Random The Random Sequencer is a circuit that produces

random control voltages. These can be completely

Sequencer random, or they can be locked into loops that repeat

every 8, 16 or 32 steps.
It was inspired by the long history of shift register

Documentation
PCB v2.1
pseudorandom synth circuits, including the Triadex
Muse, Buchla 266 Source of Uncertainly, Ken Stone’s
gated comparator and Grant Richter’s Noisering. This
September 2012 circuit includes a transistor noise generator, so should
by Tom Whitwell create true random binary sequences. However, the
structure of the sequencer and the nature of the
digital-to-analog converter used seem to give a
particular character - tones that rise and fall in a
particular way.

Block Diagram

How the module works and so on. What happens to step 16? It’s copied to
step 1, but may be transformed along the way.
This is a binary sequencer, based around a 16 bit
memory circuit called a shift register.
At any moment, the first 8 of the memory positions
are being read by the digital to analog converter.
NB: You cannot program this sequencer to play
00000000 = 0 = minimum voltage, around 0v
specific tunes. You cannot save sequences. You can
11111111 = 256 = maximum voltage, around 10v
never go back to a sequence that has changed.
10000011= 131 = around 5v
It’s designed as a sequencer that you can steer in one
With the main knob at 5pm and the length switch UP,
direction or another, not one that you can program
a 16 bit sequence will loop forever.
precisely.
With the main knob at 5pm and the length switch
DOWN the 16 bit sequence is split in half, with the 8th
Each clock pulse moves the binary sequence one step
step linked to the 1st step.
through the 16 bit memory. Step 1 is copied to step 2,
Things get interesting when the main knob is not at voltage set by a combination of the main knob and
5pm. The knob controls a randomised switch the CV input (summed by IC1A). Using a transistor
connecting the end of the loop to its start. inverter, two gates of the 4016 are set up as a SPDT
switch, routing the end of the shift register loop either
In a binary system, there are obviously just two straight back to the start of the loop, or back via
options. It can take the last bit and accurately write it another transistor inverter.
into the first bit. Or it can write inaccurately, flipping a
0 to a 1, or a 1 to a 0. Bill of Materials
Most of the components are straightforward off-the-
The knob (and the associated CV input) determine shelf parts, but the parallel PCB construction means
how likely each bit is to be flipped: some components are very specific.
At 3pm, maybe 1 bit in 10 will be flipped. The loop will Sockets: The board is designed for 3.5mm vertical
change gradually. jacks from http://erthenvar.com/ model PJ-301b. If
At 12 noon, 5 bits in 10 will be flipped. The loop is you’re not using these sockets, the lug nearest the top
filled with random data that never repeats. of the board carries the signal.
At 7pm, every bit will be flipped, and the loop
becomes a mobius loop. The DAC0800 is also known as a DAC-08. This part of
the circuit could be replaced by R-2R resistor ladder,
In effect, the sequence doubles in length. With the as seen in Ken Stone’s Gated Comparator.
length switch UP, 16 steps will play, then those 16
steps will be repeated, inverted. NB: This is a binary
inversion, not a melodic or CV inversion. The effect on
the output voltage will be hard to predict, but the loop
will repeat after 32 steps, not 16.

How the circuit works

The Clock input goes through an Op-Amp Comparator
to boost weak clocks. R9 and R3 are a voltage divider
to set the trigger voltage.

The two 4015 chips are dual 4-bit shift registers, so

there are 4 pins that receive the clock signals.

The output of each shift register is connected to the

input of the next via a 10K resistor with a 1n cap to
ground, to delay the signal. (This fix is one of many
elements drawn from Ken Stone’s gated comparator
circuit).

Eight shift register outputs are connected to the LEDs,

the optional Expansion port and the DAC inputs. The
LEDs are not buffered (Ken Stone uses transistor
buffers in his Gated Comparator, which is probably a
The LED Bar Graph module could easily be replaced
good idea, but it seems to work without them).
on this PCB by individual LEDs, which might be easier
to mount in a drilled panel - put the shorter LED legs
The DAC is connected according to the datasheet,
towards the edge of the board. The LED current
driving one gate of the TL074, with a simple
protection resistors can be adjusted according to the
attenuator pot on the output.  
LEDs used.
The noise circuit is a basic reverse-bias transistor
I’ve used an eight way 330 ohm resistor network
noise circuit, boosted by inverting op amp stages with
(RN1) for ease of installation, but it should be
capacitors to remove the offset. The front panel noise
possible to hack one together - 8 resistors connected
output is not independently buffered, so it’s possible
to one wire going to ground at the pin connected with
using this output could have an impact on the rest of
a cross (a neat example here: http://flic.kr/p/
the circuit in some situations.  
65UmMd ).
The switch part of the circuit works like this: White
noise flows into a comparator, with the reference
R19 is there to increasing the total resistance of RN1 Expansion Port
(Without it, the circuit draws rather high current, but The expansion port surfaces the clock pulses, and the
you could use a jumper if you have a higher value of 8 shift register outputs that go to the DAC. It is not
RN1). The Clock LED has it’s own current limiting buffered.
resistor, R13.

I used 1% resistors in many places, because I had

them handy. I do not think any of the component
values are very critical - nothing needs to be trimmed
or calibrated.

