Thursday, 7 November 2019

Synthesizer Build part-2: THE VCO

A word in advance: this article deals with the first VCO I've ever built and is based on the datasheet schematic combined with the LookMumNoComputer lay-out for the CEM or AS3340 chip. I personally had great trouble getting this VCO tuned over a wide range of octaves.  I could also never get really deep notes from this design. I have since found a much better VCO design so if you want to build a simple but excellent working and tunable VCO on stripboard I refer you to Synthesizer Build part-18: A Really Good VCO design.

I'm keeping this article up for my own archive and as a warning for those just starting out not to build this VCO!!!
Here's the original text for the first VCO build:
After having constructed the power supply and the power bus system it is time to move on to the next step. The Voltage Controlled Oscillator. I'm not going to go into details as to how it works etc. There's plenty info online about that.  In order to make this a complete build, not just the circuit board I needed something to mount the knobs and in- and outputs on. So I ordered a sheet of Aluminium, 200 x 1000 X 1.5 mm and powder coated gray/black on one side. That is fantastic stuff to make panels out off and I highly recommend it. You can saw off panels of the right width using an electric jigsaw with a fine toothed metal saw. Make sure you guide the saw with a straight piece of wood or metal to get nice straight panels.

For my VCO I chose the AS3340 chip which is a complete 1 volt per Octave VCO in a chip. It's a clone of the CEM3340 which were used in the 80's in synths like the Prophet 5 the Roland SH101 and many others.
The VCO we're building here will have almost all the options that the AS3340 chip has to offer and those we didn't include are not worth having anyway ;)
The schematic I used is pretty much just the schematic that comes on the datasheet.

This is the one I used:

I used the layout made by Sam Battle, from LookMumNoComputer and did a few enhancements on it. (Look to the one on the right).

For one, his layout is meant for the CEM3340 which uses a 10K pulldown resistor on pin 4, the squarewave output. For the AS chip, that's supposed to be a 51K resistor although I'm reliably informed it doesn't make a blind bit of difference what you use here. There was also a mistake in his design, namely the 10K resistor in the bottom left near the TL072. It is switched in parallel with the 10K on pin 4 making the overall resistance 5K. Just leave the bottom 10 K resistor out.
The 10K trimmer potmeter at the top left of the 3340 needs to be a multiple turn potmeter so you can set it very accurately.

Sam's layout doesn't include the High Frequency Tracking but you really need to include it in your VCO. I first build it without and at first it seemed to work fine but after having completed the whole synthesizer I couldn't get really deep bass tones out of it. That is until I included the High Frequency Tracking. Seems a paradox that something meant for High Frequency adjustments can have so much influence on the bass notes but if you look at the schematics you can see that it pulls the CV voltage on pin 15 down to ground a bit through the 20K potmeter. I kept out the 360K resistor between +15V and CV input because that kicks the VCO into really high notes. I don't know why that resistor is there but it really screws up the frequencies. I left it out but maybe I should have experimented further with that resistor in place. Anyway...
The HF adjustment pot only adjusts about half a note over its full throw so when you first test it it might look as though it doesn't work but it does when you start tuning the higher octaves of the VCO.

Furthermore I gave the buffer for the triangle wave a gain of 2 by adding two 100K resistors to the TL072. That gets the level of the triangle output up to 10V peak-to-peak, in line with the output voltages of the other two waveforms. Btw, you can use any resistor value between 50K and 1M for this purpose as long as both resistors have the same value.

I also added the Positive and Negative Hard Sync options from the Digisound 80 Modular design so that's also available on this VCO.

Here is the layout that I drew and used:

So there we have it. It's become quite a comprehensive VCO with lots of options.
I added a 100K resistor to the +15V input of the Pulse Width Modulation potmeter to get it to work over the complete throw of the potmeter and I added a switch to have the ability to decouple it from the PW Control Voltage if you have PWM controlled by an LFO for instance. You don't have to decouple it but the option is there.

I tested the finished print and everything worked as expected but there was a funny quirck in the squarewave output. Below 1.35kHz there was a strange triangular wave ringing on the downward slope of the square-wave. Here's some pictures of that from my scope:

I opened a discussion about this in the Synth DIY Facebook Group and there were many suggestions but I still haven't figured out the cause. It's not a de-coupling issue anyway.
I suspect that leaving out the High Frequency Tracking I mentioned earlier may be the cause. (Note: I did some more tests and it turns out that it does have a big influence on this issue. Including HF Track with the 360K resistor to +15V almost gets rid of the problem but on low frequencies there still is a bit of ringing on the downward slope but not nearly as much as now.
But as I mentioned before, the 360K resistor really screws up the frequency response so it can not be included. I have heared that there might be batches of chips that have this fault, so it might be the chip. I don't know and don't really care because you don't hear it and everything works fine.

