Arduino controlled oscillator

[petegaggs]

Here is my latest project. It's an oscillator controlled by an arduino.
This is a novel approach - it's a hybrid design, with a standard op-amp integrator to generate a sawtooth much like other analogue oscillator designs.
However, the reset of the integrator comes from the microcontroller, this avoids needing any tempco, matched pairs, calibration etc.
I use the 16 bit timer to generate a square wave of the correct frequency. The rising edge of this will reset the integrator.
So as well as a sawtooth, there is a square wave available as well, I'm not using this yet. If I were to use this, I would suggest buffering with an opamp before taking it off-board.
The integrator drive comes from an SPI DAC. The exponential conversion is done in software by lookup table. The integrator drive level does not need to be precise, any error here will result in a small change of amplitude of the sawtooth, rather than a frequency error, which would be the case with traditional oscillator circuits.
The sawtooth is about 4V pk-pk, and the design works over a 7 octave range.

Other features:
MIDI control
powered from +9V supply (I have a local inverter to generate -9V for the op-amp).
Nice an simple, not too many parts.
I used an Arduino nano in my prototype, since that was what I had to hand, but it was originally going to be a pro mini 5V 16MHz.
Note that you need the arduino MIDI library installed. Also the timer is used directly, so it won't work on all Arduinos. It will work on any '328p based hardware.

source code and other stuff is here:
https://github.com/petegaggs/MIDI-controlled-oscillator

---

oscillator.jpg
Description:	my prototype
Filesize:	974.01 KB
Viewed:	892 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

---

MIDI-controlled-oscillator.pdf
Description:		Download (listen)
Filename:	MIDI-controlled-oscillator.pdf
Filesize:	54.1 KB
Downloaded:	1199 Time(s)

---

[Cynosure]

Cool. I have been meaning to test this for a while, but I never got around to it. From the sounds of it, what you have created is actually a DCO. You should also adjust the voltage so that the higher frequencies do not have a lower amplitude.

[Grumble]

In the low frequencies I don't see any problems, but how accurate is it in the higher audio range?

[petegaggs]

[petegaggs]

[Cynosure]

Very cool

[petegaggs]

[Grumble]

very nice project!

[LFLab]

Nice, so a DCO? Interesting!

[Grumble]

it’s a new breed: DCAO (digital controlled analog oscillator)
Drawback is that it plays a fixed maximum number of frequencies, due to the limit of the size of the two lookup tables.
Maybe it is possible to have the Arduino calculate the two values, or is it getting to slow doing it like this?

[Sebo]

I've started to build one of these:
http://www.russellmcc.com/posts/2013-12-01-DCO.html
But never finished.
I think is same concept.

[petegaggs]

[petegaggs]

[petegaggs]

I have updated the software.
Now the values are calculated on the fly in software instead of lookup tables.
This allowed me to add midi pitchbend support, and also analogue control voltage to modulate the pitch. This is intended for LFO or envelope generator to modulate the pitch.
I experimented with a pure V/octave control mode, this didn't work very well, it seems to be a limitation of the arduino ADC which is not quite up to it. There is a bit of non-linearity and it was only usable over a couple of octaves.
I think that with a decent external SPI ADC it might work better.

---

oscillator test.mp3
Description:	Here is a recording of me testing it, showing off the midi pitch bend and analogue pitch adjustment. The latter sounds smoother. The midi pitch bend is a bit granular, this is down my controller keyboard.	Download (listen)
Filename:	oscillator test.mp3
Filesize:	907.75 KB
Downloaded:	814 Time(s)

---

[Sebo]

I did several modules in Arduino using it's ADCs as 1V/Oct input and I could get 5 octaves of decent tracking... Always using look up tables.

[Grumble]

How do you use tables and still implement pitch-shift?

[Sebo]

I interpolate the tables...

[petegaggs]

[synthjakk]

This is exactly the kind of project that I want to try. Apologies if this is a stupid question, what is the purpose of the LMC7660INNOPB Voltage converter? and why is'nt it or the 78LO5 connected to the rest of the circuit?

I have ordered the parts and want to build this project as a learning experience! Any update on the gate implementation?

[petegaggs]

[synthjakk]

Ahh I see, that makes A lot more sense! Regarding the code, you mention that the value 0.97783686 is in the attached spread sheet, however I can't seem to find this value anywhere. What is its purpose?

