Duty Cycle Configurations

6581r4AR · Joined: Apr 03, 2011 Posts: 15 Location: Texas, USA

Salutations to you all. Long time lurker, first time poster.

I've been trying to implement adjustable duty cycle on some simple multivibrators. Disappointingly, when using only one diode it will work, although it also will cause the frequency to change.

So I've narrowed it down to two configurations.
I am having trouble deciding on which configuration might be more accurate. I'm looking for the version that will affect the frequency the least... Would there be any noticeable difference?

Also, should the pots be linear or log?

Any thoughts?

Sorry if this is a "find out yourself" type question. I can't really test it myself at the moment. I'm in between my parts ordering stages. But I would like to get my schematics right while I have this on my mind.

Duty Cycle.png

Description:

Filesize:

30.13 KB

Viewed:

15965 Time(s)

Kabzoer · Joined: Feb 07, 2011 Posts: 82 Location: Belgium

I don't think it will make any difference.

tjookum · Posted: Sat Apr 09, 2011 3:39 am Post subject:

I don't think I've ever seen the right one but Im very sure the left one works without changing the frequency.

This, for me, is the best part about lunetta's, it's just 5 min on the breadboard and you have your answer. Just experiment and have fun!

6581r4AR · Joined: Apr 03, 2011 Posts: 15 Location: Texas, USA

richardc64 · Posted: Sat Apr 09, 2011 8:46 am Post subject:

I can't even begin to figure how the right circuit behaves. It looks as if the location of the freq. pot pretty much defeats what the width pot is supposed to do.

In the circuit on the left, the width pot won't affect the frequency, but each pot's value will affect the range of effect obtainable from the other. Consider if both pots are 100k and the fixed resistor is 1k. The setting of the width pot doesn't matter. Whether it's set for 1k half a cycle and 99k the other half, or 37k/63k, the total average value with respect to the timing is 50k. If the 100k freq pot is set for zero ohms the value R in the equation f=1/RC (which is only approximate for these oscillators,) is 51k, and the pulse width can be varied from roughly 2% to 98%. If the freq pot is set to 100k the total R is 151k: only 3 times what it was with the pot at 0. The freq range is only 1:3 Furthermore, the the pulse width range is reduced by 1/3rd (I think,) because the 50k average value becomes a smaller percentage of the total R.

Attempting to get a wider frequency range by increasing the value of the freq pot just makes matters worse, width-wise. Increase the width pot's value, and frequency range will again be diminished. The trick will be finding an acceptable compromise of freq range vs width range.

One possible solution may be to control the frequency not by varying R, but by varying C. This can be done with a multi-position switch, which is too crude to even consider, or smoothly, with a pot. In the sample below, you hear the frequency go from LFO range up into audio. In the schematic showing the values I used, greater resistance = higher frequency. Note that the "controlled" capacitor is 10x the value of the uncontrolled one. Other ratios are probably feasible. I did not try width control.

(Affect/Effect)

vary_c.gif

Description:

C can also be voltage controlled

Filesize:

4.59 KB

Viewed:

16032 Time(s)

vary_c.mp3

Description:

Download (listen)

Filename:

vary_c.mp3

Filesize:

528.98 KB

Downloaded:

1226 Time(s)

6581r4AR · Joined: Apr 03, 2011 Posts: 15 Location: Texas, USA

So you're saying something like this would be more appropriate? or does that put me right where I started?

And why is that feedback resistor(470K) so high? Is it just setting the highest frequency reachable?

I don't mind if it is a little crude in the end, I will be having multiple range caps on a switch in the final layout. And just so I understand correctly, in my original post, the one on the left will work except the larger the freq resistance is the less width range will work?

vary.jpg

Description:

Filesize:

85.98 KB

Viewed:

16007 Time(s)

richardc64 · Posted: Sun Apr 10, 2011 12:57 am Post subject:

6581r4AR · Joined: Apr 03, 2011 Posts: 15 Location: Texas, USA

richardc64 · Posted: Sun Apr 10, 2011 4:52 am Post subject:

6581r4AR · Joined: Apr 03, 2011 Posts: 15 Location: Texas, USA

The mosfet is 'sposed to be a switch controlled by the LFO, modulating frequency of either the main osc or the synced osc. It turns on and off the extra pot reducing the resistance of the one it is attached to. Is that wrong too? I just followed the example of the "How to make a digital synth" on the hackaday page.

Other than that, I think I understand everything a little better now. Thanks so much for the help. *big smile*

richardc64 · Posted: Mon Apr 11, 2011 5:48 am Post subject:

I found another reason not use too high a value for the feedback resistor: it affects the duty-cycle regardless of any width control that may be present.

As can be seen in the screen captures, at 100k the waveform is fairly close to being 50-50, but at 330k the asymmetry becomes more significant. With 1M (and a smaller cap to keep the freq reasonable,) the shape is most definitely non-square.

This is something the schmitt trigger datasheets don't mention. (Edit: Or maybe they do -- somewhere in the equation.)

There's probably some highly technical explanation for this, and the slope of the "1" state might offer a clue. But the effect is of little concern when the oscillator is an audio source -- especially in Lunetta's, which are supposed to be "dirty." As modulation or gating, though, the asymmetry could be noticeable and undesireable. It might cause glitches when clocking sequencers and latches, and is an added complication when attempting to control the pulse width.

equation.gif

Description:

If I knew how to inset this as text, I would have. And WTF is that lower-case script "L"?

Filesize:

5.39 KB

Viewed:

15865 Time(s)

asym_2.gif

Description:

Recordings made in Wavosaur

Filesize:

13.53 KB

Viewed:

509 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

6581r4AR · Joined: Apr 03, 2011 Posts: 15 Location: Texas, USA

WOW. That's super helpful... Now if I only knew what those variables represented. =P I'll do some searching into it.

And I went out and got myself a 40106 the other day so I could do some testing for myself. I found that replacing the not so reasonable 5M idea with a 1K did well. I found that over a certain threshold, it just cut off the control cap all together. The key is finding that threshold point and making the variable resistor as small as possible for a smooth freq sweep.

6581r4AR · Joined: Apr 03, 2011 Posts: 15 Location: Texas, USA

Just though I should update with a more or less fixed version of the schematic I posted. I didn't like the idea of having it up with no heads up on the wrong values.

This one is a tad better, but I'm still experimenting with capacitor values.

Voice 1.png

Description:

Fixed

Filesize:

30.66 KB

Viewed:

414 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.