Sequence Dancing


Following on from Atari Punx…, which was my first foray into the world of electronic sound generation, a sequencer is a very useful piece of additional kit. One can be easily made using a CD4017 decade counter (or CD4022 octal counter)… However, other ICs can be used, to provide additional functionality. For example, a CD4510/CD4516 BCD/Binary counter and a decoder or 16 channel multiplexer, such as a CD4028 or CD4067, can add up/down sequencing, pre-loading and other features.

Let’s take a look at all of these ICs as well as some others.

VCO, VCF and VCA circuits are also touched upon, be are covered more fully in Banging noises…

See also


Terence Thomas on Robots

Robot Brains Circuit and Theory

A brief introduction to sequencers

From Logic Express 9

How Step Sequencers Work

The basic idea behind analog step sequencers is to set up a progression of control voltages and output these step-by-step, typically in an endlessly repeating pattern. This basic principle helped to spawn a number of electronic music styles based on the mesmerizing effect that repeating patterns can have.

In early analog sequencers, three control voltages were usually created per step in order to drive different parameters. The most common usage was control of a sound’s pitch, amplitude, and timbre (cutoff) per step.

The control surface of analog sequencers often contained three rows of knobs or switches aligned on top of (or beside) each other. Each row commonly contained 8 or 16 steps. Each row provided a control voltage output that was connected to a control input (for a particular parameter) on a synthesizer. A trigger pulse determined the tempo between steps. A running light (an LED) indicated the step that was currently being triggered.

The running light programming concept also appeared in later drum computers, the most well-known examples being the Roland TR series drum machines.

The introduction of the MIDI standard and increased use of personal computers for music led to a rapid decline in the step sequencer and related technology. More flexible recording and arranging concepts that didn’t adhere to the step and pattern principle were possible with the far more powerful personal computer.

Despite these technological advances, step sequencers haven’t disappeared completely. Hardware groove boxes have experienced a renaissance in recent years due to their intuitive nature, which has made them a favored tool for rhythm programming.

So, by definition, a sequencer should have:

  • 8/16/32 steps
  • Each step is switchable
  • Each step should provide three CV outputs: pitch, tone (or timbre) and amplitude.  Page 59 of Sound Synthesis by Terence Thomas shows how three different parameters – pitch, tone (or timbre) and amplitude – can be controlled, by a signal generator (or Voltage Controlled Oscillator (VCO)), Voltage Controlled Filters (VCF), and Voltage Controlled Amplifiers (VCA) simultaneously,

    Multiparameter Sequencer Control block diagram - Thomas - Sequencer and three CV outputs
    Multiparameter Sequencer Control block diagram – Thomas – Sequencer and three CV outputs
  • A variable trigger pulse
  • An indication of which step is being played, usually a “waterfall” of LEDs
  • Repetetion

From SparkPunk Sequencer Theory and Applications Guide,

Raymond Scott is largely recognized as the inventor of the step sequencer, which he called the “circle machine.” You can hear some examples of it in these videos, and learn more about the man and his music at his website.

How many steps?

As a bare minimum 8


Ideally 16, from What is a step sequencer and how do I use it?

For example, if you break up a 4 bar loop that is in 4/4 time, it will have 16 beats, or steps. You can then use the step sequencer to enter the bass drum on the steps you want it to play, and enter the snare on the steps you want it to play. Then you start the step sequencer and it will play those back at the right time to give you your beat. It’s basically a way to map out exactly where each drum hit should land in relation to the time of the beat.

Having 32 just give allows for greater flexibilty, and less repetition


However, it should be noted that, from What is a step sequencer and how do I use it?

where SS falls down is if you find it hard to judge musical time, or if you wanna do complex poly-rhythmic stuff inside a bar – say 8ths for the first two beats and 6ths for the next two beats. for that stuff, i tend to play on keys then edit in PRV.

See also 32 step sequencers/drum machines vs 16 step – 8 or 16 step sequencer can become monotonous and repetitive, where as 32 allows for greater musical sequences. Also, the ability to set timing division for 3/4, 5/4 or 6/8, rather than the standard 4/4, is a good idea.

