R.G.'s sequencer

[Jason Stout]

I then made what I felt was a minute change to my breadboard layout and bam I have a clock pulse!

Hmmm... that's weird. What minor change did you do, exactly?

I just moved the 100pF cap away from the CD40106 with jumpers. I haven’t been able re-create it, so I have found nothing.. In fact, I came back to my breadboard the other day only to find random mode does not work anymore. I'm back to being stumped.

Looking at it, I wonder if I made the load pulse too short. What happens if you just short the 100pF capacitor feeding the -Ld signal inverter?

Tried it, I get a wide pulse with no switching at all. (Are we missing the threshold???)

I have tried 100pF, 200pF, 250pF 500pF, and 0.001µF  

What does work 99.9% of my time is touching the legs of the capacitor with my fingers. I thought that this was just the addition of capacitance; so, I added caps in parallel, no go. I also tried paralleling resistance with the cap, no go.

….Mother Nature may be trying to tell us something we haven't discovered yet.

I know I haven’t discovered it.

FYI I am using Fairchild semi CD40193BC

[R.G.]

What does work 99.9% of my time is touching the legs of the capacitor with my fingers. I thought that this was just the addition of capacitance; so, I added caps in parallel, no go.

One of the things MN is telling us is that the circuit can change operation somehow without changing the circuit. So... power supply or grounding comes to mind.

The other thing touching or near-touching the circuit does is tie it better to earch ground. That can do a lot of things.

Do you by some chance have inverters with their inputs open anywhere?

[R.G.]

I should explain why I'm not running out to the garage to breadboard this and make some 'scope measurements.

I can't.

The garage is packed full of stuff in preparation for moving house in the near future, and I can't lay a hand on my tech equipment.  

I'll have to do the stuff in my head for now.

[Jason Stout]

The other thing touching or near-touching the circuit does is tie it better to earch ground. That can do a lot of things.

Do you by some chance have inverters with their inputs open anywhere?

Thanks R.G. I'll focus on grounding, and also have a look at the ultrasonic side of things tonight.

The garage is packed full of stuff in preparation for moving house in the near future, and I can't lay a hand on my tech equipment.

I hope the new home has a dedicated lab room, preferably NOT in the basement. By the way, that’s one of my goals.

[Arno van der Heijden]

I tried some simulations of this circuit with Circuitmaker. The up/down modes work flawlessly, but the random part is another story.

With the 100pF cap and 10k resistor it is impossibe to get short pulses on the load line. I only get a square wave with a frequency ~500khz.
There are some strange things happening between the 100pF cap and the 10k resistor as wel. I get pulses here, but they are superimposed on a high frequency (~500khz) signal of ~0-3V peak to peak.

Also, If I run a simulation of the up/down mode with the wave shaper and inverter connected to the load pin, but not to the main oscillator (switch position 1 or 2), the results get screwed up because of the effect of the high frequency signal on the load line that is always present.

All of this can of course be caused by the shortcomings of Circuitmaker/spice, but I have no idea if this is the case here.

If I change the cap to 100uF and the resistor to 100k, I do get negative going pulses which are about as big as the pulses coming from the preceding inverter. What size should I be looking for?
I tried the aforementioned values on my breadboard, but that didn't work. Now I'm confused.  :?

[R.G.]

Arno, I think that the circuitmaker sim is having some problems. I would disregard the high frequency stuff.

But the point of this particular post is that I have looked at the circuits and I think that it might be time to give up on the multiple oscillator pseudorandom circuit. It's good for some things, but on reflection, it might not be a good place to look for things to really be random. I think the difficulties we're having here may reflect trying to stretch what is a very simple circuit.

I've come up with an alternate. It uses a 16 pin CD4105 and a 14 pin CD4070 to make a better pseudorandom generator that has 2 to the 15th (32768) different states before it repeats. It can be set up so that it goes through one state per LFO clock in this circuit. With a 10 Hz LFO, the pattern would repeat every 3277 seconds, or about an hour before there's a discernable pattern.

It's two more CMOS packages, but it plugs into the circuit using just the stuff that's already there, and should give a much more random result for the "random" setting.

What do you think?

