's - one is used to modulate the exponential input of the other. Only the modulated carrier is injected into the filter. The patch is simply the modulated VCO going into the 291, with the output going to a VCA and on through the amplifier, which adds a touch of spring reverb. Two LFO's are used occasionally - one a triangle wave to modulate the bandwidth at certain times, and the other a square wave fed to the FM input of the filter. The EG is used also to modulate the center frequency and control the amplitude of the output VCA. Check out the high pass sample - there's a good example of the bandwidth being modulated by the triangle LFO to give the impression of a repeating delay.

BP Response with FM Input 1 (515 KB)

BP Response with FM Input 2 (506 KB)

HP Response with FM Input (513 KB)

LP Response with FM Input (517 KB)

Voltage Controlled Resonance and Bandwidth

As mentioned before, one of the modifications is the addition of voltage controlled resonance by means of replacing the resonance pot with a resistor and placing a VTL5C3 Vactrol parallel to it.

Obviously resonance is highly dependent upon the bandwidth of the filter (or in the case of High Pass or Low Pass - the slope of the filter). In the original Buchla 291, there was a Q control, and was actually a very small knob on the front panel. I'm unclear whether this was a 'set and forget' calibration type of control, or if it was intended to be actively used while setting up patches. In any event, at wide bandwidth settings, the Q control has little effect, but at narrower settings it has quite an effect. In fact, if a hard voltage is placed on the Bandwidth CV input in the direction of narrowing it even further, you can coerce the module into making it necessary to go out and buy new tweeters.

The following two samples use a single sawtooth VCO connected directly to the filter. The Low Pass output is put through the VCA and on to the amplifier, where reverb is added - other than that, there are no other effects.

This sample has a slow triangle wave controlling the resonance, and a fast square wave pulsing the FM input. The frequency input is slightly controlled by the AR EG.

Voltage Controlled Resonance (320 KB)

This sample, other than exposing my ineptitude at the keyboard, demonstrates the action of the bandwidth on resonance. The setup is the same as above, only the EG is going to both the FM input and the Frequency input - this serves to give the attack a bit more punch, since I'm only operating with an AR EG. In the first portion of the sample, the Bandwidth control is set for 'wide', which corresponds to a shallower slope for the Low Pass output. In the second portion of the sample, the bandwidth is modulated slowly with a triangle LFO. In the third part of the sample, the bandwidth input again is modulated with the triangle wave LFO, and the resonance is pulsed with a square wave LFO to the point of self resonance only when the bandwidth (slope) is at the narrowest (steepest). Again, the Low Pass output is used.

Bandwidth and Resonance

This sample demonstrates the interplay between modulation at the FM input and the resonance CV input. In it, I increase and decrease the level of modulation to the FM input.

Voltage Controlled Resonance and FM Input Modulation (806 KB)

Center Frequency/Both/Bandwidth Controls

Part of our modification is to add an input, polarity/level control, routing switch and 5V sensitivity/10V sensitivity/AC couple selector to the module. This set of controls is designed to send a CV to control either the center frequency, the bandwidth, or both simultaneously. A switch is added to select the sensitivity of the control - the 10V setting is intended for +/-5V signals, the 5V setting was put in with 0-5V envelopes and keyboard voltages. The AC coupling setting inserts a capacitor in the same manner that the FM input was designed.

Today I actually implemented this portion of the circuit on breadboard, following are some samples I made whilst puttering around on the breadboard.

The signal source is a sawtooth run through my Rene Schmitz Late MS-20 clone filter, which is set in the high pass mode. I have the resonance cranked up, and the cutoff set to a sweet spot that I've grown fond of. I'm not using it as a filter in the normal sense - with these settings, the filter actually acts more as a suboscillator with the resonance in self oscillation - it makes one oscillator sound like so much more. This filter is only controlled by the keyboard voltage.

I'll write more about these samples later, suffice it to say that all four of them take advantage of this new set of controls. There's nothing overtly musical about them - just excercises in timbre =-).......

If you're in a hurry, just check out the Sample and Hold one - the VTL5C2 Phase Shifter plays a cameo in it, and there's much Vactrol juiciness.

