A Different Kind of DCO – CMOS PLL Synced Oscillator

I almost gave this post the title ‘A New Kind of DCO’, but new inventions using existing electronics don’t take place very often any more. Although I have never seen a schematic for an instrument that uses a similar method that I have used, I am sure that someone has tried it out. In fact, the CMOS chip used was pretty much created to do this exact function. I just tacked on an addition.

Lunetta Synthesizers

Before I get started, I will give a brief explanation of a Lunetta synthesizer. They are synthesizers that are created using simple logic chips. The chips are 4000 series CMOS integrated circuits. These chips operate with on or off pulses. When below the audio range, the pulses just sound like clicks. However, the pulses turn into a single sound when the frequency is in the audio range (20Hz-20,000Hz). This means that when operating in the audio range, these chips can be used as a sound source. This method of creating sounds using logic chips was made popular by an artist named Stanley Lunetta.

Learn more about Lunetta synthesizers

The Lunetta Holy Grail

As previously mentioned, Lunetta synths create sound using on and off pulses. This creates a square wave. While suitable for many situations, a square wave only contains odd harmonics and isn’t always desirable. A sawtooth wave is often preferred. However, there is no easy way to create a sawtooth wave from a square wave. There are some methods that create waves that are close to a saw, and there are very complex ways of doing it. The Lunetta Holy Grail is a simple circuit that can change a square to a saw.

The Goal

My goal was to create a method for converting a square wave to a sawtooth wave with minimal parts that was useful for Lunetta synths. Most Lunetta synths run on 5V, so the solution had to work with this power supply. This also makes it useful for use with an Arduino. I also wanted the solution to be adjustable to work on a dual rail power supply.

The Transistor Method

The most common method for converting a square to a saw is with a capacitor and a transistor. The incoming square wave is converted into a narrow pulse and sent to the base of a transistor. A capacitor is charged from a current source, and the transistor acts like a switch to discharge the capacitor on every pulse from the square wave.

Transistor square to saw conversion.

The transistor method makes a curved saw with an amplitude that decreases as the frequency increases.

This method makes a very sloppy and inconsistent sawtooth. A capacitor doesn’t charge linearly, so the saw shape is curved, especially at low frequencies. Also, because the current source always remains the same, the amplitude of the wave decreases as the frequency increases. This is because the capacitor is discharged before it has enough time to fully charge. I tried some methods of converting the frequency to voltage in an effort to compensate for the amplitude loss. It helped, but did not solve the problem.

The Standard DCO

Roland Juno-106 DCO

The DCO circuit in the Roland Juno-106.

What I consider the “standard” DCO is the type used in the Roland Juno synthesizers. It works very similar to the method previously described, but uses an opamp integrator to charge the capacitor linearly. This creates a straighter saw shape, but still has the amplitude problem. In the Juno synthesizers, a digital processor supplies the voltage that charges the capacitor. This control voltage changes with the frequency. I tried this method, but again I had difficulty creating a control voltage based on the frequency that kept the amplitude even across the frequency range.

The 4046 Phase Locked Loop

While trying to come up with a better solution for creating a voltage based upon the frequency for the Juno style DCO, I remembered that the 4046 CMOS chip does just that for its own internal oscillator. That is when the seemingly obvious idea came to me.

4046 PLL

4046 phase locked loop integrated circuit.

The intended purpose of the 4046 is to create a phase locked loop (PLL). The phase of an internal square wave oscillator is compared to the square wave from an external source. The phase comparator outputs a voltage that adjusts the internal oscillator until the two signals are synced.

My idea was to replace the internal oscillator in the 4046 with an external one.

The External PLL Oscillator

Any VCO would work, but for my experiment I had to find one that was simple and ran on a 5V supply. The LM324 opamp is a popular choice for Lunetta builders because it works well on a single rail power supply. I found an oscillator in the LM324 datasheet that creates both a triangle and square wave.

The plan was to convert a square wave to a saw, but even a triangle would do. A triangle can be converted into a saw, and this was just a test to prove that it could be done – the oscillator can be replaced with any other VCO. The only requirement is that there also needs to be a square wave output. Most oscillators do have a square wave output, but it is also very easy to create one using a comparator.

Putting It All Together

Once the oscillator was built and tested running independently, I sent the buffered source follower voltage out from the 4046 to the CV input of the external oscillator. Then I connected the square wave output from the external oscillator to the second phase comparator input on the 4046. The output from the LM324 was a little low, so I used a Schmitt trigger inverter to boost the square up to logic level. This also inverts the single, but that doesn’t effect the PLL at all.

CMOS PLL Synced Oscillator

CMOS PLL Synced Oscillator. Click the image to view the full sized schematic.

I now had a triangle VCO that was being synced to a square wave source signal by using a PLL.


Need to Test

This would be perfect to pair up with an Arduino as the square wave signal source. The Arduino has a Tone function that easily creates a logic level square wave tone. Leaving the waveshaping to analog circuitry frees up the processor to do other things (like creating LFO and envelope control voltages with the PWM outputs).
Dual Rail Power Supply
This should work even better with a dual rail power supply. The increased power supply range should improve the frequency range. The 4046 has a max supply range of 15-18V depending on the manufacturer. This means that some adjustments and level shifting may be required if used on a dual rail supply. I think that running the 4046 from 0-12V will work. Then the square wave output of the external oscillator can be shifted and brought up to the 12V logic level. This might be as simple as sending the signal into a diode and then sending the output to a comparator or inverter.
Expo Convertor
The range was just over 4 octaves, which isn’t bad considering it was on a 5V power supply. An expo convertor might improve that range, but also adds more complexity. It probably isn’t worth it for most Lunetta projects, but I might try it anyway.
Sawtooth VCO
The whole point of this experiment was to convert a square wave to a saw. I will need to try a sawtooth VCO. I will probably do this when I test the dual rail power supply. Check back in the future for that update.

Comments (4)

  1. CGamlin

    I’ve been thinking about the same problem recently (not for Lunetta, but for a different project), and found a web page entitled “How to get a sawtooth from a 4046 VCO”


    I’m in early days for my project, running LTSpice simulations to see if my ideas are viable. Unfortunately I don’t have a Spice model for the 4046 so, while I get why you can get a sawtooth using the method, I am unsure if the method solves the “increasing frequency decreases amplitude” problem.

    If you test this idea, I’d be very interested to know if this solves the problem or not.

  2. Jacob Watters (Post author)

    I talked a lot with Scott Gravenhorst (the guy who made that schematic) during this project. It solves the amplitude issue, but creates another. It adds a spike on the peak. It isn’t too bad and can be filtered out. I can’t remember, but I think I also had an issue with it leaving a bit of a gap between each saw tooth. Have a look at this page where a similar method is used:


  3. Matthias

    So basically, the pll creates a control voltage for the triangle vco from the square wave and the feedback of the vco? How fast does this work, don’t you hear any glides or something when you play some notes?

    1. Jacob Watters

      There is no audible glide. However, you can use a pot in the CV smoothing filter to create intentional portamento.


Leave a Reply to Jacob Watters Cancel reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Follow Me

Get all my latest updates by following and subscribing on these sites.

Bandcamp SoundCloud YouTube Facebook Twitter LinkedIn