Variable Duty Cycle Oscillator

[Inventor]

Let's try a little thought experiment, a twist on the Inverter oscillator. This is normally an inverter with an RC feedback and it oscillates, right? OK, so now let's leave the cap there but replace the resistor with two parallel resistor-diode branches facing in the opposite direction. Call them R1 and R2. What happens?

The charging cycle will be timed by R1*C and the discharging cycle will be timed by R2*C because of the two steering diodes. I'd guess that the overall timing would be controlled by (R1+R2)/2, and the duty cycle would be set by varying the values of R1 and R2 from each other.

Now let's take it one more step, and replace the two resistors with a potentiometer. The wiper is on the inverter's output, the two terminal connections go to the diodes, and the diodes connect to the cap and the input. We'll call the potentiometer value Rx. Let's denote the pull-up resistor as R1 and the pull-down resistor as R2=(Rx-R1). Then the frequency is controlled by the earlier mentioned expression (R1+R2)/2 which equals (R1+(Rx-R1))/2 and the R1's cancel leaving the frequency controlled by Rx/2.

What this means, if my thinking is close to correct, is that we can make a fixed frequency variable duty cycle oscillator where a potentiometer controls the duty cycle and the frequency is fixed. It's a variable duty cycle oscillator with fixed frequency.

Would that be handy for Lunettas? Is this idea nothing new? Have you seen it before?

Les

[slacker]

That's a cool idea, not new though I'm afraid, but most things aren't There's a few examples of that and similar things in Don Lancaster's CMOS Cookbook. One of the examples he uses it for is to create a variable brightness lamp, where the duty cycle sets the brightness.

If you put another pot wired as a variable resistor in parallel with the pot and diodes you can control the frequency and the duty cycle, I made a reasonable LFO like this by buffering the inverter's input to get a triangle/ramp wave form and using the output to get a variable duty cycle square.
I can't remember off hand what pot values I used but there wasn't much interaction between the duty cycle and frequency controls. The frequency did change slightly with the duty cycle but not enough to worry about. I'll see if I can dig out my notes

[Inventor]

Thanks slacker, another reinvented wheel! Well, if I'm following in the great Don Lancaster's footsteps at least for one step, I guess that's a good thing!

Cheers!

Les

[DGTom]

[slacker]

Here's what I actually built, It's part of my still work in progress, mostly CMOS synth.
If you ditch the opamp and associated components you've got something you could use in a Lunetta, the switchable cap takes it from LFO speeds to audio. Ditch the extra pot and the 1k resistor and you've got what Les originally posted.
It might be cool to hook a second inverter up to the square out then you'd get a complementary output, where the high to low ratio of the signal would be the opposite of the original output. That would give you 3 oscillators from one CD40106.
Apologies to Les if this is derailing his thread

---

lfosnippet.png
Description:
Filesize:	35.14 KB
Viewed:	9726 Time(s)

---

[Inventor]

Consider the thread derailed!

What made me think of the diode-based oscillator originally is something like what you mentioned earlier, the possibility of using multiple input gates to create oscillators. For example, using a NAND instead of an inverter, the extra input becomes an active high enable. But what would happen if you took say an XNOR gate and put different time constants in the two feedback paths?

Would you get some kind of galumpy galloping crazy oscillator or what? If that worked, how about more inputs? This one would have to be built or simulated, or at least calculated, as I can't think it out...

Les