Green Ringer "Nulling" Mod

[Eb7+9]

I wanted to verify what I said in my Impedances post about the Green Ringer - so to try it out I built one up and introduced a 22k pot in series with C2 0.047u ( ... from JD Sleep's schematic) ... well, it works !  Fundamental Nulling is a possibility in the Green Ringer.  You can can insert a trimmer and set it at the center of the null range and add a switch to short it out so that you can have both from a switch only.  I built my pedal with the pot external so I can turn it with my foot - it can be used to give the octaver effect some animation.

http://www.lynx.net/~jc/greenRingerModded.gif

You can definitely hear a nulling going on as the pot is turned from short to around 8k (in mine I measured 7.4k at roughly the null center) ... the disappearance of the fundamental, perceived as a drop in volume body, also seems to alter the perceived quality of the octave a little in the process - it is sweeter, more ringy and drier ... the sound seems a little more focused ... I still like the stock sound with its fair bit of fundamental feedthrough so, the pot or switch then can give you both - a Richer or Leaner octave effect ...

Here's how my idea works :

Most octaver circuits produce their output by adding (mixing) the inverted versions of "harmonically coloured" signals to cancelling out the fundamental (or carrier) and produce a residual harmonic signal that has strong second harmonic content.  The Green Ringer produces its harmonic salad by adding the rectified versions of an inverted and non-inverted replica of the input.  Because this is tantamount to full-wave rectification the output waveform cannot consist of a pure second alone and so this circuit can be classified as a Dirty-2nd Octaver, as opposed to circuits which are intended to produce a pure 2nd harmonic.  Still, the Green Ringer circuit can be tweaked so that the fundamental is nulled out leaving behind only the harmonics - like the circuit is inherently meant to produce.

Let's go through the stock Green Ringer circuit.

The Input gain stage centered around Q1 acts as a voltage inverter with a gain magnitude of roughly 3v/v ... I used more common values to bias this gain stage in my pedal (R1=470k, R2=120k, R3=15k, R4=4k7) - they preserve the gain and headroom specs to within 10%.

The output gain stage based around Q3 is an emitter follower, or voltage buffer, with a voltage gain just below unity.  What interests me is the input impedance seen by the circuit at the base of Q3, which is determined by all AC loads in the emitter circuit and the hfe of the device as it sits in the circuit.  Assuming the output isn't loaded significantly then the parallel combination of the 10k and 4k7 resistors (ignoring the 0,047u cap) leaves a combination value of around 4k - 4k5.  For an average bipolar device with a small-signal current gain of 100 (ballpark) that leaves us with a reflected impedance of 40k to 45k seen looking into the base of Q3.

Q2 is wired as a unity-gain heterodyne circuit where the "unloaded" voltage swing at the emitter and collector should be equal in magnitude - that's provided the two 10k loads are well matched.  These two signals are passed through equal capacitors and matched diodes to the mixing node at the input of Q3.  Notice that the diodes are barely biased "on" by the base current of Q3 - I used 1n60 Germanium diodes and one measured 19.7mV of quiescent voltage across it while the other measured 15.3mV - both barely on.  The reason for biasing the diodes, which is what makes this circuit smooth sounding in my opinion, is that it's not up to the signal to supply ALL the current to charge-and-discharge the diode parasitics before it can change states from on to off and to on again.  This "trickle bias" makes for faster and more seamless diode turn-on - very smart !

Now, it would seem that if all the components are measured and matched everything should be fine and a clean effect would be produced - but there's one thing that's been overlooked.  Namely, the balancing of driving point impedances at the source nodes due to Q2 transistor action.  The DPI at the collector looking into it consists of the collector load in parallel with the DPI of the collector itself, which is typically high (100k-1Meg) - so it is dominated by the collector resistor which is 10k-ohm.  The DPI at the emitter is dominated by what the base sees devided by the current gain of the device hfe.  The base is basically looking at a collector circuit that is dominated by a 15k load and the hfe oif Q2 can be assuned to be 100 (ballpark) - this gives a DPI of 15k/100=150-ohms.  So the DPI at the emitter is 150-ohms while at the collector it is very roughly 10k, the DPI at the emitter is much lower and so its signal component will add in a much stronger way than the sigbnal component at the collector.  

If the mixing doesn't occur in an even manner than the full-wave rectified waveform will have half-cycles of uneven height and a fundamental will register through this pattern.  Above I estimated the equivalent mixing node impedance to lie around 45k so it's easy to see that with the 68k bleeders and 10k DPI at the collector source that a DPI of 150ohms doesn't jive - it's way too strong "in relation".  If we add resistance (somewhere around 10k) to the emitter circuit then we've made it more compatible with the other drive node, and the mixing circuit impedance levels without changing other aspects of the circuit.

