#cracklebox

So this is another Eurorack patch using external devices: here, my homemade cracklebox.

I built this thing quite a while ago; it's a copy of the classic STEIM Krakadoos design. Its oscillator exploits bugs in the design of the ancient μA709 op amp chip to produce squelchy and crackling sounds when you touch the brass buttons on the left hand side. I've been intending to supplement the audio output jack on it, which is directly wired into the discrete amplifier circuit, with an output transformer; I've been told that doing that will make it so that plugging it into something doesn't couple that amplifier to the internals and change the sound.

But I decided tonight that before making any modifications, I should try plugging it into the sync input of my VCO 3340. This audio clip is the result: a solid chip oscillator being intimately tied to a weirdo.

In the late 1960s, Michel Waisvisz (later of experimental music/art lab STEIM) developed what he called the crackle circuit. It originated in poking with his fingers at the exposed connections in a broken electric organ, and then radios, coming up with what later "circuit benders" call "laying on hands". In 1973, he joined STEIM and created the Krakadoos, or, translated into English, the Cracklebox. It was simple, perhaps deceptively so — the interface is six exposed copper plates on a printed circuit board, the whole in a wooden box, powered by battery and with an included speaker. It made interesting sounds, particularly in those early days before electronics miniaturization really kicked in, and the thing gained a kind of mystique. STEIM has begun making them again in the last several years, but they're pretty expensive.

The Cracklebox works by using the performer's flesh as part of the loop feeding the circuit, which is composed of an early operational amplifier chip which lacked internal compensation and thus was easy to send into oscillation, producing the characteristic whine/squeal/crackles. The circuit diagram shows the whole thing feeding into a second amplifier (the four transistors on the right), and then to the speaker.

Fast forward to a little over a year ago, when I found the diagram above and a source for the μA709/LM709/MC1709 chip required. (Modern opamp chips — which is to say, those introduced after 1980 or so — apparently fix the runaway oscillation "problem" that lets any of this work.) Now, I considered having a circuit board made to mimic the interface of the original, but decided to be a little more inventive with my reproduction, and not incidentally use materials I had on hand.

This is my Wax Wolf Cracklebox. The case is a plastic box intended to house a walkie-talkie back in the '80s; I picked up a dozen of 'em on the surplus market years ago, and enjoy using them when I want an enclosure with a speaker inside. (Surplus Sales of Nebraska still sells them for 35¢ each, last I checked.) The brass knobs on the front, taking the place of the PCB plates, were originally stand-offs, intended to slide into slots in a metal case and allow a circuit board to be screwed down onto them without contacting the case and shorting anything out; here each has a wire from the internal stripboard screwed in, connecting it to the circuit. The top has a slide switch and a power LED, as well as an output jack that switches the speaker off, and I made the labels on the local makerspace's vinyl cutter.

Cracklebox and Clouds, down a long, long hallway.

Here's a quick diagram of this portion of the circuit so you can follow along:

The oscilloscope trace starts with nothing touching the contact points, and the result is a big, long smear of rolling signal. The next bit is from me touching the two right-hand contacts — the two input frequency compensation leads — and gives a distinct mountain-looking waveform. Then I'm touching the two top terminals — the ones that I've found most often give a semi-controllable note — and we get something close to a square wave. (Because this is an old analog 'scope, a square wave that's got a sharp enough rise or fall won't show the upwards or downwards stroke, leading to the parallel dotted line you see.) There's some further noodling after that, and different combinations of all the contacts, but this was the "aha!" portion.

So basically, the core μ709 op amp is supposed to amplify greatly the difference between its inverting (-) and non-inverting (+) inputs. With those floating (not connected), and without anything on the frequency compensation leads, the output wants to amplify any stray voltage, so it slams the output into its maximum (9v here) and its minimum (0v) over and over again, basically outputting a square wave signal high above the range of human hearing. (How high? I haven't measured yet.)

When the frequency compensation leads get joined via the variable resistance that is human flesh, that oscillation slows down and gets captured as the jagged wave you see. And when the top contacts are bridged, the signal stays a square-ish wave but runs much more slowly than the normal runaway, as varying amounts of the output feed back into the frequency compensation. This lowers the pitch into an audible note. Pressing harder on the contacts, or contacting more of the surface, decreases the resistance between the two, letting more and more of the output back into the frequency compensation, driving the pitch lower and lower.

I've posted before about my Cracklebox — a homemade noise box based on the early electronic musical instrument by STEIM — and how I built it. It's a fun little gadget, which you play by touching the brass knobs on the front; depending on which ones you press, and how hard, it makes a variety of sounds that range from tones to squeals to chatters. But I've long had a nagging suspicion that I made the wrong decision when I made a change to the circuit when assembling it.

The bit in question is the amplifier. The schematics I got have a strange arrangement of transistors. At the time that I built it, I didn't understand the layout, or have the specific transistors the circuit called for. So I gave up on it and instead used a popular audio amplifier chip, the LM386, which gets used in everything from musical birthday cards to guitar amps and pedals. This worked, but I wasn't getting the same range of noises out of the thing that I'd seen others achieve; either the amplifier is a bad substitution, or I just wasn't playing it right (or both!).

The amplifier circuit is above on the left. Now, I'd seen the middle circuit, known as a push-pull amplifier. It uses two transistors of opposite polarities — that is, one NPN and one PNP — to get better efficiency than a single transistor amplifier. But the Cracklebox amplifier isn't exactly that — it's a push-pull amplifier made of two Sziklai pairs, which are seen above right. This layout uses two transistors of opposite polarities to pretend to be a single transistor with a very high current gain. So when you stack the two together, you get a very high gain, efficient current amplifier, at the expense of some distortion and, according to what I've read, a small "dead band" that ends up filtering out some low-level noise.

Anyway, I finally decided to find out if the amplifier really did make a difference. I like the setup I have, so I didn't want to build a whole new circuit and enclosure -- and I'm not entirely sure where I put my LM709 opamps -- so I ended up using a scrap of protoboard to build the amplifier as a daughterboard add-on, and (since I know you can do this now) subbed in 2N3904 and 2N3906 transistors for the BC547 and BC557 transistors, respectively.

Here it is, just sort of hanging off the side of the main circuit. I wanted to be able to go back to the LM386 if things went badly. It didn't work at all when I first turned the changed device on, but some quick troubleshooting revealed the connection I'd neglected to include, and it started making noises.

And it's much different. Some of the knobs are much more sensitive (and it's a little clearer what each one is doing), there's a greater range of sounds, and it's quite a bit louder. I may have to go back in and put in a volume knob. But it was definitely the right call to fix it like this.