Random spikes and pulses..

Our collaborator Laura Maes at Logos Foundation got a commission for an audio art installation to be installed at Dartmouth College in 2017. Her idea was to have an entire ceiling filled with little circuits producing spikes and clicks powered from solar cells, placed outside. Designing simple circuits to produce pulses seemed pretty simple and our first idea was to use a simple UJT circuit, an oldie but still a valid option. The UJT's are still on the market (here is the datasheet for the 2N2646), but their price (ca. 4 Euro) is much higher than that of another classic, the infamous NE555 chip. Here is a practical circuit: Five components only... The 555 chip at Farnell and at the time of writing was priced at only 20 eurocents if bought in quantities of at least a hundredth. So we decided to go for it, as we needed some 200 of these circuits working in parallel anyway.

However, the problem that raised was how to make a circuit producing randomly varying pulses... Evidently, in the analog domain, we would look for devices having a natural random behavior (such as noise diodes, or transistors configured to operate as such) or transducers for natural random phenomena. The first idea in this direction we came up with, was to use natural radioactivity and the Geiger-Mueller tube used for detecting and measuring it. Every time a particle (alfa) hits the tube,a discharge takes place and a pulse can be observed and made audible. The time between particle hits is random and only the probability of hits gets higher with higher radioactivity levels. It would have been a great idea, if it weren't for the price and complexity... Geiger-Mueller tubes need a pretty high voltage to work well and need a good high impedance amplifier stage behind them. Way too expensive for this project. The classic noise diode at the other hand, required a lot of analog (or digital) low pass filtering containing a lot of quite expensive capacitors in the 1 to 47uF range or, as an alternative, an expensive switched capacitor low pass filter (LTC1563 or similar) followed by a comparator or a Schmitt-trigger..

Thus we examined the old and proven 555 concept. Laura suggested to add a small LED on the PCB that would flash with each pulse generated. This brought us to the idea to use an LDR as a source of some degree of indeterminacy, by creating an optical and leaky coupling between the LED and the LDR. LDR's -soon they may become obsolete parts, as they contain some cadmium- are notoriously slow reacting devices... We figured out that if we have a cloud of some hundreths of these pulses, given they are not synchronised and due to component spread, the result would be pretty random. Moreover, the circuits being relatively close to each other, would certainly pick up some of each others light. The first prototype circuit was as straightforward as it could be:

The formulas to be used for component selection are:

f = 1 / (R1 + 2R2) .C1. ln(2)

duty cycle: ta / tb

ta = ln(2).(R1+R2).C

tb = ln(2).R2.C

T = ta + tb

It will be clear that reducing the value of R2 will make the pulses shorter. When the LDR is dark, the pulse frequency will be slowest. If cost is an issue, the 47n capacitor on pin 5 can be dropped at the detriment of some stability. The minimum voltage for this circuit to work is 4.5V. If the Cmos version of the 555 chip is used (7555), it would work from 2V on, but in that case a white LED cannot be used. On such low voltages it should be red and the current limiting resistor should be lowered to 47 Ohms. Obviously, sound level will go down a lot when lowering the voltage, even if an 8 Ohm speaker is used. The speaker we used for this first prototype is a tiny Visaton K16 (Art.nr 2816, www.visation.de). When a lot of these circuits are made to work and sound together, we obtain a nearly random result, as the oscillators are not synchronised and their LDR's show up quite a spread in resistance value.

Here are some PCB versions (scale is 200%) we made for experimenting with component values and specifications:

Here are a bunch of etched and drilled prototype boards:

Although the circuit as shown performs quite well, Laura found the frequency under full light condition too high. So, we added a fixed resistor in series (82k) with the LDR and a parallel resistor (10M) as well, to limit the slowest pulse rates. Now the circuit became:

A new PCB to accomodate this circuit was drawn as well.

As cost was a major concern in this project, we tried to examine alternatives for the 16 mm 50 Ohm Visaton loudspeaker (ca. 4 Euro's a piece). Magnetic buzzers came up as they can be bought for less then 1 Euro. Their problem however is that they produce a definitly pitched sound. So, in order to make them produce pure spikes, the pulses generated by our circuit must be made so short that pitch becomes unnoticable. Hence the value for R2 in the circuit, taken as 150 Ohms. Something to watch out for: the output of the 555 chip is not short circuit protected! Make sure the current never exceeds 200mA. This is why we use a 50 Ohm speaker. If you want to try 8 Ohm loads, a capacitor in the 1uF to 4.7uF range can be placed in series with the speaker, or the LED can be placed in series with the speaker, thus saving even more on components:

Needless to say that we also considered solutions in the pure digital domain: we could design a circuit around an 8-pin Microchip PIC controller. Randomness is certainly easier to achieve, but costs will be quite a bit higher unless we could go to industrial quantities of say over 5000 pieces...