Sourcing the parts

One could just purchase a bunch of CD4017 decade counters, or some other counter from the 4000 series, here is a list, or obtain a simple kit for around a dollar:

  • NE555+CD4017 LED Light Water DIY Kits Electronic suite Water lamp module Red
  • There is an SMD version, but this is not so easily hackable for use with an APC

    Decade sequencer from eBay - SMD
    Decade sequencer from eBay – SMD
  • The CD4017 decade counter
    • NOTE: If you are going to be cascading the CD4017 decade counters, to make greater than an 8 step sequencer, then some NAND or AND, logic gate ICs are also required, CD4011 or CD4081 respectively (or CD4572 Hex gate: quad NOT, single NAND, single NOR). I would recommend getting NAND, or NOR, gates, as these are universal gates which can be used to make any subsequent gate (AND, NOT, OR, NOR, XOR, etc). From CD4017 ic 16step seq troubleshooting…need help, it can be seen that an AND gate is required. This is the manufacturer’s recommended way of cascading two CD4017 decade counters:
      16 steps
      16 steps

      Take note of this slightly different schematic for cascading three (or more):

      Counter cascading

  • You could use 74HC(T) series too, but beware (see CD4017 ic 16step seq troubleshooting…need help):

74HC…. logic is only rated for 5V operation (5.5V max). You *CANNOT* mix it with 4000 series logic running from a 9V supply. Its only safe to directly interconnect 4000 series and 74HC…. if they are BOTH running from the same supply of under 5.5V.


To be picky the 74HC series will work from 2V to 6V. I think you’re thinking of the 74HCT series is 4.5V to 5.5V. Either way, you’re right that the 74HC11 and CD4017 both need to be running on the same power supply between 3V and 6V.

You should use a 5V regulator. A low drop-out regulator will give plenty of battery life from a 9V battery. You could try the LM78L05 but the output voltage will fall below 5V when there’s still plenty of energy left in the battery. This may not be a problem, as the circuit will still work below 5V.


From this post, on 16 Step 4017 sequencer?, this describes how the 4017 cascade using AND gates operates:

Once the last output of the first 4017 goes high it enables the EN-pin (13), thus stopping the first 4017; at the same time – through the 4081 AND-gate – it makes the second 4017 start counting (because now the clock signal can come through). Once the last output of the second 4017 goes high it resets the first 4017, so this one starts again at pin 3, while the second 4017 is stopped again as the last output of the first 4017 goes low (thus disconnecting the second 4017 from the clock).


The original sequencer Baby8 article can  be found here, DIY 8 Step Sequencer (original image):

Original article on the baby10 - Captain's Analog
Original article on the baby10 – Captain’s Analog

8 step

From Atari Punk Sequencer: Breadboarding and Planning, this schematic shows a fantastic switchable 1-8 step sequencer,

Atari Punk Sequencer (with 4017)
Atari Punk Sequencer (with 4017)

See also Atari punk console + Baby sequencer 4017 IC, for a pretty messy 8 step schematic.

From CGS resetty clock circuit?

Unfortunately, the link of the image above has broken/died/rotted. However, given the context of the message, the CGS Gate Sequencer may be this Ken Stone schematic (original):


Which is (apparently) obsolete, and replaced by this (original (there may be another source?)), from Gate Sequencer for music synthesizers:


Also referenced in CGS resetty clock circuit? is the CGS Pulse Divider (which again in the thread is a dead link), but here is another instance (original):


Another Ken Stone schematic is the CGS Gate Sequencer CV adapter (original):


Another interesting design uses CD4016/CD4066 quad bilateral analogue switches, again from simple 8 step sequecer for SL, from this post (original image):

Simple 8 step sequencer for Sound Lab
Simple 8 step sequencer for Sound Lab

And this modification

Sequencer to SL gate input
Sequencer to SL gate input

10 step

From Baby 10 – Analogue sequencer, image and article

Sequencer - 10 step
Sequencer – 10 step

Baby 10 step sequencer help! for this nice, yet simple/basic, schematic of a sequencer (along with an APC)


From simple 8 step sequecer for SL, this post is the revision 2 of the Baby 10, and contains some elements of Terence Thomas’s sequencer, including manual step and reset, as well as 2 CV outputs, gate outputs .


and here is the associated oscillator

Here is revision 2.1, which is three  years later, from Fonik’s BABY_10 mini sequencer. It has an improved reset circuit (PDF link)

Baby10 - version 2.1
Baby10 – version 2.1

8 step – Schmitt trigger NAND implementation

From CMOS 4017 Based 8 Step Sequencer, uses one XOR gate and four Schmitt trigger NAND gates (CD4093), plus a NAND gate for the oscillator. The single XOR gate could be replaced by a NAND gate IC, using all four gates. Either way, three ICs are required.