[Edit: In the second VCO I built and now use (see article 18) this ringing is still there but it is much less then in this design. The new VCO has at most 3 spikes in the downward slope of the squarewave. Anyway it has proven to be not a problem what so ever. You can't hear it and it doesn't affect the working of either VCO in any way.]

Although the connection is there in the layout, I did not use the Soft Sync input on my final build. I don't think I'll need it. I did use the FM input. You can connect a second VCO to that for instance.
Here's a look at the finished product, panel and all. The powder coated Aluminium was a great choise and looks so cool. It doesn't scratch easy at all, it's perfect for this project.

I'll explain what's on the panel.
We have the FM input at the top left. The big knob at the top is the Coarse Tune potmeter, below that on the left are the CV1 and CV2 inputs and on the right are the Triangle-, Ramp- or Sawtoothwave and below that the Squarewave outputs. ( I always put inputs on the left and outputs on the right.)
Then there are three inputs to the left of the blue knobs. Those are the Pulse Width Control Voltage input, the blue knob next to it controls its level. Underneath that are the Positive Hard Sync and Negative Hard Sync inputs. The bottom blue knob is the Manual Pulse Width control if you don't use a control voltage. The switch with the diode symbol let's you choose to put a diode in the external Pulse Width input line which de-couples it from the internal PWM control or to bypass that diode and get more range on the PWM control knob. [edit] I have since scrapped this idea and I took out the diode. The switch is now used to turn off or on the manual Pulse Width Modulation potmeter as described above. (It's not necessary but the switch was there so might aswel use it for something). I added one more output which isn't in this picture and that is a "CV out" function to connect the second VCO to the first one. It's simply switched in parallel with the CV-1 input.

Okay, that's it for this one. If you have questions or suggestions please write them in a comment. Next part will be about a filter, probably the Prophet One Low Pass filter.
Stay tuned!


  1. Is it possible to have the outputs to be bipolar(+/-) by using capicators like you mentioned in the article about the digisound VCO?

  2. Hi Eddy. Mi name is Mikel from the Basque Country and I am triying to build a VCO with the AS3340. I bought two of them and i followed the Look Mom No Computer schema for the Pulse Width Modulation version of his VCO circuit. The saw and triangle shapes outputs give more or less the correct output the pulse output gives a variable duty cicle output but always the high state is up from the 85% and I can not make the shape less in high state than that. I changed the 10K resistor from pulse output to ground with a 51K one, I traced every component position and connection finding that except for the inherent component error margin they are acuratelly close to the indicated values and I can not resolve the issue.
    Do you have any idea why could this behavior occur? I ask you this because i saw your circuit is slightly different from the one i am building and perhaps you noticed issues that conduct you to change and improve the "original" circut design. Also I have to mention that I use a +12 G -12 volts power source.

    I wolud appreciate your expertice help.

    Many thanks!

    1. Hi Mikel. The Pulse Width input of the chip on pin 5 needs to vary between 0V and +5V to get the correct Pulse Width control. I suggest you take a look at my other 3340 VCO design (the 'Really Good AS3340 VCO") project 18 on this website. It is a much better design and it will give you the correct control over Pulse Width. Changing the resistor on the Squarewave output doesn't do anything. You can use 10K or 47K or 51K. It doesn't matter. But the way the PWM is done in the Digisound 80 VCO (project 18) is the right way to do it. I hoped this answers your question. Good luck!

    2. Hi Eddy. I have assembled part of the Very good VCO circuit and it works very good, even PWM part wich at first did not work because a stripboard's fabrication error!!! I allmost got mad trying to determine why when i was moving up and down the PWM shape pot it didn't change PWM shape but the frequency.
      After trying all kind of changes of components, layout rechecking and oaths i realized that i had no checked for shorts. Aparently there weren't but when i decided to use a continuity test i found the problem. I cut that underlaying short and suddenly allmost all went right! Now i have new issues with triangular wave, too low, pulse too high (i don't implement zener voltage limitation), freq. sweepinng too small, no range enogh... Lots of things but in the end i am moving forward. Thankiu!

    3. Sorry. I have forgotten to say that i am Mikel.

    4. Hi Mikel, I'm glad you got so much closer to getting it working. The squarewave is close to 13,5V if you don't use the zener diode and the triangle is only 5V when it comes out of the chip and it is amplified by the opamp stage with a gain of 2. So if you concentrate on the opamps and make sure the resistors have the right value then it should really work as it should. But I had the same problems in the beginning and if I couldn't solve it I would just start again and then it usually works. And that's how you learn building modular synthesizers ^___^

  3. I will follow that schematics.

    Many, many thanks!

    1. You're welcome Mikel. Let me know if you have more questions.