I was gifted a pair of Benjolin PC boards by Alan Rosetti. The Benjolin was originally designed by the late Rob Hordijk. There’s a good article about it on Perfect Circuit. You can find many different implementations of the Benjolin, including the Forestcaver Benjolin files on github. The circuit for the board I received is based on an open source design posted by Jolin. This board is a through-hole version, with Jolin’s enhancements, some of which I decided not to use (external filter audio input, vactrol VCA on the filter output). I did use the 5X gain amplifier on the filter output.

Alan included a layout for the PCB, showing both sides. This was invaluable during the build and testing.

My design choices

A Benjolin is comprised of two LFO/VCOs, a Rungler (another of Mr. Hordijk’s inventions), and a state variable VCF. In a typical Benjolin the two oscillators provide the clock and data inputs for the Rungler, which is a shift register, sometimes offering switches for alternate external inputs. Filter input comes internally from a PWM output, which is a comparator looking at the triangle wave outputs of the two oscillators. Some of the Rungler output is also mixed into the filter. These interconnections are fixed.

I wanted to separate out the oscillators, Rungler, and filter into fully patchable modules. In line with this approach I have designed a 2U 19-inch panel for a dual Benjolin with isolated components. It will be a banana jack panel.

This is an image of the prospective Front Panel Express panel design. It’s a symmetrical design. Going left to right are the modules Oscillator 1, Oscillator 2, Coupler, Filter, Rungler, Rungler, Filter, Coupler, Oscillator 3, Oscillator 4.

Oscillator

The four oscillators are identical. Each will have an initial frequency knob, plus two attenuated CV inputs. Bipolar triangle and square wave outputs are available, each approximately +/-4.5 volts. The oscillator frequency ranges from LFO up to about 5KHz.

Coupler

This is the PWM output, renamed. It’s the comparison of Oscillator 1 and 2 triangle waves. The output is bipolar, +/-4.5V. The Coupler features three bi-color LEDs, indicating the two triangle waves and the coupler output. The LEDs each have their own driver, running off separate power rails, and buffered to avoid loading of the output signals.

Filter

The Filter has initial Frequency and Resonance knobs, two signal inputs with attenuators, two CV inputs with attenuators, the output, and a switch to select between Low Pass and Band Pass.

Rungler

The Rungler has Clock and Data (called XOR_IN on the schematics) inputs, XOR and Rungler outputs, plus three additional pulse outputs, Q6-Q8. The pulse outputs are bipolar and fully buffered (my modification, see below) and feature unipolar LED indicators. I decided to use the on-board LED drivers, but build my own buffers for Q6-Q8, as these are CMOS outputs. The Rungler output is an 8-bit step wave created with an R-2R resistor ladder network driven by Q6-Q8. XOR is the resolved data signal for the shift register. Normally it comes from the Data input, but when looping is taken from Q8.

The right side of the panel duplicates the left; the right Coupler comparing Oscillators 3 and 4.

Circuit modifications I made

This is an unusual circuit in that it runs on a +/-9 volt power supply. Alan’s PCB includes positions for three types of fixed voltage regulator. Minus 9V fixed regulators are difficult to find. The one I used is DI79L09ZAB, Diotec Semiconductor LDO Voltage Regulator. The positive 9V regulator I started with is UA78L09ACLPE3. Both of these regulators are in TO-92 packages. But the small +9V regulator didn’t work right. It could have been defective, or just not suitable. Luckily I had also ordered some super duper +9V regulators, the Texas Instruments LM2940CT-9.0/NOPB LDO regulator, a pin-compatible match for the old LM7809.

The only design flaw I found on the PCB was an almost complete lack of bypass capacitors. I remedied that by adding 10uf electrolytic caps on the 15V power inputs, 1uf film caps on the regulator outputs, plus a smattering of 100nf caps on the back of the board. N.B. The LM2940CT-9.0/NOPB LDO regulator has a different requirement for output bypassing than the old 78xx series. It needs at least 22uf at the output terminal. I replaced the 1uf film cap with a 100uf electrolytic.

The four pulse outputs were originally going as close to the power rails as the op amps would do. I modded the output structure to have a 2K/2K voltage divider, cutting the voltage in half to around +/-4.5V. These mods are evident in the back side photo, below.

Schematics

I redrew the schematics to show the mods I made. First is the oscillators and coupler, and includes the LED driver schematics. NOTE: I ultimately used a different LED driver design. See that schematic on my later post Benjolin Project Progress.

One oscillator schematic is drawn; both are identical.

The second schematic is the Rungler. I used the onboard LED drivers but added off-board op amp buffers for the three bit outputs, Q6, Q7, Q8. These buffers are powered by +/-15V, and divide the output voltage for a resulting +/-4.5V pulses on the jacks. When the LED is off, the output at the jack is around -4.5 volts.

The third schematic is the filter. I added two attenuated inputs, and also a switch in front of the 5X gain amplifier to choose between BP and LP outputs.

Testing

For testing I wired up the frequency pots for the oscillators and filter, the filter resonance pot, and the loop offset pot. Also the loop switch. I temporarily soldered three LEDs for testing. Power comes from my +/-15 volt bench supply.

I am glad I used sockets, because it greatly simplified diagnosing the bad 9V regulator. (It was putting out about +8.8 volts before adding the chips, but when all the chips were added the positive rail weirdly sat at minus 0.9 volts!)

Careful as I was to try and follow the schematic, alas most of the pots ended up wired backwards. Well that is pretty easy to fix, as I will later, when panel assembly begins. Now it’s on to building the second PCB. The second is always easier than the first!

A Sound Demo

Here’s the video.