See also diyaudiocircuits/eagle-cad-files, although the files seem a bit incomplete or missing.

The clock schematic, implemented using a NAND oscillator is not shown on the page, linked to above, but on the same site a, presumably, similar oscillator, or rather two, can be seen, on Atari Punk Junk Console, which uses CD4093 Schmitt trigger NAND gates:

Schmitt trigger NAND gate oscillators
Schmitt trigger NAND gate oscillators

Errors with this circuit

  1. Note that if the pot is turned “down” fully then there will be zero resistance between the capacitor and the output of the gate and that it would be better to include a minimum resistance for charging the capacitor, say 100 Ω.
  2. The Hold circuit, using an active low input to the HOLD NAND gate, will not work as shown, and will actually inhibit the clock unless the hold push button, or switch, is pressed, or closed, respectively.  Therefore the logic logic is inverted. An additional NAND is required to invert the output of the Hold circuit. Better still, without requiring an additional NAND, invert the active input, by using an active high input to the HOLD NAND gate (parallel the resistor and capacitor pair and switches/buttons go to VCC) as shown below (taken from When to use pull-down vs pull-up resistors):mg5fz
  3. In addition, with the original circuit, when the hold button is released and the input to the HOLD NAND rises, an additional (rising edged) clock signal is generated, even if the clock is low. So, the HOLD does not pause on the current step, but rather the next step. It would be better to apply HOLD to the Clock inhibit input – after all that is what the input is for.
  4. The best method, with the original circuit, is to have the output of the HOLD NAND gate go to a three input XOR gate, then the rising edge effect of the HOLD circuit is avoided (check STEP function). Ignore this as it only works when STEP is held active LOW, and step doesn’t actually work.

16 step

It is possible to chain two CD4017 ICs together to make a 16 step sequencer. Again there are a number of designs out there:

From 16 step baby sequencer improvement- help needed-> schemat, one of the simplest designs, a non-switchable 16 step sequencer,


and the same circuit but with an oscillator and a reset circuit (see also the original Instructables for the schematic below at 16-Stage Decade Counter Chain – Using Two 4017 Chips)


Again, from 16 step baby sequencer improvement- help needed-> schemat, the same circuit, and is switchable 2×8 or 16 steps (although it may suffer from skipping step one in 2 x 8 mode)


The resistors are there because you cannot leave a CMOS input floating so it is tied low until pinged momentarily high with a pulse through the caps.


Add a rotary multiway switch on each set of 4017 outputs and as Daverj says, you can have any bar length you fancy. Alternating 5/8 – 7/8 time. 13/8. Do a beat count of Tubercular Balls. Loads of music has been in odd time signatures and some of it remarkably popular, but not so much in recent years since the doof-doof-doof-doof groove set in and people forgot they could count up to anything else but four.

Any bar length you can think of can be broken down into sets of two and three beats for some interesting variations.

To shorten the beat:

Do not wire Q1 as 1st step and Q0 as 8th step, so when external reset is applied, it jumps to Q0, it all gets out of sync if you do. The ninth step either immediately jumps to and outputs the first step in 2×8, or it mothballs itself with no outputs and resets the alternative counter in 1×16.

If you want a shorter number of beats, then make a step earlier than the 9th do the resetting. An 8 way switch on each 4017 (in other words 8 of the positions of a 12 way switch).

However, this switchable circuit has some issues, apparently it skips step 1 in 2 x 8 mode, from Yet another sequencer ((edit: answered) question) thread (from whence the above schematic actually came), specifically this post

Dukester (who has yet to make an appearance on the forum) send me a message that he tested the switchable version I posted and although the 1×16 mode worked the 2×8 mode skips step 1. I did a test myself and that is indeed the case.