Because the Green Ringer produces dirty-Octaving through a process of full-wave rectification it follows from a simple Fourier argument that because of the shape of the FullWave waveform produced a pure second cannot be produced at the output ... so instead, along with a dominant 2nd harmonic other higher harmonic terms will be present at the output.  The "Nulling" Mod that I present here allows the tuning out of the fundamental (carrier) so that upper harmonic terms can be heard more on their own.  This idea can be adapted to other Octaver units that employ a similar "mixing" principle to achieve production of harmonics.

copyright JC Maillet (c) 2003

[B Tremblay]

Very intriguing!  Thanks for sharing this!

[bwanasonic]

Thanks for the Green Ringer info - inspired by this, I grabbed mine and swapped out 1N914 diodes I had in there with some NOS germaniums from smallbear. My unscientific conclusion is it sounds "more betterer"  :wink: A bit less *gatety* and smoother - maybe. Didn't do a real A/B, Just went for it. The question I have about the GR is, what is the main factor in it's overall frequency response? Mine has a bit of a notched wah tone that responds mostly above the 12th fret. I'd like to rock the "wah pedal" back a hair so it resonates a bit lower and responds better around the 7th fret if possible.

Thanks again
Kerry M

[moosapotamus]

:shock: Um... wow!?!

jc - Nice mod... awesome analysis. Time to build another GR.  

bwanasonic - If you put a simple, high gain booster in front of it, I think it will respond much lower down on the fretboard.

~ Charlie

[Tim Escobedo]

I often find a trimmer quite beneficial to such circuits, allowing nulling to really be maximized. Some circuits are such that they offer results that are "good enough" using fixed values. But some methods really need to be "tuned" for maximum effect. Here the magic trimpot comes to the rescue.

[Eb7+9]

The question I have about the GR is, what is the main factor in it's overall frequency response? Mine has a bit of a notched wah tone that responds mostly above the 12th fret. I'd like to rock the "wah pedal" back a hair so it resonates a bit lower and responds better around the 7th fret if possible.

Thanks again
Kerry M

... aside from the rectifier section Fourier'ing out harmonics there's nothing else to shape tone in the signal path ... I'm not sure, if you're talking about effect sensitivity or effect quality up and down the neck it seems like a lot of octaver circuits respond this way ... just remember that the amount of fundamental nulling that occurs by trimming the added emitter-resistor in the Green Ringer will depend on how well matched the other "parallel" components are to each other ...

With Octavers I tend to hang around either the 12th fret and play strings against each other there (with lots of dual string vibrato or whammyism) or low strings near the nut with a mutted palm for a keyboard effect - I get a nice strong mutted effect at the bottom of the neck with the Green Ringer so I don't think this octave effect is any less sensitive there or anywhere else ... with the nulling I notice a quality in these two sounds right away - there's more ring bite it seems ... but I agree the "in between" region might sound challenging a little, I think a varied pick attack is the key there - a booster like Charlie said is probably a good idea too ...

From talking about this with Mark Hammer this summer we came to the idea that this sensitivity-perception issue in octavers might be related to a "frequency difference" term that is heard when playing up high on the neck but disappears into sub-audio land when playing below a certain point ... It's fairly easy to provide analysis for this in a multiplier based circuit but for a FW-rectifier based octaver I'm not sure how to argue that such a circuit automatically creates spectral side-terms at F+F and F-F like a multiplier circuit does ... the F-F term typically cancels to becomes a residual DC term except when the cancelling is not perfect ... if imperfect cancelling of the difference term occurs then an added low-frequency "component" would be heard as part of the output for notes that are high enough in pitch ...

100% pure conjecture ... jcm~

[Eb7+9]

The question I have about the GR is, what is the main factor in it's overall frequency response? Mine has a bit of a notched wah tone that responds mostly above the 12th fret. I'd like to rock the "wah pedal" back a hair so it resonates a bit lower and responds better around the 7th fret if possible.

Thanks again
Kerry M

... aside from the rectifier section Fourier'ing out harmonics there's nothing else to shape tone in the signal path ... I'm not sure, if you're talking about effect sensitivity or effect quality up and down the neck it seems like a lot of octaver circuits respond this way ... just remember that the amount of fundamental nulling that occurs by trimming the added emitter-resistor in the Green Ringer will depend on how well matched the other "parallel" components are to each other ...

With Octavers I tend to hang around either the 12th fret and play strings against each other there (with lots of dual string vibrato or whammyism) or low strings near the nut with a mutted palm for a keyboard effect - I get a nice strong mutted effect at the bottom of the neck with the Green Ringer so I don't think this octave effect is any less sensitive there or anywhere else ... with the nulling I notice a quality in these two sounds right away - there's more ring bite it seems ... but I agree the "in between" region might sound challenging a little, I think a varied pick attack is the key there - a booster like Charlie said is probably a good idea too ...

