Balanced modulator

[brett]

Hi.  Over the last couple of years I've built a variety of ring modulator circuits, mostly based on transconductance amps (CA3080, LM13600 etc) or transformers (the Dalekator).  They all have issues with modulation signal bleed thru, and I've given up looking for a workable circuit.

But the other day I saw a schematic for a *balanced* modulator.  This fairly simple circuit uses a 4066 analogue switch to shift between an inverted and normal signal.  Doing this at a rate of 30 to 200 Hz apparently gives a similar effect to a ring modulator.

Does anybody have experience with balanced modulators?  Do you get lots of switching noise, and can this be controlled?

Thanks for any help

[bwanasonic]

The Craig Anderton EPFM ring modulator is actually a balanced modulator.

That's about all I know about it!

Kerry M

[R.G.]

But the other day I saw a schematic for a *balanced* modulator. This fairly simple circuit uses a 4066 analogue switch to shift between an inverted and normal signal. Doing this at a rate of 30 to 200 Hz apparently gives a similar effect to a ring modulator.

You mean the polarity reverser at GEO http://geofex.com/Article_Folders/polarity_reverser/polarity_reverser.htm?

You still have the problem. You have to make either the signal large enough or the clock bleedthrough small enough that you can't hear the clock, which means perfect switches, etc.

One thing that might work just fine is to compand around your modulator. That would bring the signal up and the clock feedthrough 'way down when there's no signal. I suspect that an NE570/1 would do the trick.

All of the ring modulators I've ever seen have problems with feedthrough. The simplest is probably the OTA versions or the MC1496 modulator.

[brett]

You mean the polarity reverser at GEO http://geofex.com/Article_Folders/polarity_reverser/polarity_reverser.htm?

No, but thanks for the link.  The schematic was in an Australian kids electronics book called "Short circuits" (get it?).  The project was called something like "Dalek voice".

Anyway, I've bought the kit and will put it together tomorrow if I take the day off work.  It's fairly simple.  Mainly needs the mic and mic pre-amp bypassed.  Actually, I might as well use the pre-amp, but reduce it's gain (from 400 to 4).

The output section employs a low-pass filter with an fc of 3.4 kHz to reduce the clock (555) and switching noise.  I might need to raise that (to 5 or 6 kHz), and convert it to 2 poles to allow more treble through but still cutting high-frequency clock noise.

I'll let everyone know if I can get a decent signal to noise ratio out of it.

cheers

[idlefaction]

here is a 1496 based ring mod that i have built up at home - it's just waiting on a power supply.  i have a max1044 and some caps and diodes, but i'm lazy.

http://www.ele4music.com/rmp1b/rmp%201b.pdf

[amz-fx]

brett,

I've seen the circuit you are talking about...  it is an amplitude modulator that switches the output on/off at an audio rate...  not exactly the same as a balanced modulator but it has a somewhat similar sound, though a little rough.

regards, Jack

[toneman]

the old Paia 4710 Balancedmodulator has a null trimmer in it.

the new 9710  dual VCA & balanced modulator also has
a modulation null trimmer.

Both of these will take a lot of input signal amplitude.
the 4710 has a clipping indicator.
the 9710 does not.
the 4710 needs plus9&minus9&plus18V.
probably could work on plus/minus12.
the 9710 needs just raw 15Vplus/minus as it has onboard 12Vregs.

by your command
tb

[R.G.]

The output section employs a low-pass filter with an fc of 3.4 kHz to reduce the clock (555) and switching noise. I might need to raise that (to 5 or 6 kHz), and convert it to 2 poles to allow more treble through but still cutting high-frequency clock noise.

555? It's not all that widely known, but the bipolar 555 is a power supply pig. It eats spikes out of the power line and those couple into ... everything...

If the cmos version of the 555 can be used there, it might clean up some of the clock noise.

it is an amplitude modulator that switches the output on/off at an audio rate... not exactly the same as a balanced modulator but it has a somewhat similar sound, though a little rough.

I believe the old Jen Modulator did the same, although with discrete switches.

[ExpAnonColin]

Two quick (well, maybe not) things-

The way that the 3080 and 13700 work are that they are 2 quandrant multipliers.  Ring mods are 4 quandrant multipliers.  With the OTAs, or 2 quandrant, you get A, B, A+B, and A-B.  With 4 quandrant, like the 633, 1496, etc, you get just A+B and A-B.  Remember that you're adding the waveforms, not the frequencies, so with the OTA the carrier is naturally coming out while with 4 quandrant you aren't getting the inputted signal or the carrier signal, just their funky frequencies.

And second, a balanced modulator and a ring modulator are exactly the same thing.  Just different words for the same thing.

Other than that, check out synth DIY pages.  Lots of nice building blocks for this sort of thing.

-Colin

[Peter Snowberg]

And second, a balanced modulator and a ring modulator are exactly the same thing.  Just different words for the same thing.

Ring modulators are balanced modulators, but balanced modulators are not necessarily ring modulators. The ring seems to refer to the use of a diode bridge for mixing. If you look around... some have it, and some don't.

