An oldie but goodie - Anderton ring mod

[Mark Hammer]

I was listening to an older Jeff Beck album - You Had It Coming - today, and he makes excellent use of a ring modulator in the tune "Roy's Toy" (Try around 2:10 - https://www.youtube.com/watch?v=xLrshCi1vHs ).  That inspired me to dig into the barrel of unfinished pedals, and pull out the board for the old Anderton EPFM ring modulator, built around the LM565 chip. (Project #9: https://msu.edu/~dougl126/Electronic%20Projects%20for%20Musicians.pdf )

If you've read any of my posts regarding RMs over the years, one of the things I tend to natter on about is that ring modulators started out in the synth world using oscillators, and often sine-wave oscillators.  The sum and difference sideband products derived from combining such waves would be easily perceivable in the absence of extraneous harmonic content.  Once you diverge from sine waves, though, you start getting sums and differences with the modulating frequency and all the various harmonics in the input signal.  And that tends to create a lot of hash that buries the sum and difference with the fundamental.

So I decided to "improve" the unit this evening and fire it up. 

The first thing I did was to limit the high end at the input stage and at the mixer stage (see schematic on page 102), by adding a feedback cap at each stage., and rolling off top end around 3.5khz.  I figured there was no need for  any content above that from either the clean or modulated signal paths.

The path from the input stage to the 565 chip consists of a 33k/6k8 divider, with a back-to-back diode pair (iN4001) in parallel with the 6k8 to ground.  I did what I have often done in delay-pedal feedback paths, which is to  split a larger fixeed resistor into component values, and run caps to ground to make one or more lowpass filter sections.  In this instance, I split the suggested 33k resistor into 10k, 10k, and 12k (in that sequence).  After each 10k, I ran an 8200pf cap to ground.  This rolls off top end just under 2khz.  With only two poles of lowpass, there is still some upper treble reaching the 565 chip, but the majority of potential fundamentals are more prominent/obvious,with harmonic content pushed way in the background.  As Craig suggests, I made the cap value that sets the modulation range, larger, to keep the sidebands within a reasonable distance of the fundamental

I don't know that it is perfect...yet...but I found the unit more "musical" with the top end severely limited.  Anderton provided us with a decent basic design.  The modifications adapt it better to guitar, so that it sounds more like a rubber band, and less like a big cymbal.

If this interests you, note that JD Sleep has graciously provided re-draws of the PCB layouts from the EPFM book, replacing 14-pin 4739 dual op-amps in the layouts with standard 8-pin dual op-amps.  http://www.generalguitargadgets.com/effects-projects/extras/epfm-book/

[BetterOffShred]

It interests me    I love Jeff beck, and ring mods.  Thanks for sharing and I'm going to check it out!

[sergiomr706]

Hallo, video sounds amazing, was the modified CA Ring Mod sounding more o less like that? If so, thats a winner!
when you say,

"The first thing I did was to limit the high end at the input stage and at the mixer stage (see schematic on page 102), by adding a feedback cap at each stage., and rolling off top end around 3.5khz.  I figured there was no need for  any content above that from either the clean or modulated signal paths."

are these capacitors parallel to R19 and R17? which values would that be? Because I see how you calculate low pass when C follows R and C goes to ground,
like in your next step,
" In this instance, I split the suggested 33k resistor into 10k, 10k, and 12k (in that sequence).  After each 10k, I ran an 8200pf cap to ground.  This rolls off top end just under 2khz".
but I m a little confused when C and R are paralleled.
Thanks in advance

[Mark Hammer]

I put 1000pf in parallel with R14 (47k feedback resistor in the output/mixing stage), producing a rolloff around 3.4khz on the final output.  I replaced R17 (normally 100k feedback resistor on the input stage) with 470k for 5x gain in the front end, and put 150pf in parallel with that resistor for a rolloff of both clean and effect around 2.2khz.

I'm going to experiment later this morning with a series back-to-back pair of Schottky diodes ahead of R8 (the mixing resistor for the effect signal).  My hunch is that it may be able to gate out any of the remaining oscillator whine that the nulling control doesn't fully eliminate.  It will introduce some crossover distortion, but in the face of all the other changes to the signal that a ring modulator produces, I doubt it will be noticeable. 

[stringsthings]

I was playing with this circuit a few years back and I had one question about Craig's description
in the "How it Works" section. 

"R1 and R2 drop the power supply voltage to about +-6V for feeding IC2 (LM565) .."