Also, what does the 1000000.0 and the -1 represent here?

uint16_t timerSetting = round((1000000.0 / freqHz)-1.0);

My parts arrived today so I will upload a video once it is all working properly! I want to try and fully understand the code to help comprehend the project

[petegaggs]

>Regarding the code, you mention that the value 0.97783686 is in the attached spread sheet, however I can't seem to find this value anywhere. What is its purpose?

Sorry, I forgot to update the spreadsheet after my last changes, it's there in the git repository now. So originally, I worked out the values for timer and DAC in advance in a spreadsheet for each midi note. This is now changed so the arduino software calculates them itself (in function updateNotePitch), this allows analogue voltage control to be used, and pitch bend.
If you look at the sheet called 'voltage control' you'll see how it works. The constant DAC_MULTIPLIER is a function of
a) desired ramp peak voltage
b) integrator R
c) integrator C
d) DAC bit resolution
e) DAC full scale voltage
TBH, it doesn't really need to be a precise value, 1.0 would probably have been close enough, and would have saved a few cycles, however having this constant does allow the above parameters to be changed about.


>Also, what does the 1000000.0 and the -1 represent here?

uint16_t timerSetting = round((1000000.0 / freqHz)-1.0);

We have to multiply by one million because freqHz is a floating point number representing the frequency in Hz, and the timer value is effectively in microseconds (the timer is actually running at 2MHz, and the timer period is half the note period)
The -1 is needed because the timer is zero indexed, e.g. loading it with a value of 0 will make the timer run for 1 x timer clock, so to make it exactly right the timer is loaded with the note period in microseconds, minus one.


I look forward to hearing your results!

[synthjakk]

>This is now changed so the arduino software calculates them itself (in function updateNotePitch), this allows analogue voltage control to be used, and pitch bend.

Thank you for taking the time to answer, so by analogue voltage control do you mean an external input such as a LFO? I am really only interested in producing a saw tooth for now, so maybe I can factor this out. baby steps lol.

>a) desired ramp peak voltage

Was the max ramp voltage of 4.442 calculated from the equation V=i*t/C?

How was the DAC multiplier constant equation derived? is there a source that may explain this? I have added an image of what it appears to be on your spreadsheet. I get that 2^12 - 1 is the amount of available steps from the DAC.

I hope you do not mind answering these questions for me. I start building tomorrow!!!

---

Screen Shot 2018-03-18 at 18.21.59.png
Description:
Filesize:	14.16 KB
Viewed:	461 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[petegaggs]

Hi Synthjakk,

> so by analogue voltage control do you mean an external input such as a LFO?

Yes, that's exactly what I had in mind for this. It is reading the analog voltage on pin A0. If you don't need it for now, just comment out this line:
#define ANALOG_CONTROL

If this is not defined, then the analogue control will have no effect.

>a) desired ramp peak voltage

>Was the max ramp voltage of 4.442 calculated from the equation V=i*t/C?

Yes. That equation is right, V is the level of the ramp after time t, and i will be V_dac/R, since the opamp input is a virtual ground.

I was looking for around 4V or so pk-pk voltage on the ramp waveform. (much more than that you would hit the limit of what the opamp output can swing to, though this depends on the supply used, I'm using +/-9V, so I want to keep it quite small). I chose some component values that gave me about that when the DAC was at max (5V) and the frequency was highest (4186Hz).
I can't remember how I got to 4.442 volts, I think it just worked out that way when I chose some standard values for R & C.

>How was the DAC multiplier constant equation derived?
Your equation is correct, and that is how I derived it in the spreadheet. However now I come to think about it, there is a simpler way to explain it:
the value we have to send to the DAC is proportional to the frequency, and this constant defines that relationship.
Think of it this way, at the highest supported midi note (108) the frequency is 4186Hz, and the DAC is at full scale, which for a 12bit DAC is 4095 (or hex FFF). 4095/4186=0.978.


>I hope you do not mind answering these questions for me. I start building tomorrow!!!

No problem at all.

[synthjakk]

Thank you again, you are being very helpful. I have built the circuit and am just waiting for the 5 DIN connector to arrive so that I can test it.

Something that I am trying to work out is the frequency range that is outputted by the microcontroller PWM out. the spread sheet shows a range between say 300 to 30,000 values sent to the timer.

The timer is prescaled from 16Mhz to 2MHz.

So what frequency would the PWM output with the value 300 say ?