Take care,
-Peter

[Paul Perry (Frostwave)]

The way that the 3080 and 13700 work are that they are 2 quandrant multipliers.  Ring mods are 4 quandrant multipliers.  With the OTAs, or 2 quandrant, you get A, B, A+B, and A-B.  With 4 quandrant, like the 633, 1496, etc, you get just A+B and A-B.

While it is certainly easy & natural to get a 2 quadrant multiplier from a 3080 or 13700 OTA, it IS possible to make a full 4 quadrant multiplier from them, by adding some offset & nulling out the unwnated components.
i remember it was a bit of a mindbender trying to see from the 3080 aplication notes what ws going on (and even more of a strain to actually DO it!)
The CA3080 app notes (1.8meg!) are at www.intersil.com/data/an/an6668.pdf

[ExpAnonColin]

The way that the 3080 and 13700 work are that they are 2 quandrant multipliers.  Ring mods are 4 quandrant multipliers.  With the OTAs, or 2 quandrant, you get A, B, A+B, and A-B.  With 4 quandrant, like the 633, 1496, etc, you get just A+B and A-B.

While it is certainly easy & natural to get a 2 quadrant multiplier from a 3080 or 13700 OTA, it IS possible to make a full 4 quadrant multiplier from them, by adding some offset & nulling out the unwnated components.
i remember it was a bit of a mindbender trying to see from the 3080 aplication notes what ws going on (and even more of a strain to actually DO it!)
The CA3080 app notes (1.8meg!) are at www.intersil.com/data/an/an6668.pdf

Right, that's why I mentioned that the OTAs naturally output the carrier-althoughI should have been more clear about how you can destroy it.  The Moog ring mod uses an OTA and there's no carrier bleed at all.

-Colin

[ExpAnonColin]

And second, a balanced modulator and a ring modulator are exactly the same thing.  Just different words for the same thing.

Ring modulators are balanced modulators, but balanced modulators are not necessarily ring modulators. The ring seems to refer to the use of a diode bridge for mixing. If you look around... some have it, and some don't.

Take care,
-Peter

Yes, but the point is that as people used newer techniques (OTAs, 633's, etc) to create the same effect, they still called them ring modulators, not referring to the diode ring but instead the effect that it produced.  So, the terms are very much interchangeable, even though a 633 based balanced modulator is not a ring modulator.  Sort of like how people can call overdrives overdrives even though the original was produced by overloading your amp's tubes.

-Colin

[R.G.]

It occurs to me that the essence of the OTA ring modulator is to feed the carrier into the high-impedance output node to cancel the carrier. The cancellation is rightly ticklish. The fed-forward signal provides the non-inverted signal compared to the OTA's inverted signal and the linear mix at the output node gives the cancellation.

It might work better if you used two OTAs, one inverting, one noninverting with both outputs tied together into one load resistor. If fed identical Ibias, the contributions are equal, and so there is no output since one output is eating anything the other puts out.

If the Ibias for the two are fed out of phase, the one receiving the higher bias current predominates and the output polarity follows it.  As Ibias reverses, the other polarity predominates.

The nice thing about this is that the Ibias static point can be identical for both if they're fed from opamp buffers as in the CCA article.  One buffer inverts, the other doesn't, and you have a high impedance input to one buffer for capacitively coupling in signal to modulate.

I ... think ... that the inaccuracies in one OTA's feedthrough cancel with the other OTA under these conditions, especially if you use a dual OTA which offers some matching between them.

... just a thought...

[Jason Stout]

... just a thought...

That will take some time to digest..

[ExpAnonColin]

... just a thought...

That will take some time to digest..

When RG posts, it's sort of like you're sitting at the dinner table with a 30lb, tasty, cooked turky in front of you.  It's going to take some time to eat, and if you eat it too fast, you'll throw up, but once you're done you feel quite satisfied.

-Colin

[Peter Snowberg]

I ... think ... that the inaccuracies in one OTA's feedthrough cancel with the other OTA under these conditions, especially if you use a dual OTA which offers some matching between them.

... just a thought...

As long as the inaccuracy is is linear, I would think the same thing. Fantastic observation on the use of a dual OTA in this roll.... that just may be a key to much nicer result. Very cool!

(check out figure 8 as a building block in the CA3280 datasheet. It's digital in that example, but the plots look really nice for linearity. It's speced for use in demodulators too. )

I think it's time to order some chips.

Take care,
-Peter

[Paul Perry (Frostwave)]

The Moog ring mod uses an OTA and there's no carrier bleed at all.
-Colin

Does anyone know whether the Moog ring mod has a noise gate built in?
And as far as RG's posts being like a cooked turkey.. sometimes they make me feel like I'm the turkey    keep them coming!

[Tim Escobedo]

A simple noise gate would cover a multitude of sins in ring mod design.

I had a Mold Spore ring mod, which seemed to be a Craig Anderton LM565 based ring mod with a Vactrol noise gate. This setup worked quite well getting rid of feedthrough.

I suspect the Lovetone may have used a noise gate, too, as they boasted using a "traditional" ring modualtor design rather than a multiplier chip.

[puretube]

those multi-turn trimpots (or what`s the name? , Helitrims) will help a lot in cancelling out unwanted things...