simple, clean-ish octave down

[earthtonesaudio]

Here's an idea for a much-simplified "clean" octave down, like the Boss OC-2 and similar circuits.  Seems to work fairly well in simulation...


Import into  Falstad's Circuit Simulator Applet:

Code Select Expand

$ 1 5.0E-6 2.3728258192205156 50 5.0 43
v 432 560 432 480 0 1 750.0 0.5 0.0 0.0 0.5
v 432 608 432 560 0 0 40.0 5.0 0.0 0.0 0.5
g 432 608 432 640 0
a 640 496 720 496 0 9.0 0.0 1000000.0
a 624 320 704 320 0 9.0 0.0 1000000.0
w 432 480 432 336 0
w 432 336 432 304 0
r 432 336 512 336 0 100.0
r 432 304 624 304 0 5000.0
r 624 256 704 256 0 10000.0
w 624 256 624 304 0
w 704 256 704 320 0
c 432 480 528 480 0 1.0E-8 -0.021983906524478236
160 528 368 528 336 0 20.0 1.0E10
w 544 336 624 336 0
c 528 368 528 400 0 1.0E-10 4.500017112089536
g 528 400 528 416 0
w 528 480 640 480 0
r 528 480 528 560 0 100.0
w 432 560 528 560 0
w 528 560 640 560 0
w 640 560 640 512 0
f 720 496 784 496 0 1.5
w 784 512 784 608 0
w 432 608 784 608 0
w 784 480 864 480 0
r 784 480 784 416 0 100.0
R 784 416 784 368 0 0 40.0 5.0 0.0 0.0 0.5
155 864 448 944 448 0 0.0
w 864 448 864 416 0
w 864 416 992 416 0
w 992 416 992 512 0
w 992 512 960 512 0
160 1008 240 976 240 0 20.0 1.0E10
w 992 256 992 416 0
w 976 256 704 256 0
w 976 224 432 224 0
w 432 224 432 304 0
w 512 352 960 352 0
w 960 352 960 448 0
c 1008 240 1104 240 0 1.0E-5 0.6555383584867753
r 1104 240 1104 608 0 10000.0
w 784 608 1104 608 0
o 0 8 0 34 0.5846006549323611 3.6537540933272573E-4 0 -1
o 41 8 0 34 9.353610478917778 3.6537540933272573E-4 0 -1

Report to moderator   129.79.160.200


[Quackzed]

i just ran it, thats pretty slick!
-i had to change the 10k in the opamp fb loop to 5k to make the top and bottom part of the wave even, but aside from a little crossover distortion, its pretty clean.
nice

[earthtonesaudio]

Oh yeah.  I made it uneven on purpose for the simulation so I could easily check what half was being inverted.  Definitely for a real build it should be symmetric.

[Astronaurt]

I'm totally unfamiliar with a circuit like this, Could you explain to me how the flip flop chip works and where one is supposed to take the output of all this business? Hate to n00b out, but I'm really interested in the octave down thing so I have to ask. thanks!

[earthtonesaudio]

There are some details left out that will most certainly trip you up if you still have your Noob shoes on. 

There are several different types of flip flops.  This one is a toggle flip flop which changes its output state (logic "high" or logic "low") every time it sees a negative-going edge.  Er... it might be the positive-going edge but that doesn't really matter.  Essentially if you have a square wave going into the flip flop with a frequency X, then the frequency coming out of the flip-flop is X/2.

In that circuit I used an N-channel MOSFET to translate the output of the op-amp (0-9V) to a 0-5V signal compatible with the flip-flop.  Some flip-flop ICs that use CMOS technology are compatible with 0-9V control inputs so the MOSFET may be unnecessary.

The output (referring to this updated schematic) is the voltage across the 100k resistor.  Note that this resistor is referenced to 4.5V, not ground, and I've left out a lot of details regarding biasing and AC coupling signals. 

