Benjolin schematics

[Rob]

The Benjolin schematic to download as pdf

---

BenjolinSchematic.pdf
Description:	Schematic of the Benjolin	Download
Filename:	BenjolinSchematic.pdf
Filesize:	41.38 KB
Downloaded:	11230 Time(s)

---

[Rob]

Notes on the oscillators:

A SSM2164 gaincell is used as a current source to charge/decharge the cap of the integrating opamp. The next opamp toggles its output of ~+/-5V is exceeded. On the output of the integrating opamp is the triangle wave, on the output of the toggling opamp is the pulse wave. Then the pulse is fed back into the SSM2164 gaincell to reverse charging polarity after the toggling opamp toggles. The SSM2164 has an exponential control curve, so no additional exponential converter is needed. Frequency control range spans some 18 octaves.
The summing opamp that controls the CV input of the SSM2164 has two extra diodes to prevent its output voltage to give the gaincell a higher gain than unity gain (which wouldn't work well on this osc)
See datasheet of the SSM2164 for further details. The SSM2164 is available through Jameco.com.

[Rob]

What the $#%$ is a rungler?

The purpose of the rungler is to create short stepped patterns of variable length and speed. One could categorize the circuit somewhere halfway between a plain S&H and a shiftregister-based pseudorandom generator. It needs two frequency sources to work and basically creates a complex interference pattern that can be fed back into the frequency parameters of the driving oscillators to create an unlimited amount of havoc.

The rungler is basically a CMOS shift register clocked by one oscillator and receiving its data input from the other oscillator. The output bits of the shiftregister are used as a binary code 'to do something with'. E.g. in the Benjolin the last 3 stages of the shift register for a 3 bit code that is fed into a 3 bit DA converter. This DA eight level output voltage is fed back to the oscillator frequency control inputs. The output of the DA is the 'rungler CV signal'. To describe the rungler waveform in similar terms as like a sine wave or pulse wave I call it a 'stepped havoc wave'.

When the rungler signal is fed back to the frequency parameters of the oscillators it will change the triangle waveforms and pulse widths of the oscillator outputs, making other types of havoc waves, like a 'pulsed havoc wave' and a 'sloped havoc wave'. Note that it is these properties of stepped, sloped and pulsed that are of interest in the waves.
(The Dutch composer Jan Boerman formulated an idea in the 1960s about audio signals that are inbetween pitched and unpitched. Havoc waves are probably somewhere in that region, maybe a bit similar to granular synthesis stuff. I haven't really thought deeply about this myself, but Boerman has certainly always been an inspiration to me to try to go into that inbetween territory.)

The rungler will try to find a balanced state. In this way it behaves according to principle from Chaos Theory. There seems to be an unlimited amount of possible balanced states and when a balanced state is just slightly disturbed it can be noted that it takes a little time to find the next balanced state, with noticeable bifurcations, etc. Note that a new balanced state is defined by the exact position of the control knobs plus the previous state it was in.

The first rungler experiments I did were back in 1980 I think, and there are quite a lot of variations possible on the rungler idea. In the Benjolin design the data input for the shiftregister is not just the pulse from the second oscillator but the XOR of this pulse and the last bit of the shift register (inspired on the pseudorandom generator). The XOR is the transistor/opamp combination that actually forms a controllable unity gain/minus unity gain amplifier, a very simple ringmodulator, so to speak.

Tip: An interesting option is to feed the three bits at the end of the shiftregister into the 3 'selection' inputs of a CMOS 4051 eight-to-one/one-to-eight analog switch and e.g. quickly switch between eight audio signals. You can take these three bits from the pins 2, 3 and 12 on the 4021.

The shiftregister used should not be too long, four to eight stages already does a perfect job. Some CMOS shiftregisters can recirculate, which would hold the pattern.

One can expand by having multiple parallel shiftregisters alternatingly clocked on positive and negative flanks of the oscillator pulse and e.g. using the triangles from the oscillators to crossfade between multiple DAs on the multiple shiftregisters, etc. By expanding the number of oscillators and shiftregisters the number of available havoc waves explodes. Basically the rungler is an open ended circuit that can be expanded and chained into multi-rungler networks.

In the current issue of Leonardo Music Journal (issue 19) is an article about the Blippoo Box, the Benjolins Big Brother, and there is more on the rungler circuit in the article as wel as some other thingies I used in the Benjolin as well.

Imho a rungler circuit works best in an analog electronics implementation. It is definitively more alive and surprising due to the slight instabilities in the analog circuitry. I did digital implementations, but they can't beat the 'organic behaviour' of the analog versions. But this is just personal taste...

[Rob]

Notes on the filter:

The filter is a plain vanilla 2-pole state variable with some added tricks. Two SSM2164 gaincells drive integrating opamps to create the poles. Like in the oscillators care is taken that the gaincells cannot amplify more than unity gain. (The oscillators can take a little over unity gain but filter poles really don't like this at all). The cutoff frequency range is again something like 18 octaves.

The input signal into the filter:
The two triangle waves from the oscillators are 'compared' to create one single PWM wave. If one osc is tuned into the LFO range and the other is in audio range you wil hear the classic PWM effect. When both are tuned into the audio range it sounds like a ringmodulator effect. To this PWM wave is added a bit of the stepped rungler wave and this final mix is the input into the filter.