I think this actually makes sense if you look at the internal structure of the 4017. The CLK INH pin is internally connected to an AND port (after being inverted) together with the CLK input. If you look at the circuit in 2×8 mode then up untill the 8th step (Q7) it works fine. When Q8 goes high two things happen:
The CLK INH pin goes high which disables the CLK input but it also gets immediately reset. Because of this the CLK INH pin goes low again. However because the CLK IN is still high this high to low transistion of the CLK INH pin creates a CLK pulse through the internal AND gate. The result is that it skips step 1. I am slightly confused why it works fine in 1×16 mode.

and there are three variations on the theme:

I did some tests to see if I could come up with a solution that wasn’t too complex and I did manage to get something working using a 3-pole switch (V2.0). It doesn’t always work well when you switch to 1×16 mode though.

Cascaded 4017 - Three pole switch
Cascaded 4017 – v2.0 – Three pole switch

Because I didn’t like the idea of using a 3-pole switch I tested if I could replace it with a mux so that only a single pole switch could be used (V2.1). This worked and I managed to use the same switch to send a reset pulse to the first counter which makes the 1×16 mode work every time. Dukester has also tested this version and confirmed that it works. Very Happy

Cascaded 4017 - Mux
Cascaded 4017 – v2.1 – Mux

So I was going to post it (actually I am doing that now or I did when you read this) and I looked up the thread where I noticed
the schematic by DUBmatze. It doesn’t work how it is drawn but I was curious what it would need to get it functioning. It first got a bit more complex needing more diodes and a 3-pole switch but when I compared that to version I drew I noticed it could be simplified. When I drew V2.0 I did had the suspicion that there was an easier way but just didn’t see it, so thanks DUBmatze!
The end result is a version (V2.2) that needs a 2-pole switch and no mux and seems to work every time, at least it does on my breadboard. (you could probably replace the switch with a mux).

Cascaded 4017 - Two pole switch
Cascaded 4017 – v2.2 – Two pole switch

Personally I would go for a different approach; 2 seperate sequencers that have individual CLK and reset inputs and a switch to select the number of steps with an option to chain them. I think it would be possible to make it so you can chain more than 1. I do recall drawing something like that once before so I will look into it or maybe come up with a new version.

One can incorporate an AND gate (CD4081 AND or CD4011 NAND), from the MCP14017 data sheet, and this is the usual. and manufacturer’s recommended) method of cascading

Counter cascading

A similar method (except that the “feed forward” signal goes to the next CD4017’s ENABLE pin, rather than the CLOCK) using inverters, from 16 Step 4017 sequencer?, requires an additional hex inverter (CD7474 or CD4069)

From 18 Stage LED Sequencer, this circuit uses two diodes and a 15 k resistor to form an AND gate


When power is applied, the 15K resistor and 10uF cap at pin 15 will reset the counters to the zero count where pin 3 is at +12 and all other outputs are at zero. The 2 diodes (1n914) and 15 resistor form a AND gate so the clock pulse will be passed to the right side counter when the sequence starts. When the right counter reaches the 10th count, pin 11 will move high enabling the AND gate on the right to pass the clock pulse to the left side counter. As the left side counter advances, pin 3 will be low so that clock pulses cannot advance the right counter. When the left counter turns over and pin 3 again moves high, the sequence will repeat. Thus we get 18 total counts, 9 from the first counter, and 9 from the second.

Note that the 4017 counter will not deliver much current, and so the LED current is set to about 6mA using a 1.5K resistor in series. For more current, you could use transistors on each output as shown in the drawing above, (10 Channel LED Sequencer). But some of the newer bright LEDs are fairly bright at 6mA.

