SINE wave from CMOS?

[aerogramma]

Hello!
i've been building my first CMOS based devices in the last weeks and had a lot of fun... and probably avoided to go bankrupt too

I really like the rawness that comes out but I'm also an indestructable SINE wave lover

Is there a simple way to get a sine wave from a CMOS IC?

thanks and all the best from here

aero

[Uncle Krunkus]

CMOS are inherently digital, and sine waves sound like they do because they are not. A very steep low pass filter would reduce a square wave to mostly just it's fundamental harmonic, which would be close to a sine wave, but it wouldn't be spot on. Almost certain to contain higher harmonics to some degree. And even then, it would have to be re-adjusted for every note.
Maybe you can waveshape a sine from a square, but I think they tend to use saw or triangle because it's a lot easier.
Sorry.

[jnuaury]

theres two ways to do this
one follows the concept of bit depth and the other is similar to sampling rates


"bit depth" method
you use a divider and make a "dac" with resistors after a square wave to get a sine wave essentially with a bit depth of however many divisions you use

also youll need to clock the square wave at a muh higher frequency than the pitch you want to be the sine-ish wave

link
http://www.robthefiddler.com/electronics-audio-diy/circuit-experiments/big-johnson-signal-generator/


"sampling rate" method
use a counter after the square and then run the separate count outs through resistors (or pots) set up as voltage dividers... its more or less a really fast sequencer and once again you need your square to be much higher than the sine-ish wave

link
http://www.milkcrate.com.au/_other/sea-moss/#8a

[urbanscallywag]

Not quite simple, but lookup Walsh functions. You don't need a DAC.

[Pehr]

Check out this thread:
http://electro-music.com/forum/topic-23878.html

The first post from para includes a sine/tri cmos oscillator.

brgds

[jnuaury]

[urbanscallywag]

No I don't think you are over complicating things, its just bizarre that the Walsh generators don't need a DAC.

[blue hell]

[aerogramma]

[urbanscallywag]

[andrewF]

haven't tried these, sin_0001 looks the more interesting of the two -

---

sin_0001.jpg
Description:
Filesize:	49.76 KB
Viewed:	453 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

---

sin_0002.jpg
Description:
Filesize:	53.24 KB
Viewed:	459 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[urbanscallywag]

The problem with these circuits is making them tunable (let alone voltage controlled) right?

[jnuaury]

[urbanscallywag]

I guess both of them have pots, but they just aren't included in the equation. Oops!

[jnuaury]

well the first circuit has the pots wired as variable resistors (rheostats)
but the second one the pot actually divides the voltage meaning you could remove the pot and toss a cv in where the wiper was connected (pin 14)
theoretically at least

[urbanscallywag]

Yeah, I see that. But when I looked at the equations I didn't think they were tunable even though I did see the pots.

[blue hell]

The first one seems to be a drum osc, where the sine fades out at a rate controlled by the variable resistor.

[urbanscallywag]

I saw the trigger input and thought it was like a bootstrap to get the thing oscillating.

[blue hell]

That's about the same thing ... when it can be set to oscillate forever.

[urbanscallywag]

I see, the pot is decay control, and with the suggested 22k resistor it should oscillate with continuous amplitude?

[blue hell]

When its really a drum osc, but I'm < 100% sure, it would likely decay to zero with 22k. So basically I don't know

[urbanscallywag]

OK if LTSpice has the appropriate components I will try it.

[loss1234]

OK..i am no expert. BUT i thought that walsh functions at least as outlined in the Electronotes supplement s-008 definitely DO need a DAC.

I quote from the heading "Digital to Analog Conversion" from supplement s-008 (which is a great read by the way)


"it is of course necessary to combine the various walsh functions in proportion and in the proper sense to get the proper waveform for the calculated walsh coeffecients. Four possible methods are shown below"


one of the methods the show is just a simple opamp summer but even that if used with a resistor for each walsh function is summing the separate functions into one analog waveform...

or at least that is how i understood it. (since all wash functions..like squarewaves are just a combination of high and low, or on and off/1 and 0, and if you can get a sine out of that there has to be a D to A process somewhere.

really not that much different than the Counter method with the resistor ladder someone already mentioned. except walsh functions (i think) could end up a lot more complicated if you want to build the whole first 31 functions. you need a lot of xors and flip flops. and then they still recommend filtering the result.



i find these walsh things really interesting and hope to build a working generator soon.


but really, SO MANY circuits use a simple DAC that i think the term DAC sounds much scarier than it is.

there is a great article free to the public on useful walsh stuff... and it has examples of getting SINES from just a few functions at

http://www.wiseguysynth.com/larry/schematics/walsh/walsh.pdf


great topic! thanks andrew for putting up those cool schematics! are they from the cmos sourcebook?

[urbanscallywag]

Yeah I started looking into Walsh functions after you asked on SDIY (I asked if anyone had samples). I built/simulated the 31 function generator in software. It doesn't need a DAC. Its bonkers.

[loss1234]

i am very confused. so you mean in software it doesnt need a DAC? what software did you use? i would love to hear some samples!! is it worth getting into?

thanks