Keyboard CV on 5V sensitivity (234 KB)

1 LFO for 'Both Input", EG to CF, LFO to BW (597 KB)

LFO for 'Both Input', EG to CF, LFO to BW (354 KB)

Another Sample and Hold for Vactroholics (515 KB)

Notch/Inverted/Mixer Output

The Notch/Inverted/Mixer output function is a means to extend the range of timbres produced by this module. In effect, it is an internal Vactrol crossfader that has the inverted Bandpass signal as one input, and the output of the input mixer as the other input. The crossfader, BTW, is another Mark Verbos creation (Verbos strikes again!).

The idea is that inverting the bandpass signal and mixing it back with the input signal produces a notch filter response. Because this setup allows any ratio of inverted bandpass to original signal, it will go from pure inverted bandpass output to pure signal in, hitting all the timbres in between.

The control, with no CV applied, will control the mix of inverted BP to input signal. With a CV applied, the control will determine the level and polarity of the CV controlling the crossfade between inverted BP and input signal. 0V CV will result in a 50/50 ratio of inverted BP to input signal. -5V CV will result in pure inverted BP, and +5V will result in pure input signal.

In this sample, a sawtooth signal is applied to the filter. The sawtooth is run through the VTL5C2 phase shifter, which is controlled by the same LFO that is feeding the sample and hold, and is also controlled by the sample and hold voltage (note the VCO is not being modulated by an LFO - the pitch bending is coming from the VTL5C2 phase shifter). The sample and hold trigger is controlling the EG. The EG is controlling the center frequency. Via the extra BW/Both/CF input, the sample and hold output is controlling the center frequency and bandwidth, as well as the VCO frequency. At the beginning, there is a 50/50 mix of inverted notch and mixer output. Around the 26 second mark, I bring up the amplitude of an LFO controlling the inverted BP and mixer output mix.

Inverted BP/Mix out modulated by LFO (508KB)

In this next sample, the S&H voltage that is controlling the VCO frequency is also controlling the ratio of inverted bandpass to input signal (which in this case is a single pulse wave). It is patched so that on the higher notes, the ratio of inverted BP signal to input signal is lower, and on the lower notes, the amount of inverted BP is higher. The filter center frequency is modulated by a slow LFO into the CF input, by a fast square wave LFO into the FM input, and by the EG, which is AC coupled to the center frequency control as well as the bandwidth via the CF/Both/BW input.

Inv BP/Input Ratio Controlled by S&H (593 KB)

The rest of these samples only use the 291 and either a single sawtooth or a single pulse wave input. No external modifier, other than the VCA and a touch of reverb, is used in order to make the effect of this feature more apparent. This is a tall order, mainly because so many things affect what comes out of this connector - center frequency, bandwidth, resonance, and the amount of inverted BP to mixed signal all play a role here.

These samples employ two LFO's, an EG, keyboard and a sample and hold at various times for control. At various times the center frequency, bandwidth, and notch mix are modulated.

Notch/Mix Sample 1 (249 KB)

Notch/Mix Sample 2 (508 KB)

Notch/Mix Sample 3 (451 KB)

Notch/Mix Sample 4 (462 KB)

Notch/Mix Sample 5 (384 KB)

Notch/Mix Sample 6 (533 KB)

Notch/Mix Sample 7 (182 KB)

Notch/Mix Sample 8 (511 KB)

Notch/Mix Sample 9 (455 KB)

The Schematic

Below is a link to my schematics page that contains the schematic that Jeff and I used to breadboard the filter. It was derived from Mark Verbos' version of the Buchla 291 that replaced the discrete FET's with op amp buffers. There are a few minor changes, mainly in power supply filtering, the addition of the Low Pass and High Pass tap points, the addition of reference designators (to help us coherently discuss the circuit over email), and the value of C1 was changed to 1000 pF from 980 pF.

This isn't the final version of what we're doing, but it's mainly what is coherently down in a schematic. We added voltage controlled resonance (using a VTL5C3), extra VC points, and a simple input filter, plus a few other odds and ends. We don't have all of the extras tacked down, but I'll post schemo's when we do. Please remember that the LP tap point is inverted, and you may want to buffer/invert it. We'll probably wind up buffering the HP tap point for GP, too, though all of the samples taken above were made just by connecting directly to the tap points. Above all, remember this:

This schematic does not follow Mark Verbos' PCB layout - the pin assignments on the op amps may be entirely different (I haven't checked yet). Use it as a guide to find the tap points, mainly on one side of the resonance pot for LP, and on the output of the first buffer for HP.