There is a relatively small resistor over the resonance feedback pot to give the (linear) pot a slight antilog curve, a curve that feels more natural on a resonance control knob. When resonance is increased there is also a bit of the filter input signal subtracted from the input signal through the resonance pot, to balance the output level between low resonance and high resonance settings.

Finally, there is a simple trick that takes just one resistor but creates some all-harmonics distortion ('tube-like' distortion) by skewing the output waveform; the original more sine-like resonance peak is skewed to a more sawtooth-like waveform. This is explained in detail in the Leonardo Music Journal article. It makes use of the sine-cosine relationship between the outputs of both poles, the explanation why it works is quite complex but to implement it all it takes is only one resistor. And it makes a significant difference in the sound of a filter.

[Rob]

The three two-pin headers:

These headers are extra CV inputs for oscillator 1 and 2 frequency and the filter cutoff frequency.

Tip: A simple expansion to the Benjolin is to feedback the oscillator 1 triangle wave from the 16-pin output header, via an extra potmeter, to the oscillator 2 frequency input. And a second knob for feedback of the oscillator 2 triangle to the oscillator 1 frequency input header.

I tried to get these two extra pots on the board, but there simply was not enough space on the board. So, it is a nice DIY option, one of many.

[Danno Gee Ray]

Thank you Rob.

This set of circuits will, I predict become famous in the EM world.

Your sharing them with us is generous and majestic.

[oscilloclast]

Thanks Rob! I picked up some of this info in your talk in Baltimore but this give me much more to think about.

So what is the one resister modification to the Filter?

David

[fluxmonkey]

[TekniK]

Thanks Rob!

[Rob]

[oscilloclast]

Rob,

The one control on the Benjolin that I don't understand is the filter sweep. What does it do? Is it an attenuator for the filter CV in?

I went ahead and ordered the Leonardo Journal (really enjoyed the old issue on the Gizmos).

Thanks
David

[Rob]

[oscilloclast]

Rob,

Thanks for the explanation. What a great feature.

David

[SteveBull]

Rob,

Thank you so much for the schematics and the workshop NYC:Harvestworks 2009.

Q1. I'm not sure I see where the three two-pin headers appear on the four pages of schematics. Please enlighten me.

Q2. If one pin of the two pin headers were ground and I was using CV output from the board itself, then am I correct to assume I would only need to connect the pot between the CV out and one header input pin?

Cheers,
Steve

[Rob]

Q1. In fact they don't appear in the schematics, I added them later while routing the PCB. But they are connected through 51k resistors to pin 6 of U5.B, pin 6 of U6.B and pin 3 of U5.A.

Q2. Correct, the CV out would go to the 'top connection' of the pot, ground to the 'bottom connection' and the CV in to the 'middle taper connection'.

[TekniK]

Rob,can you post the pcb layout also?

[TekniK]

nothing?

[fluxmonkey]

[TekniK]

asking shematics in PM with promesses to share other,and then once u get them weaving bey bey is one of your speciality as i was victim myself dealing with you.

So you better keep your ....

[electri-fire]

Ooh, what's going on here? I don't think I want to know.

I do want to recommend a Google search when you need something, and read the EM archives.

This link (pics of the PCB and parts layout):

http://web.me.com/klangbureau/DIY/Benjolin.html

was already here somewhere.

Again, for your convenience.

[TekniK]

[TekniK]

5U adapted version redraw of layout with component implementation and
wiring.

Layout and components checked with shematics ,should be all fine,but its actualy not yet tested.

COPYRIGHT ROB HORDIJK COMMERCIAL USE NOT ALLOWED.

---

RHBJLN-5U.pdf
Description:		Download
Filename:	RHBJLN-5U.pdf
Filesize:	815.41 KB
Downloaded:	3710 Time(s)

---

[mono-poly]

Awesome!

[slabman]

"The SSM2164 has an exponential control curve, so no additional exponential converter is needed."

Rob - I'm always interested in VCOs that don't use the conventional transistor exponential converter. In theory this VCO could be abstracted from the Benjolin and built as a VCO for a modular synth - how well do you think it would perform, ie in temperature stability?

[Rob]

Oscillators built with the SSM2164 work very well, but the 1V/Oct curve is not exact enough for tracking a keyboard. As temperature stability is primarily an issue when tracking a keyboard, and the SSM2164 has no means to exactly trim the curve, the issue of temperature stability is not really important. More important is that you can use such an oscillator for anything except playing perfectly in tune. The thing is that if you trim such an oscillator to produce the proper frequencies for two notes five octaves apart a note played halfway could be like 50 cents off, which is pretty out of tune to the ear. Normally a quality VCO needs three trimmers to tune it properly: a trimmer to stretch the scale, a trimmer to offset the frequency range and a trimmer to compensate for the inherent and accumulated DC offsets in opamps. This last trimmer shifts the exponential curve slightly up and down in a linear way, as the DC offsets do the same. The SSM2164 has no provisions to do this and thus one can never be sure if the scale isn't shifted. Well, you can be sure it ís shifted but not how much and if it is acceptable for keyboard play.