From CD4017 ic 16step seq troubleshooting…need help

CMOS 16 step sequencer - EEVBlog
CMOS 16 step sequencer – EEVBlog

From  simple  8 step sequecer for SL, this post, uses 4016 bilateral analogue switches


Also worth of note are two of Osamu Hoshuyama‘s designs (from Unusual Synthesizer Circuits, which I discovered from the 16 Step 4017 sequencer? thread)

First, from 16 Step Sequencer Using CD4017s, original image:


Secondly, from Basic 16-Step Analog Sequencer, original image:


A neat veroboard layout from Baby-8 Sequencer/LED driver (original image)


Three CV outputs

As this sequencer is predominately designed for use with a Atari Punk Console, how do the two go together? As stated above, a typical sequencer has three outputs, for pitch, amplitude (which, here, we will interpret as duty cycle, or pulse width), and timbre (or tone – according to Page 59 of Sound Synthesis by Terence Thomas). This image shows how three different parameters – pitch, tone (or timbre) and amplitude – can be controlled, by a signal generator (Voltage Controlled Oscillator (VCO)), Voltage Controlled Filters (VCF), and Voltage Controlled Amplifiers (VCA) simultaneously, using the three CV outputs of a sequencer:

Multiparameter Sequencer Control block diagram - Thomas - Sequencer and three CV outputs
Multiparameter Sequencer Control block diagram – Thomas – Sequencer and three CV outputs

These three stages are examined in more detail in the article, Banging noises…, in the section A closer look at Voltage Controlled Circuits.

Using ICs other than a 4017


The 4022 is the same pin for pin as the 4017, except that it is an octal, rather than a decade, counter, so you could still have an eight step sequencer, see simple 8 step sequecer for SL.

However, it is not possible to directly replace the 4017 for a 4022, as 4017 based circuits trigger on D9, whicb the 4022 obviously does not have, see 4022 Sequencer.

Also, 4017 vs. 4022 for simple sequencers?

Using BCD instead of decade counters

Using a simple 4017 decade counter is fine, however, it does not offer much flexibility. If you want to reverse the sequence or use a specific sequence, then Binary Coded Decimal (BCD) counters could be the way to go.

From the very long thread in simple 8 step sequecer for SL, using a 4516 (Presettable Binary Up/Down Counter) and two 4028 (BCD to Decimal decoder). Up and down is also possible, as is pre-loaded sequences. See this post:


This is the core of the ARP 1601 sequencer and what I’m using for my current 32 step sequencer project. I’ve changed it a bit – ARP didn’t use an up/down counter. It works like a bat out of hell. It’s not much more complex than what you have and offers some flexibility in what the sequencer can do, with little added circuitry.

Using the U/D counter will allow you to step up or down in your sequence. You could put in a switch that directed your ‘reset’ command, which is used to set the number of steps, to instead change the direction of the sequence. That’s where another stage of the Schmitt trigger (hell, you have five left over) would come in handy, in addition to a flip flop.

The CD4516 would let you pull a lot of stunts (preload, reverse count, etc), but of course you cold just use a simple up counter, too. I’ve found this general setup to be a lot easier than the CD4017 route.

If you set up a switch to disallow the inversion on the ‘D’ input of the second CD4028, then badda-bim, badda bam, you have two 8 step rows in parallel.

Note – to get the two 8 steps in parallel, you’d actually want to prevent both CD4028 input D’s from going high – a simple enough matter. You could do it with a simple switch and MML logic using the same 74C14/CD40106 package that the selection toggle resides in. Just diodes and a couple of resistors configuring up some AND gates.

Preload is something to look at too – makes a good way to randomize a sequence with additional circuitry.

Having the ability to parallel the two decoders (creating two 8 step sequencers) would come in handy in a number of apps.

The first would obviously be that you could have two separate notes per step.

The second would be that you could have a preset cutoff for a filter per preset note. Reference Tomita’s rendition of “Golliwog’s Cakewalk” from “Snowflakes Are Dancing”.

The third would be that you could accurately tune an operator and carrier VCO for tuned FM sounds (bells, etc.) – this can be problematic otherwise. It’s fairly difficult to get linear tuning of FM with analog synths for a single CV driving the operator and carrier. That takes a high degree of accuracy on the VCO’s. being able to tune the differences out per step makes it a very doable thing.

Of course, two separate decoders would not be necessary for this. One could just as easily rig two or more outputs per step with any number of steps by sending each decoder output to two or more tuning pots and having separate CV output busses. Peter Grenader’s “Milton” sequencer does this with 16 steps and four separate tunable CV’s per step.