I measured much closer to +-9V running on 2 9V batteries.
Did I mess up the circuit?  Or was his description wrong?
And why would he want to drop the power supply voltage?

[Mark Hammer]

1) Did you measure at the  IC pins, or at the diodes?

2) I gave the LM565 datasheet a quick scan to see if there was anything that jumped out at me, regarding recommended supply voltages, but my eye is not well-trained enough to see anything of that sort.

[sergiomr706]

Thanks a lot, i ll have to try this one. Looking at the schematic, would this work turning it into voltage, volt reference and ground? Or is +-9v necessary?

[stringsthings]

1) Did you measure at the  IC pins, or at the diodes?

at the IC pins.

[Mark Hammer]

In the book, Anderton notes that the recommended voltages (and the components that provide them) are important for keeping the carrier feedthrough low.

[Mark Hammer]

So, I just finished booting up the one I've been finishing off.  I made up a little daughter board with a TC1044 charge pump to provide the necessary +/-9v out of an external +9v supply.  I did the pin 1-to-8 bridge on that chip to raise the clock on the 1044 safely out of range.

I inserted a back-to-back pair of 1N5817 schottky diodes in series with the 4k7 mixing resistor for the ring modulator side to provide a gate, or rather a "hurdle", past which the low level whine from the carrier would not pass (assuming the nulling trimmer had already been set optimally).  Worked like a charm.  The unit was dead quiet.  No whine.

Nothing particularly special about the Schottkys.  I selected them simply because I figured the background whine was small enough that it was unlikely to exceed 150mv.  If one has a harder time nulling out the carrier, for whatever reasons, you can use a pair of germaniums with a higher forward voltage, or sift through your pile of Schootkys for a pair with the highest forward voltage.

I should note that the diodes DO take away from the ring-side signal amplitude, although there is still plenty of signal left.  On mine, I installed a gate/no-gate toggle to compare.  The drop in level with the gate is easily noticeable, but I can still set the ring level well above bypass level, so minimal harm done.  And, as noted in my earlier post, if it was a "cleaner" effect, one might notice the coloration introduced by the crossover distortion.  But given the clanginess of the effect itself, any crossover distortion is undetectable.

Nice to have a more musical quiter version of this project.

Ring on.

[Digital Larry]

I've asked this a few times, but I wonder why active op-amp full wave rectifier circuits, which would seem likely to do a better job simpler, are not favored for this application.

[Mark Hammer]

I'm not sure in what way you mean.

That said, the circuit is over 35 years old, well before the web allowed people to forward suggestions, and I imagine Craig used what he used.  I certainly won'tclaim there isn't better out there.  I just boxed up something I had laying around for years.

[stringsthings]

The improvements sound very good.  I may have to dig my old pcb out and
give this one a try.

[Mark Hammer]

Turning the volume up, I will make a retraction and say that it isn't dead quiet, but much quieter when I'm not playing.  Maybe using diodes with a higher Vf (like Ge or even Si types) would reduce audible whine even further.

Still, ring modulators are not exactly set and forget effects that one leaves on, even when not strumming.

[Digital Larry]

I've asked this a few times, but I wonder why active op-amp full wave rectifier circuits, which would seem likely to do a better job simpler, are not favored for this application.

Ah woops sorry, this comment should be about an octave fuzz.  My bad, too much or too little coffee.   

[Mark Hammer]

Or bad aim.  That's what I tell people when they wonder why our two kids are ten years apart. 

[~arph]

You can use a XR2206 too if you want to do a nice ring mod with sine wave modulation and I'm not talking about using the XR2206 as just the oscillator but also as the modulator. Dead simple circuit too.. too bad the XR2206 is also obsolete and I have not found anything modern to replace it with.

[Mark Hammer]

There was an old XR2206-based ETI project called the Soundbender.  ( www.fenris.talktalk.net/images/Sound%20Bender.pdf )  I actually happen to have one of those etched and stuffed, sitting in the bin-of-boards, but not wired up.  Its chief virtue seems to be that one can select a sine wave carrier, in addition to the triangle that the EPFM provides.  I have to do some grocery shopping, but if there is time left today, I might try and fire it up.  It wants a 12v supply.

[~arph]

Wow never saw that eti soundbender. Really  similar to what I came up with. As always I have been predated by over 30 years.. one thing I can assure is that it will run fine on 9v. I will try the mixer stage proposed here as I can get some phase cancellation in the simpler blend setup I used.