[Arno van der Heijden]

But the point of this particular post is that I have looked at the circuits and I think that it might be time to give up on the multiple oscillator pseudorandom circuit. It's good for some things, but on reflection, it might not be a good place to look for things to really be random. I think the difficulties we're having here may reflect trying to stretch what is a very simple circuit.

Hmm, I really liked the simplicity of this circuit. With a 10 Hz LFO it works really well.

I've come up with an alternate. It uses a 16 pin CD4105 and a 14 pin CD4070 to make a better pseudorandom generator that has 2 to the 15th (32768) different states before it repeats. It can be set up so that it goes through one state per LFO clock in this circuit. With a 10 Hz LFO, the pattern would repeat every 3277 seconds, or about an hour before there's a discernable pattern.

Sounds interesting. Do you really mean CD4105? Isn't that some kind of diode?

But isn't there a way to make the original one work a little better? Maybe by tweaking the R and C values? What are we aiming for here?
Or tweaking the high frequency oscillators? BTW, I used 1k 680R 560R for this. Isn't that too low for the CMOS input resistors?

BTW, I found this interesting site: http://www.musicfromouterspace.com/analogsynth/mmlogic.html

But they don't use the diodes in the wave shaper. What's the point of using them?
And wouldn't it be possible to save some components and an inverter by using the pulse generator on the bottom of that page instead of a square wave generator and a wave shaper?

[R.G.]

Hmm, I really liked the simplicity of this circuit. With a 10 Hz LFO it works really well.

I did too. But the first builders seem to have run into problems with it, and I can't imagine what happens when a first-time builder or two get hold of it. The three-oscillators stuff is not really random, so it makes sense to put in something that's better. Maybe just as an alternate. The two-IC version shares only power, ground, clock, and the three outputs with the circuit, so changing one for the other is simple.

Do you really mean CD4105?

... can't type... it's really the CD4015 eight stage shift register.

But isn't there a way to make the original one work a little better? Maybe by tweaking the R and C values? What are we aiming for here?
Or tweaking the high frequency oscillators? BTW, I used 1k 680R 560R for this. Isn't that too low for the CMOS input resistors?

There may be some magic tweaks, but I am not in a position to tinker and find them right now, and I want what you guys build to work.

The resistor values really don't affect the CMOS inputs; or rather the CMOS inputs don't affect the resistors. The CMOS inputs are a layer of metal on high purity glass. The inputs are really just "watching" the resistor/capacitor charge/discharge without taking any part in it.

BTW, I found this interesting site: http://www.musicfromouterspace.com/analogsynth/mmlogic.html

Yep, that's a pretty good sendup of MML.

But they don't use the diodes in the wave shaper. What's the point of using them?

They use them, they just don't show them. The diodes they count on are inside the CMOS chips as protection diodes on the inputs. When the MML monostable is inactive, the resistor lets the cap build up the full power supply across the chip. When the driver changes state, the cap is pulled to the other power supply, "pumping" the voltage on the cap into the CMOS input. My diodes clamp this to the power supply  (+ or ground, as appropriate) to dump the cap charge fast. The ones on the MML page let the protection diodes do this. It ...might... work forever with tiny, fragile internal diodes, but 1N4148's are so cheap, I like to use them explicitly outside the chip.

And wouldn't it be possible to save some components and an inverter by using the pulse generator on the bottom of that page instead of a square wave generator and a wave shaper?

Yes. You can even pervert that pulse generator with a pot and two diodes into a variable pulse width generator. But I played conservative. It's worth messing with.

[The Tone God]

Its hard to create psdueo randomness without using many more parts. With the Vanishing Point all I had left after the chase and bounce modes was one opamp. Since I don't like having left over unused stages and I like to as many features as possible I made the second clock which feeds a serial data input of the shift register making the "pattern" mode.

While this did not create a random mode it did provide something that can be random-ish if set correctly or atleast random enough to most listener's ears. Three resistors, a cap, a pot, and a switch was all that was needed. This was much easier to do especially when I was trying to keep the part count down so the board could fit in a BB.

Andrew

[Arno van der Heijden]

R.G., I'm interested in your alternative pseudo-random circuit. I hope you can find some time to draw it up.

[Jason Stout]

I am too!

[R.G.]

I'll sketch it out. It's not too bad - only the two IC's needed, no resistors, caps, or other parts.

[Arno van der Heijden]

Any updates yet?