Feel the Music: Questions About Feedback Mechanisms

I like the question, and to be sure that you know where my answer goes, I will sumarize. A device that picks-up the string vibration, amplifies it and returns to the string, as excitation, in order to maintain a indefinite oscillation, identical with the initiator oscillation.

Letțs talk about one string. By pressing the left hand fingers in the right spot on the string (in order to get this end anchored) you create a sound. It would be a single fundamental frequency (first harmonic) and some others, depending on the ghitar configuration, wood, how hard is the string presset and other factors tha you, surely, know better than me. If you look at the oscilloscope, you might not see a sinusoid but a periodic wave with a shape that differ from it. Electronically, in order to reproduce that oscillation identically, your capting sensor has to be a wide band one(not only the 16 to 15,000 Hz), as well as your amplifier. The periodic wave that we were talking about, might have a multitude of harmonics, starting with the first one (or fundamental) and going up to n-th (that means n multiplied with the fundamental frequency) if the sound started abruptly (therefore a very fast transition from the silence to sound). To reproduce it perfectly, all the chain must have a wide band capability.

Is the string capable to reproduce the (possible) complex wave? My common sense tells me that, if it was possible to generate the sound, it would possible to sustain, by feedback the same one. If you try to do the trick with six cords at the time, unless you create a similar signal (as spectral components content), it is not possible to sustain six different signals.

I think that your idea of individual pick-up coils and individual exciters is a good one.

But what do I know? I played trumpet in high school.

I guess the system has to be wide band though, as its capable of sustaining any one individual string. It's just that when all are played simultenously, one will end up being more dominant than the others.

Phase relationships become significant when attempting to sustain oscillation in an amplifier. Phase relationships in an amplifier or a system including an amplifier and other components may not be consistent for all frequencies. Consequently, some signals may be preferentially sustained and create feed back where as others do not meet the required phase relationship. It may be possible to parallel several amps (properly mix the outputs!) with staggered phase relationships so that one is always responsive to the desired signal and the composite is fed to a common amplifier in order to circumvent the need for individual string sensor/actuator elements. I would experiment with four amps with different propagation delays and see if it produces the desired polyphonic response. If you try this please let me know how it works out for you.

Yeah, this is quite noticable with the ebow (movable stand alone unit - single string sustainer). For any given note, there are specific "hot spots" and "dead spots" where the string is being driven in phase or out of phase.

I like the idea of paralleled amps with staggered phase relationships. The thing is, the one amp which has a fixed filter characteristic (and hence deterministic phase response) is capable of sustaining all notes on all strings to some degree (some more efficiently than others). I guess as long as the driving effort isn't completely out of phase with the string's oscillation, the exciter can do work on the string and hence continually feed it energy.

Intuitively, if the one amp is capable of driving any of the strings, then a combination of the string signals being transduced simultaneously should superimpose and excite the strings simultaneously. Each string would see six superimposed signals trying to excite it though, and so somehow this means that one string preferentially resonates over the others.

I may try independent amplifiers and see if that works, as it's relatively easy to test.

Does anyone have any good feedback/control system analogies that may shed light on this problem?

Hi rcapper

Phase relationships become significant when attempting to sustain oscillation in an amplifier. Phase relationships in an amplifier or a system including an amplifier and other components may not be consistent for all frequencies.

Do you agree that tuning could be part of stabilising these Phase relationships? Even with an acoustic instrument, these play a major role. Why not in this case, were the dudes get their rocks off.

A lady bird singer of mine (way back then) had the gab to kill my voice with a certain overtone that only she could create/sing. Although her input was as lovely as, it just eliminated my vocal cords. Just the one tone and my voice would be zero. Her sound waves cutting mine out. Even after rehearsing many times she could make it happen. Would that be one of the occasions were Phase relationships went wrong? I still wonder about it.

But then again my singing could have been out of tune. Ever so slightly. Hope you get there guys. Ky.

You may be seeing the affect of positive feedback and filtering. As Indel implied, the market exciters are not going to produce all the components to exactly reproduce the sound. What will happen is the fundamental frequencies are going to be preferentially introduced. Over time the sub-harmonics will get muted, and possibly destructively interfered as they vary in their amplitude and have some but not all re-inserted. The fundamentals will also cause harmonic excitation in the neighboring strings so that this "preferential" sustaining will cause positive feedback to the harmonics.