Read up on "AC coupling" and "biasing op-amps" and "single supply voltage cirucits" to help find your way with the analog ins and outs, and also check out "interfacing logic families", "TTL to CMOS" or similar topics for how to connect the op-amp to the flip-flop correctly.  In the updated schematic I did not use the MOSFET and it still works.

[pinkjimiphoton]

hey eta, any chance you could post an actual schem of this? not sure how to translate the app to an actual graphic.

thanks bud!

[artifus]

high light and copy the above code, follow the link and allow the java applet to run. click file, import then paste in the code.

[pinkjimiphoton]

i get that part, artifus, but how do you save it as a schem? and (newbe alert) what are the chips involved? 

[artifus]

i haven't actually looked yet but i'd guess op amps.

[earthtonesaudio]

Yep, regular op-amps.  I have added some extra stuff (output resistance on the op-amps, AC coupled input) to make it a bit closer to a practical circuit.

Here is the link.

[Gurner]

I would think the biggest problem (which is prevelant with most octivider type circuits) is the bastard strong octave harmonic embedded in freshly plucked real world guitar signal (vs a nice sine showing in your sim) breaching/triggering the zero detect circuitry.

[earthtonesaudio]

Absolutely.  There's nothing in my schematic to discourage the first harmonic from becoming a false trigger.  I'm not really sure what to do about that except perhaps a lot of signal conditioning which would make the circuit much more complex.  Not that there's anything wrong with "more complex", of course. 

[~arph]

What would be cool, is a single DSP chip solution, just for fundamental extraction.  Something like the taptation (which is probably just a PIC or AVR.). Little preconditioned guitar in. then square/sine fundamental output. But yeah, then why not do the octave stuff too 
That would open up a lot of posibilities for the diy community I guess.

[CynicalMan]

Hmmm... How's this?

Low-pass filter -> Peak Detector -> Attenuator -> Comparator
                |                                      ^
                |                                      |
                |--------------------------------------|

That would probably allow you to ignore most of the junk zero crossings by measuring crossings at a specific percentage of the wave height. You'd have to throw something in to prevent noise, but this is the rough idea. You could probably do this with a dual op amp. Disadvantage: the symmetry of output pulse would be different from the symmetry of the input. This means that it wouldn't work for Alex's idea.

[~arph]

Another problem is that harmonics can be louder than the fundamental.

[earthtonesaudio]

Another problem is that harmonics can be louder than the fundamental.

Yes, it's clear from the sim that the sub-octave disappears when the input contains a strong harmonic. 

[PRR]

> how do you save it as a schem?
> post an actual schem of this?

Java has limited interaction with the system. Maybe only screen-grab?



> what are the chips involved?

Probably "ideal". The top triangle is acting like an op-amp. The bottom triangle is working as a comparator, but an op-amp can do this job. The FET is not needed. The box is a flip-flop, a standard logic part available under many chip numbers. Nothing here is very fast nor very precise, no exotic parts needed.

It's just another way to full-wave rectify.

[Earthscum]

Ya forgot the cmos switch, Paul.  

You know, looking at parts count vs. circuits/package count, you could build a pair of these in a single circuit if ya wanted. Sub a quad amp for the dual, flip-flop comes in a dual package, and you would already be using a quad analog switch. You could do a double-down.

Question... what's up with the (battery?) symbol between the ground and signal/+ in? I use Ubuntu... Java fights me, so I am going off of Paul's screenshot.

[brett]

Hi

There's nothing in my schematic to discourage the first harmonic from becoming a false trigger.  I'm not really sure what to do about that

An Auto Gain Control could remove the "fresh plucked" effect (ie compressing the dynamic range to half or quarter (log, dB) really diminishes the undertones and overtones).
cheers

[earthtonesaudio]

quad analog switch.

A quad SPDT analog switch?  I've not seen one of those... ?

Question... what's up with the (battery?) symbol between the ground and signal/+ in? I use Ubuntu... Java fights me, so I am going off of Paul's screenshot.

In the simulation I used 0-9V for the op-amp supply and the battery is 4.5V to bias the op-amp inputs at the midpoint.