Potentiometers: Both pots are voltage dividers, any Pinout:

reasonable value should work. I’ve used 50k & 100k. 1-8: Data channels from the shift register. 1 = Most
significant bit, 8 = Least significant bit
The PCB is designed for 9mm vertical pots from 9: Ground
Alpha, Bourns or other manufacturers like this: 10: Clock

Building the circuit

I’ve found the 4015 shift register chips to be slightly
unpredictable. If the sequence won’t loop properly,
swap the chip and throw out the old one. I’ve no idea
why this happens, but Navs reported the same thing
building his Bitsy. This is why the two 4015 chips are
on the back of the board, not hidden behind the
panel.

Quirks
Feeding the module from a multed clock input can
sometimes give strange results - clearing a looped
sequence. I’ve no idea why this happens.

PCB Versions
PCBs versions are marked in the silkscreen

I used one with a 25mm D shaft for the main control, v1 Personal prototype, not released
and a 12mm knurled for the smaller control.
v2 Original shared PCB design, May 2012. Includes
Switches: These red-bodied solder tag switches are incorrect polarity of C6, Please read the addendum at
pretty common. Two ranges are Salecom T80-T http://musicthing.co.uk/modular/?p=211 before
switches from Rapid or Bitsbox, or E-Switch 100 range building this board.
from Mouser.
The LENGTH switch is a standard SPDT ON/ON v2.1 September 2012. Fixed the polarity of C6, and
switch. removed a couple of spurious rectangular milling
The WRITE switch MUST be centre off SPDT - and instructions from the gerber files.
ideally a momentary (ON) OFF (ON).
Hardware, pots, switches, lights:
1 5+5 pin 0.1 inch Power Header (Mouser 649-69190-410HLF)
1 10 pin 0.1 inch header (Mouser 649-67997-210HLF)
1 Length Switch: SPDT ON-NONE-ON (Mouser 633-M201202)
1 Write Switch: SPDT (ON)-OFF-(ON) (Mouser 633-M201802)
2 50k ALPHA-9MM vertical Pot CHANGE, SCALE
(Main ‘change‘ control: 25mm D Shaft Mouser 652-PTV09A-4025FB503 Smaller ‘scale’
control: 12mm knurled shaft Mouser 317-2090F-50K )
4 PJ-301B 3.5mm socket (from erthenvar.com)
1 DC10 Kingbright DC10 LED Bar Graph (Mouser 604-DC-10EWA)
3 16 pin DIL socket (.3 inch)
2 14 pin DIL socket (.3 inch)
1 8 pin DIL socket (.3 inch)

Capacitors:
2 0.47uf Polybox Cap C10, C11
4 1n Ceramic Cap C1, C2, C3, C4
1 10n Ceramic Cap C9
2 10uf Electrolytic Cap C5, C6
5 100n Ceramic Cap C7, C8, C12, C13, C14

Resistors:
2 1k Resistor R17, R18
1 6.8k Resistor R26
12 10k Resistor R1, R8, R14, R15, R16, R23,
R24, R27, R28, R30, R32, R35
2 15k Resistor R2, R9
1 51k Resistor R5
7 100k Resistor R3, R4, R10, R12, R22,
R25, R29
1 150k Resistor R6
1 330R 8-commoned Resistor Network RN1
2 470k Resistor R31, R33
1 560R Resistor R19
1 820R Resistor R13

Semiconductors:
1 1N4148 1N4148 Diode D2
2 4015 Shift Register IC2, IC3
1 4016 Bilateral Switch IC5
1 DAC0800 Digital/Analog Conv U$2
1 TL072 TL072 IC1
1 TL074 TL074 IC4
3 2N3904 Transistor: 2N3904 T1, T2, T3

Full Mouser Project: http://bit.ly/MRGqRF access id: bc8a4a43bf

The top of the board...
... where most of the components go. In this
picture, the component values are missing from
the silkscreen, but they should be there if you
make a board from the Gerber files

Start with the resistors

I usually check each one with a multimeter,
because I can’t read the coloured bands

Add the capacitors.

C5 doesn’t fit very well
IC Sockets and Transistors on the
top of the board

The back of the board

The expansion header is optional -but you won’t be
able to add it later, as it fits under the LED.

LED Bar Graph

One corner of the LED bar graph module is
flattened - make sure you align this with the
silkscreen. You might want to lift the module up a
little, so it’s closer to the acrylic panel, but it’s
certainly not necessary.
The big bits Soldering complete.
Add the switches, pots and sockets. The sockets At this stage you may want to power up the module
are fiddly - try to hold them in place, tack some and use a multimeter to check the expected +12v,
solder on, then add plenty more solder to keep -12v, GND connections to the IC sockets, according
them secure to the data sheets.
Once you’re happy, insert the ICs. They go, from the
LED down: DAC0800, 4016, TL074, TL072, with
the two 4015s on the other side. The 4016 and
TL074 are rather snug, but they do fit in.
Open Source Hardware
This is an open hardware project. You are free to build this
hardware, sell it or make any changes to it, so long as you
share your changes and release them under the same license.
You should also credit me, Tom Whitwell. All of the original
design files (Eagle CAD, Illustrator etc) are shared. These files
are licensed under a Creative Commons Attribution Share-Alike
license, which allows for both personal and commercial
derivative works, as long as they credit me and release your
own interpretation under the same license.

This work is licensed under the Creative Commons

Attribution-ShareAlike 3.0 Unported License. To view a
copy of this license, visit http://creativecommons.org/
licenses/by-sa/3.0/ or send a letter to Creative
Commons, 444 Castro Street, Suite 900, Mountain
View, California, 94041, USA.