Tunable noise from a fixed source

[richardc64]

I am adding sounds to a device that already has a noise source, an MM5837, which has recently gotten some bad-mouthing here that I feel is undeserved. In any case, I thought it would be nice if the circuits I added could operate on differently "pitched" or swept noise. Not wanting to build multiple PRBSs, (or SN voices, for that matter,) I sought a way to get different pitches from the 5837, which does not have a clock input. My solution is partly inspired by synthmonger's method of using several oscillators to Clock a 4520 4-bit counter.

First thing I had to do was level-shift the 5837 output, because in this partcular device it's powered by Gnd and -9volts, and I like to keep digital stuff on the positive side of 0 volts. The circuit I used is a schmitt trigger sine-to-square converter straight out the National CMOS data book. I used it because of the capacitor coupling. (This should work equally well to square up the output of a reverse-biased transistor noise generator.) This squared up noise clocks the 4520.

By itself this gives different "octaves" of noise -- quotes used because the sound is essentially pitchless. But the 4520 counters also have enable inputs. When Enable is high (1), the rising edge of Clock increments the count. When Clock is low (0), the falling edge of Enable increments the count. Apply an oscillator on Enable and the two signals fight over which one gets to advance the count. With a variable frequency oscillator one can Tune the noise. And if it's a VCO...well, that just adds to the fun, doesn't it?

I found the best sounds at the Q1 and Q2 outputs, (f/2 and f/4 respectively,) though the other outputs may be useful in some circumstances. In the sample the output is from Q2. I used a schmitt oscillator with a 100k pot as R and .1uF for C. Power was 9Volts.

The beauty -- to me, at least -- is that the other half 4520 can use the same noise on Clock and a different oscillator on Enable for a second tunable noise source.

Another possibility is to also inject a signal into Reset. I didn't find this especially pleasing, the periodicity of the resulting sound being too noticeable. Still, it's something to try. The R-C combination shown gives a TC of 10uS so that Reset doesn't stay low too long.

---

swept_n1.gif
Description:
Filesize:	3.13 KB
Viewed:	1420 Time(s)

---

---

t_noise56.wav
Description:		Download
Filename:	t_noise56.wav
Filesize:	237.04 KB
Downloaded:	349 Time(s)

---

[JovianPyx]

I think you're eluding to at least one thing I tried and found underwhelming, that being the use of an LFSR (pseudorandom shift register noise generator) with a VCO as the clock. What I hoped for was that moogy pitched noise sound. That's not what happens. All you get is a shifted spectrum of noise, lower spectrum from lower VCO pitches. It's really not a whole lot different sounding than using a white noise generator patched into a lowpass VCF and varying the CV to the filter.

More research on the subject of pitched noise informed me that what I really wanted was to use a VCO with noise modulation of the pitch control voltage.

Since my analog equipment is linear (Hz/volt) that isn't easy. I figured out a way to do it and never implemented it in analog circuitry due to the complexity/effort required. I did, however, use the same idea in a digital implementation on an FPGA (way off your topic, but included for completeness of my response) and it works quite nicely.

If you have expo VCOs, you can get this effect by simply mixing (summing) some noise with the pitch CV.

[richardc64]

I have expo VCOs, but not in the unit I'm working on. It's very simple, and what I want to add has to be even simpler. Precision V to F is not a requirement.

Still, I tried using the level-shifted noise to FM a simple 40106 oscillator. The result sounded pretty much the same as the 4520 circuit but without "octaves." However, since it uses one less chip, it wins in the simplicity department and has the additional benefit of control of the depth of modulation. So it's very much worth exploring.

Thanks.