Preload could be used for a couple of things.

Probably the easiest use of it would be to enable a random step selection function to a sequencer. Say you had an additional clock that was running 20 or 25 kHz, feeding the preload inputs. In this mode, the clock pulse would be directed to the Preload input rather than the clock input (this could be done with some logic switching). When the clock went high, it would inhibit the fast counter so that a whatever count it was at would be present on the preload inputs. The preload going high would then load that value into the counter, and that would put your step sequencer to whatever step was represented by those four bits. Thing is, you’d want to put a little propagation delay in so that the fast counter froze slightly ahead of when preload went high, so the data would not be moving when the 4516 wanted to load the bits (not a terribly hard thing to do).

ARP used something like this with latches and a fast counter on the 1601. They further randomized things by playing with the speed of the fast counter a bit. I think even without that additional manuver, you’d still get a pretty decent random step selection – the randomization would be a function of the relationship of your sequencer clock and the fast counter clock, rather like the relationship of a the frequency of a S&H clock to the frequency of the sample source.

Not saying you want to do it on this particular sequencer, but that’s one thing preload would be good for.

From this post, using a 4516 (Presettable Binary Up/Down Counter) and a 4067 (16-channel analog multiplexer/demultiplexer)


Here’s another way to get a 16 step sequence with relatively few components. The CD4067 16X1 mux seems well suited for it. A bloody big IC, though – it’s a ‘double wide’.

In the schematic, I’ve got a dual comparator set up – one section handles the up/down switch, and the other section allow you to use an LFO (preferably sharp edged) from the SL to clock it. It’ll accept a bipolar LFO.

Of course, missing is a manual step, and also a manual reset would be nice. Both of these come in handy when programming the steps. That would be fairly easy to put in with another dual comparator, though I’d have to play with it to see if diodes would work suitably as an OR for the clock and step inputs, or if it would be better to resort to an IC (which means, um, another IC would be needed). A debounced momentary switch would be good for both a step and reset function. Check out how Ken Stone handled momentary switches on his Pulse Divider/Boolean Logic module here:

That’s where I lifted the comparator scheme anyway Very Happy . I’ve built the Pulse Divider, works pretty well (and comes in handy for sequencing, too).

The same thing could be applied to the dual decoder schematic I previously uploaded. That configuration would make it possible to switch from a 16X1 sequencer to an 8X2 sequencer. If you don’t have CD4028’s, then a pair of CD4051’s could be made to do the same thing by feeding +V into the input (as done with the CD4067 on this schemo).

From the amazing 16 Step Sequencer (+/-9V to +/-15V) by Ray Wilson, using a 40913 (Synchronous 4-Bit Up/Down Binary Counter)

and 4514 (4-to-16 line decoder/demultiplexer)


and 4094 (8-stage shift-and-store bus register)


Also, equally amazing, 16 Step Sequencer Auxiliary GATE/LED Driver PCB
 by Ray Wilson, a similar circuit with a 4514 (4-to-16 line decoder/demultiplexer)

and a simplified schematic

A couple of other links above, mention alternative ICs:

The approach with CD4514 & 4516 looks much better I’d say.

By which, the poster is referring to the second post in that same thread

Also,i wanted in the past to do sth like 2×8 or 1×16 and fonik suggested 4516/4514 combo ,

Which, I believe, refers to this, 4516 / 4514 based CMOS Sequencer, which is the Superseque by Thomas Henry, which features up and down, hold, single step…

Superseque - Thomas Henry
Superseque – Thomas Henry

Note that, as shown, pin 1 of 4514  (the input) is tied to Vcc make the outputs act as gates. However, if the clock (or pulse) is fed to pin1 then the outputs could act as triggers.

I still would use two 74HC595s, and cascade them after the column ‘595s. I mentioned one reason in my other answer: the ‘595 is cheaper, and you won’t need the AND gate.


If you are going to use a bunch of chips, you might as well use a binary counter and a 4067 multiplexer.

Auduino Sequencer

Following on from the Auduino, see Build an Auduino Step Sequencer


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s