The solution would be calculate the number of harmonics to identify, and correctly reproduce for the length of time you are interested in maintaining the sustain. Needless to say the longer you want the sustain to be accurate, the more harmonics you will need to reproduce. This will quickly become an acemdemic exercise as getting too deep in the harmonics will not be feasible for the pickup, amplifier and possibly the driver.

Having individual pickups, very wideband amplifiers, selective filters (mechanical or otherwise on the pickups) to reduce the impact of neighboring strings, and individual drivers will reduce the depth of harmonics you will have to identify and reproduce. But this may not be simpler or cheaper.

Dear Guest

I just left from a post at CR4 were automatic tuners are involved. This is maybe part of your problem. A Guitar will not "sing" if it is not tuned. The un-tuned instrument could be giving "false" signals which would be hard to correlate and for the electronics to read. Try tuning to the max and see what happens. Certain waves cancel out others.

Jimi used to tune more than anyone on stage and he is still hard for some to imitate. ( I know it's completly impossible but there is hope). He didn't even have the gimmicks mentioned. What he also had was some kind of electro static thing going, you could see the sparks fly if you know what I mean.

Our lead guitarist (1970) used to play with his instrument un earthed so to get and trigger this excitement. In the slower parts he would be standing on a foam mat.

What ever you do, don't do that. Ky.

Dear all,

our human ear is not phase sensitive, so forget about phase and phase relations.

There is only one seeming exception: very sharp pulses.

In your basic description: "pickup, gainstage, magnetic driver" there must be an additional component that establishes the damping to be slightly above zero, else the loudness will rise until the gainstage or magnetic driver satureates.

If your pickup is a very good quality microfone this would hear all vibrating strings plus the noise from everywhere. If you add a digital active filter stage to identify at which tone any individual string is vibrating at this moment , then store the frequencies and amplitudes and excite with these this would be then near ideal setup. The exciter will be the worst part introducing harmonics (as loudspeakers are the worst contributors of harmonics).

In the existing models the actually existing total vibration (of the air or ghitar body) are measured and at some place the excitation is coupled into the system.

This is deteriorating the total quality. But in principle a multi-tone exciter is possible.

Any loudspeaker is performing this task: to excite our hearing system at many frequencies at the same time.

RHABE

"our human ear is not phase sensitive, so forget about phase and phase relations."

So at what point does someone's ear get involved in the feedback loop to determine whether feedback is sustained? I don't see as how it is a component in the feedback loop.

Are you saying that phase relationships do not affect oscillation in an amplifier?

The answer you are seeking is contained within your question. As with any acoustic system (guitar body, neck, nodes - nut, bridge, and main resonating surface: the top) there are natural places in a span of frequencies that are resonant points. This means the the system is very sensitive to a specific frequency and its related harmonics, and not so sensitive to others. The same is true of electrical circuits...the challenge for sustain/feedback circuit designers is to pick up the frequency you generate and then perform the work without having the system go so wild it burns up. Resonance in the circuit and resonance in the mechanical system may not relate well to each other - so the circuit can perform better when the physical system is matched harmonically. Since the circuit designers cannot possibly know the physical resonance of the system you are using, they have a problem making the electrical resonance respond equally througout the range of frequencies possible on a musical instrument.

The other electrical piece of this equation is contained in the "old" way of generating feedback - the volume setting, circuit resonance, and sound loop distance between you and your physical amplifier...it's a 3rd factor for success or failure when trying to generate that long, hot, satisfying sustain you get with an overdriven circuit.

Hope this helps - if you have an oscilloscope, you can look at the basic waveform of your instrument and see it's circuit resonance points. Put a white noise generator on the pickup and look to see where the signal peaks in spite of the uniformity of the signal.

Wonderful discussion:

Just a few areas to review here with EM. The phase of a component is analogous to the distance, frequency analogous to velocity, and acceleration analogous to velocity of frequency. In case you have not had physics, velocity is the derivative of distance over time, and acceleration is the derivative of velocity over time. Simply put: the change in distance over time is velocity, the change in velocity over time is acceleration.

With this in mind one can differentiate the phase to obtain the velocity, and differentiate the velocity to obtain the frequency's velocity (or phase acceleration). This is very interesting when playing with MathCAD or Matlab to see the results, especially when considering resonances.

Simply put, the entire electronic circuit and physical aspects have a characteristic impedance (complex resistance, if you will) that will explain the resonant frequencies--some stable some not. With an amplifier, pre-, and post-circuitry each have their own resonant freqencies, but combine to one large characteristic equation. To resonate is to tweek the entire circuitry for a specific frequency (OR NARROW BAND). The pre- and post-circuits can have inidividual resonating circuits tuned to the narrow band of each string on a guitar. Whereas the amplifier, more costly and power hungry circuit, typical will amplify a band of frequencies, but will have standing wave ratios for input and output as well as impedance mismatches--all of which tie into the characteristic equation.