From talking about this with Mark Hammer this summer we came to the idea that this sensitivity-perception issue in octavers might be related to a "frequency difference" term that is heard when playing up high on the neck but disappears into sub-audio land when playing below a certain point ... It's fairly easy to provide analysis for this in a multiplier based circuit but for a FW-rectifier based octaver I'm not sure how to argue that such a circuit automatically creates spectral side-terms at F+F and F-F like a multiplier circuit does ... the F-F term typically cancels to becomes a residual DC term except when the cancelling is not perfect ... if imperfect cancelling of the difference term occurs then an added low-frequency "component" would be heard as part of the output for notes that are high enough in pitch ...

100% pure conjecture ... jcm~

[bwanasonic]

I'm not sure, if you're talking about effect sensitivity or effect quality up and down the neck it seems like a lot of octaver circuits respond this way ...

Yeah, I know the *sweet spot* behavior. If I really want to freak it out, I'll play an arpeggio of perfect fourths at  the 12th fret. Maximum Ring Mod effect. I'm trying to describe more the effect quality. If I play "Who Knows" at the 12th fret, it sounds like a slightly sped-up cartoon version of BOG. Well. the Roger Mayer Octavia is on my to do list...

It's fairly easy to provide analysis for this in a multiplier based circuit but for a FW-rectifier based octaver
I'm not sure how to argue that such a circuit automatically creates spectral side-terms at F+F and F-F like a multiplier circuit does ...
the F-F term typically cancels to becomes a residual DC term except when the cancelling is not perfect ... if imperfect cancelling of the difference term occurs then an added low-frequency "component" would be heard as part of the output for notes that are high enough in pitch ...

:!: :?:  What is the sound of one head spinning?:wink:
Kerry  M

PS- Frank Zappa had his tech build a Green Ringer into at least one of his guitars. Anyone know of a recorded instance of him using it?

[Tim Escobedo]

Part of the effective octave up tone comes from having a input signal with the least amount of harmonic content. Typical formula is:

Play around the 12th fret
Use neck pickup
turn down the tone control
pluck the strings somewhere close to where the center node would be[/list:u]

These things de-emhasize the higher harmonic content of the typical electric guitar signal. As a result, fewer of those harmonics get multiplied, which means a "clearer" octave up tone with less buzz and fizz. The further you deviate from these conditions, less clarity you'll have. Depending on the conditions, this can mean more fizz and/or fundamental at the output.

If this is what you mean when you say "frequency response".
:wink:

[moosapotamus]

Part of the effective octave up tone comes from having a input signal with the least amount of harmonic content.

That's interesting... makes sense, too. So, would a band-pass filter do this too... reduce the harmonic content more effectively (or rather, more selectively) than turning down the guitar's tone control, and allow a stronger, more distinct 'octave' effect?

~ Charlie

[Tim Escobedo]

I'm guessing a bandpass might present problems if it causes any ringing in the waveform it's filtering. Such as the typical wah-type BPF. However, it's a interesting idea that I've never tried.

[brett]

Hi.  I've tried low-pass filtering before rind-modulators and octaves.  Cutting out stuff above, say 4kHz definately cuts down on those high bell-like tones.  So it's purer octave...but it also sounds kinda "flat".

I found this discussion interesting because I once tinkered with a Roger Mayer octave, and set it up to have less-than-normal fundamental.  In the end, I decided that it sounded worse than in original condition.  As RG said in a recent post, the human ear is sensitive to even a small amount of octave-up, so the 25:75 mix of fundamental and octave (or thereabouts) that comes out of a RM octavia SOUNDS like HEAPS of octave over a small amount of fundamental.  To me, more octave up and less fundamental seemed like a good idea, but when I heard it I realised that I was kinda missing the point.  

However, I haven't built a Green Ringer - maybe it puts out heaps of fundamental (like 90% or more), and so there'd be heaps of room for more octave than in an RM or Tyco.

