Op Amp Frequency Doublers

[dwmorrin]

I came across this paper titled "Sinusoidal Frequency Doublers Using Operational Amplifiers:"
http://www.kmitl.ac.th/msplab/Journal/12.pdf

I was thinking of trying this out to see if it had an octavia/superfuzz type tone lurking within:


The output is hard to see, but it is labelled "V_O" and it appears below R1.
It is an unusual arrangement, where the voltage output is taken from the V+ pin of a dual op amp.

I will try and decipher what the resistor values should be, but so far the paper loses me in all the math equations.  Anyone have some suggestions for resistor values?  Anyone experienced with these circuits?

[R.G.]

It works by forcing a lot of current through the power supply on positive excursions of either opamp. Both opamps work from the same power supply, and R1 converts the current pulses pulled from the + power supply to a voltage. The same technique is used in other places, although I've never seen it done with two opamps. The voltage across R1 is a negative-going full wave rectified sine for sine waves in.

This will need some additional circuitry to convert it to a usable DC average voltage and peak-to-peak level. The unseen gotchas are that its output sits near the + power supply and that the opamps are polluting their own power supply, so they have to be made stable with high impedance power supplies. This may or may not be tricky, depending on the opamp.

Ought to work, needs some tinkering to get the parts and levels right.

[PRR]

Note that many classic SPICE opamp models will NOT simulate this correctly. They do not model dynamic current from output to supply pin. You can get a rough approximation by adding NPN PNP emitter followers inside the conceptual "opamp", because SPICE sure will model the collector-emitter current of a BJT.

> decipher what the resistor values should be

The voltage across Rc equals the input voltage. The induced current in Rc must come from the rails through R1 R2. We are only using R1 output. If you pick Rc so the peak input voltage causes an Rc current much greater than opamp idle current (for 1V in try 100 ohms), and pick R1=Rc, then the peak voltage in R1 is also 1V.

I would not try to get large outputs because the opamps are not immune to supply rail wobble. Best take a fraction of a Volt and add a post-amp if needed.

This needs a very tame opamp. TL072 may be tame enough.

[PRR]

Actually I may have that backward. Near Eq 12 we want K2<0.2. If I got it right, we want signal current smaller than idle current. For TL072 and 1V signal I get a value like 1K5 or 2K.

The value of R1 has no effect on distortion (assuming pretty-good opamp PSRR). You pick this for a small DC drop but reasonable AC output. 1K is perhaps an upper limit for TL072 at decent supply voltage. For 9V system even 300 Ohms may be a bit high to keep the TL072 happy.

This is of course just a push-push doubler which can be constructed in simpler ways. What the opamp deal seems to offer is stable idle current and intrinsic matching. And an opposing pull-pull output if you can think of a reason to want that.

[dwmorrin]

Success!  A quick try yielded positive results with +/-12V power, 1k for R1 and R2, 100Ω for Rc.  1Vp-p sine input => 4Vp-p output.  Thanks for the info guys!
Classic octave up sounds are happening with the guitar.  Guess I'll be tinkering with this for awhile...

Update: +9 single supply working as well.  1k for R1 still. 680Ω gets max output so far.  300Ω was not as good.  1V in gets 1.4V out.  Using a TL072.  R2 is a jumper, as it actually was above as well.

Whoa... just got it to straight guitar in -> BIG signal out.  1.5V p-p output with passive guitar input.

Getting pretty close to a dual op amp superfuzz.  Will have to post an updated schematic after refining...

[mth5044]

That is so cool.

[deadastronaut]

is the 'octave' still just up the 12 fret region?...(that type of octave)

[dwmorrin]

Correct, Rob.  It's old fashioned '60s "octave."  You'd think it was an octavia, or maybe a superfuzz if you just plugged in and listened.
The interesting twist is doing it with just a single dual op amp.  Might be good for "easy" versions of the classic octave up fuzz.  Add some diode clipping and a filter of your choice...

[deadastronaut]

yeah thought so, sounds cool using 1 opamp,  have you tried one into another?...just out of curiosity.

[dwmorrin]

Not yet, but it did cross my mind! 

[deadastronaut]

  yeah worth a try...might be pretty cool.

[dwmorrin]

Here's an updated, +9 volt version if anyone else would care to try.
I'm currently using big caps on the ins and outs, along with a 100k Rin.  Rout has probably been a 1M from the next amp input.  Any suggestions on fine tuning or modding this are very welcome!
I put R1 and Rc as trim pots for the moment, but I was getting good results with 680Ω for R1 and 10Ω for Rc.  Seems like the Rc made variable could function as a "gain" control.  R1 is probably best simply set to greatest output and then left alone.
Warning it can be a loud (>1V) output as drawn, so tack a volume control at the end, or turn your amp real low.

[dwmorrin]

[dwmorrin]

As an octave up fuzz, this thing is great.  I finally picked some values for everything, and I'm calling it quits for now.
I was using a humbucker neck pickup for the best octave up tones.  A single coil may require some tweaking.

I settled on 18Ω for the "Rc" resistor that connects the 2 op amp outputs.  At around 20Ω a good amount of drive occurs, and around 2Ω it gets really compressed and thin.  Lower this if you don't have enough "drive."
"R1" has settled to 1k, which puts pin 8 at 7V, for a nice and loud output.

And, Rob, I did try 2 in series... it didn't do much of anything!  I got nearly the same output as input.  So, that can be skipped.

[deadastronaut]

righto, glad you tried it though. 

looks like a cool nice and easy project for people to breadboard /build..nice one.