Click the schematic to go to the schematics page.

Go to schematics page.

Features of the 291 Band Pass Adaptation Project

Note: Eventually the full modifications will be published on this Web Page. No big secrets - just lazy ;0)


RESONANCE - (Part of original filter, originally labeled 'Q') - Sets the amount of initial resonance of the filter.

RES CV - (Modification to filter) - Controls the polarity and level of the CV present at the RES CV input.

IN 1 - Controls level of signal applied to IN 1 connector.

IN 2 - (Modification to filter - original filter had one input) - Controls level of signal applied to IN 2 connector.

BANDWIDTH - Controls bandwidth of filter.

BW CV - Controls polarity and level of the CV present at the BW CV input.

BW/CF CV - (Modification to filter) - Controls level and polarity of CV present at BW/CF CV connector (applies CV to either bandwidth, center frequency or both as determined by the BW/+/CF switch).

10V/5V/AC Switch (left side) (Modification to filter) - Selects sensitivity of resonance CV input to 5V or 10V level, or selects AC coupled input for the resonance CV input.

10V/5V/AC Switch (right side) (Modification to filter) - Selects sensitivity of BW/CF CV input to 5V or 10V level, or selects AC coupled input for the BW/CF CV input.

BW/+/CF Switch (Modification to filter) - Selects routing of signal applied to BW/CF CV input to control either center frequency, bandwidth, or both.

CENTER FREQUENCY - Sets initial center frequency of filter.

CF CV - Controls polarity and level of the CV present at the CF CV input.

FREQ CV - (Function labeled FM on original filter) - Controls the level of the signal present at the Freq input.

NOTCH/IN - (Modification to filter) - Dual purpose. With no CV applied to the N/I CV input, controls a balance between the product of the two signal inputs and an inverted copy of the bandpass output. With CV applied, controls the level and polarity of the CV present at the N/I CV input.


IN 1 - Signal input 1

IN 2 - Signal input 2

RES CV - Resonance CV input

BW CV - Bandwidth CV input

CF CV - Center Frequency CV input

FREQ CV - Frequency CV input (AC coupled)

BW/CF CV - Bandwidth/CF CV input

BAND - Band pass output

HIGH - High pass output

LOW - Low pass output

N/I CV - Notch/Inverted mix CV

N/I OUT - Notch/Inverted signal output

My finished version of the filter, which isn't nearly as ergonomical as Jeff's, but rather is in my own cramped little format of MOTM size dimensions, using 3.5mm jacks will have a couple of extra functions:

1. An additional audio input.

2. A 'multiple' for the CF CV input.

3. It retains the original level control of the Buchla design, which in this case acts as a master level control, because it comes after the input mixer.

4. Multipled outputs (three connectors each for BP, LP, HP, and Inverted BP/Mixed audio).

5. External inputs to the internal Inverted BP/Mixed audio crossfader.

With number 5, the connections are normally switched so that the Inverted BP/Mixed Audio function is the default mode. Connecting to one input will replace the mixed audio portion of the cross fader, connecting to the other will replace the Inverted BP portion of the crossfader. Because of the nature of the circuit, any input to this crossfader will come out inverted.

So, if one were to replace the inverted BP with another signal, this signal would be subtracted from the mixed audio input as the crossfade is implemented.

If a signal is connected to the other connector, this will replace the mixed audio portion of the function. In that case, the inverted bandpass will be algebraically added to the input, so that they would sum together and come out inverted.

If signals are inserted into both connectors, the function is acting like a stand-alone crossfader independent of any filter function (except that the output will be inverted).

A number of possible uses occur to me - for example, the LP and HP outputs could be patched into these inputs, and one could crossfade between LP and HP.

Even without this external input, I can see the Inverted BP function being used to loop a signal through some other device and applied to one of the audio inputs.

Another use of this function would be to use the input mixer in a patch that actually isn't using this filter - in that case, with the control set for only Mixed output, the input mixer is acting as a stand-alone mixer. This would come in handy if one had run out of audio mixers.