Evaluation table for the different prototypes build up and tested:

Nr. board C1 Rs Rp R2 Rl /Cl transducer remarks
1 breadboard         - 50 Ohm - 50mm speaker no LDR
2 PCB V1 1uF 1M - 10k - 50 Ohm Visaton K16 50 Ohm no LDR - soft
3 PCB V1 1uF 1M - 10k   Piezo disk 20mm too silent - needs resonator
4 PCB V1 1uF - 1.8M 10k   KPX-G1205B 5V/30mA 3.1kHz LDR - too fast -->musicbox
5 PCB V1 10uF - 10M 150   Multicomp MCKPI-G2210L-3672 LDR - high pitched
6 08.02.2017 10u 82k - 150 - KPX-G1212UB 12V/30mA 2.3kHz ok
7 07.02.2017 10u 82k - 150 82 KPMB-3008E 8 Ohm, 150mW ok - non pitched
8 08.02.2017 10u 82k - 92 - KPX-G1205B 5V/30mA 3.1kHz ok
9 08.02.2017 10u 82k - 102 - KPX-G0905B 5V 2.,7kHz ok
10 08.02.2017 10u 82k - 92 - KPX-G1205A 5V/30mA 2.3kHz ok
11 07.02.2017 10u 82k - 76 - 50 Ohm Visaton K16 50 Ohm ok soft
12 07.02.2017 10u 82k - 75   Piezo disk 20mm too silent
13 07.02.2017 10u 82k - 82 47 ABT408-RC magnetic single pulse ok - non pitched
14 07.02.2017 10u 82k - 68 2u2 ABT408-RC magnetic single pulse ok ++
15 08/02.2017 10u 91k - 82 - KPX-G1205A 5V/30mA 2.3kHz ok beetje lager dan nr.10
16 08.02.2017 10u 91k - 76   KPX-G1205A 5V/30mA 2.3kHz ok
17 14.02.2017 10u 82k - 150 82+LED KPMB-3008E 8 Ohm,150mW too soft
18 13.02.2017 10u 91k   68 1u5 ABT408-RC magnetic single pulse 19 pcb's made
19 08.02.2017 10u 82k - 82 2u2 EKULIT LSF15WS 112 15mm ok dry click
20 13.02.2017 10u 82k - 75 1u5 ABT408-RC ok

Here is a picture of a handfull of assembled test boards: After some experimenting, we concluded that we could save yet another component: the 10M resistor, in parallel with the LDR. The penalty is that the pulses will stop completely when it is completely dark. Seen the application, this is rather an advantage.

Here is a picture of prototypes 7 to 12: For the prototype 13 and 14 we used a non-self oscillating magnetic sounder (ABT408-RC). This required some changes in the circuit as the sounder is designed for operation on 3Vpp maximum. In prototype 13 we used a series resistor (47 Ohm) at the detriment of sound volume. Prototype 14 uses a capacitor on the output pin in series with the sounder, with much better and louder results. Here is the circuit: Here is a picture of prototype nr. 13 to 16:

Prototypes 17 and 18 came out to look like this: .Prototype 19 uses an 8 Ohm, 0.8W speaker (Ekulit LSF-15), as small as 15 mm in diameter and therewith even 1 mm smaller than the Visaton type. They are priced at 2.09 Euro a piece at Conrad.

Clearly the PCB's could become quite a bit smaller if we would have designed them using SMD components only. The prospect of handsoldering 200 boards with SMD's however was prohibiting.

It goes without saying that all these circuits can be used for a wealth of different purposes. If the timing capacitor is taken a factor 100 to 1000 smaller, they become audio oscillators. With the LDR it becomes very easy to bulld an optical theremin or a cracklebox, if you use your fingers -an even cheaper component- instead of the LDR...



Godfried-Willem Raes

Ghent, February 8th-14th, 2017

References and data sheets:

to index-page Godfried-Willem Raes