This has been an interesting thread.  Makes me think about going back to the RM and turning the octave down to 50:50 or thereabouts (I find that both the Tyco and RM are both a bit "full-on" for regular use).

thanks for the insights

[Eb7+9]

pluck the strings somewhere close to where the center node would be
[/list:u]

I tried and it works good ... by moving my picking hand down it totally cleaned things up around the 7th fret ... when playing around the 12th fret then normal picking position is already about half way :!:  :!:

that's kewl Tim, what a great tip !  

thnx ... ~

[GFR]

You may find the article below interesting:

http://www.geocities.com/gfr.geo/octave.html

[Mark Hammer]

It is now about 12:20 as I start to write this.  At roughly 10:00PM, I stopped watching the "West Wing" (a favourite of policy wonks), took the stack of schematics I was perusing during commercial breaks and sauntered into my office/shop, determined to make a Green Ringer.  In approximately 2hrs 20min I have a working one, built on perfboard, and installed in a 1590B, true bypass, wall wart jack, battery snap and all.  All that remains is to paint and label it tomorrow.  I wouldn't have believed it could be done unless I saw it myself.  If the problems with administering my site were ironed out, I'd have posted a picture.

I built it as per JC's schematic.  The only substitution was replacing the 7.4k resistance with a 10k trimpot (if I had dug around for a 22k, it would have taken 2hr 30min), and replacing the 47k output resistor with a 50k level pot (I had all that chassis space staring at me and just had to put *something* there).  Transistors are 2N5210 and 2N3906 as per spec, and diodes are some generic unlabelled Ge (likely 1N34's) that read about 250mv forward voltage.  The 2N3906 read an hfe of around 76 so I'm wondering if that might be a tad low-ish.  I have a mountain of suitable PNP Ge's, courtesy of Mike Irwin, so I may try one of those in there, or perhap another Si PNP.  In retrospect, wish I had socketed Q2.

It works fine, though I haven't discovered the "sweet spot" on the trimpot yet.  I would not describe it as my favourite octave-up unit.  That place is presently reserved for the Foxx Tone Machine.  Certainly not raspy, though.

Through dumb luck, the 5% resistors I had for R5 and R6 were matched well within 1% (both read 9.79k on the meter, if you can believe it).  Bernie Hutchins indicated in Electronotes years ago that use of a bipolar for a rectifier-like circuit such as this would work better if the functional equivalent of R5 was just a hair larger value than the equivalent of R6.  That is, the two resistors should be *almost* but not exactly identical.  Since it's late, I'm heading to bed, but I may try sticking a 220R resistor or something similar between R5 and V+, just to see if it makes a difference, although it may well be the sort of "improvement" that Rbal is intended to address already.

Thanks JC.  Thanks Dan Armstrong.  Thanks Mr. Step-bit.

[Eb7+9]

Mark ... if you pluck a string and roll the 25k/22k pot you should hear the bottom coming in and out a little as you pass near the center - it's pretty wide and subtle ... depends on how well matched the rest of the feeds are ... I made sure to measure the caps mine and chose two that were the closest to each other ...

[Mark Hammer]

I'll try that tonight.  I was trying it out with a little 386/3"-speaker battery-powered practice amp last night at low volumes; probably not the optimal arrangement for identifying sweet spots in *anything*.

[Steben]

Verrrrrrrry old thread I know I know. (but a fraction of our age)
I've put this beast in a sim. But with an opamp as first stage, driving a phase splitter BJT stage directly. I guess a lot of the nulling problem is gone. The original BJT stage is very non linear, which is bypassed with a linear stage an opamp is ment to be.

Is this over due or did it already popped up??

[Mark Hammer]

Not that it is a terrible idea, but sims assume that a nominal value is the true mathematical value of a component; something which might not be the case in "real life".  JC's mod is to compensate for what may be the case (and unpredictably so) in real life.

The trick to getting an easily audible octave is equal-amplitude signals from the emitter and collector outputs, such that when they are "folded over" two equal-amplitude peaks are created for each peak that initially existed.  If the doubled peaks are not the same amplitude, we don't hear it as a doubling in pitch.

There are a variety of ways that one can arrive at rectified signals of equal amplitude.  The compensating trimpot illustrated is perhaps the simplest way, since it doesn't involve measuring or matching components, simply listening and tweaking.

[Steben]

Not that it is a terrible idea, but sims assume that a nominal value is the true mathematical value of a component; something which might not be the case in "real life".  JC's mod is to compensate for what may be the case (and unpredictably so) in real life.

The trick to getting an easily audible octave is equal-amplitude signals from the emitter and collector outputs, such that when they are "folded over" two equal-amplitude peaks are created for each peak that initially existed.  If the doubled peaks are not the same amplitude, we don't hear it as a doubling in pitch.

There are a variety of ways that one can arrive at rectified signals of equal amplitude.  The compensating trimpot illustrated is perhaps the simplest way, since it doesn't involve measuring or matching components, simply listening and tweaking.

Yes, that is very true.
Yet the original BJT stage is non linear nonetheless... Of course it is 1% talk.

And it does give the option of adding a soft clipper in the loop.... Giving birth to a smooth "dynamic" compressor you talked about.