So the answer to your question is this: Separate circuits can be built for each string, but that is not economical (but perhaps better). Due to economics one may use individual passive circuits such as a tank circuits to enable resonating frequency bands at fixed distances.

Even more interesting is the use of a comb filter or notch filter to keep the strings frequency bands separated. See wiki at http://64.233.167.104/search?q=cache:rXGIVLzeK34J:en.wikipedia.org/wiki/Band-stop_filter+wiki+%2Bnotch+filter&hl=en&ct=clnk&cd=1&gl=us. Enjoy!

BTW: the ear can hear down to 10ms delays, where the perception is spatial acoustic spreading such as that with "hall", "wide", and other psycho-acoustic effects using around 10ms to 40ms delays. These delays will typically incorporate several cycles between delays, typically in the mid to high frequencies.

I'm not sure why your post was in reply to me, but it does contribute to the general discussion.

Perhaps there is a DSP solution here...how about a feedback device that detects each frequency on the fly for every note being played. It then performs an FFT and based on numerical values produced by an algorthm for the harmonics of each note plus the change of amplitude of the harmonics as a function of time and the envelope of the signal, knows how to throw in just exactly enough "push" to create the illusion of feedback (overdrive) by synthesizing the necessary frequencies at the correct amplitudes. Oh, it's already been done! It's called a guitar synthesizor. I just bypass the whole durned thing by playing my music through a Yamaha FX6 and telling it what to tweak and how much to tweak it.

HMMM...wonder if I've ever learned any Physics?

Thanks for the insight guys. I guess it makes sense that the dominant resonant modes will cause one string to win out over the others.

I also have a guitar synthesizer. I was going to attempt to feed an exciter two pure sinusoids to see how effectively this could resonate two strings simultaneously. If this is effective, then I could use a guitar synthesizer to drive the exciter and thus achieve polyphonic sustain.

One of the problems with this approach is, its a bit sluggish. Guitar synthesizers (FFT) aren't great at tracking, and so the system wouldn't feel as tactile as a true analog feedback loop.

There's no feedback like natural amplifier feedback. It's the only one that "screams". Of course it's only one note, that's what resonance is, single frequency (at a time).

Actually, these "artificial" sustainers are well capable of screaming when they're set to their harmonic modes (which presumably employ a high pass filter so that a higher harmonic mode is excited instead of the fundamental)

I agree with Parsec...there is a tactile delay or loss when processing through all that gear to get polyphonic feedback. I've been thinking about the elements of that hot, overdriven sound - think from an aural standpoint it has three main components:

1. The amplitude envelope is reshaped...when the feedback starts, the normal decay of a normal note on a guitar is now stretched from a point just after initiation to the end when the musician changes the pitch.

2. There is a change of harmonic emphasis ...the harmonics present at the moment the feedback kicks in are very prominant, so instead of diminishing, they add energy to the to the fundamental which in turn generates more of the harmonic.

3. At the beginning and end of the feedback pitch, here is that clippy, teeth chattering white noise that might be a super amplification of the pick noise. This is rarely present in an artificially induced feedback system.

Since we are creatures of habit when it comes to music, I wonder if listeners would "buy it" if you produce feedback on pitches that aren't related to the tonic or dominant of the music being played.

Now you had me thinking. Cause I'm genuinely interested in a feedback system that would sound like the real thing without any big speaker involved. My sound guy hates a loud guitar amp that he can't control, so if I get the feedback without the loudness everybody would be happy, I could play in headphones and still get a big sound.

I finally got around to plugging my guitar into a CRO and having a look at some of the signals.

It's interesting to see that a plucked string is very harmonically rich, whereas the sustained string tends to be a pure sinusoid, which is probably why sustain sounds so much thinner than plucking a string. I will try synthesizing these additional harmonics to see if I can create rich sustain that mimics the initial string pluck transient.

I tried using a hexaphonic pickup (one signal per string) and routing each of the signals through a mixer. I can achieve polyphonic sustain by balancing the signals right, but it's a bit knife edge.

The multiple sustainer idea seems to do the trick though (I used my sustainiac in tandem with a power amp driving a bass shaker).

I'm going to try the six pickup signals (per string), six amplifiers and a single driver. I can't think of a good bridged configuration for driving a single driver/loudspeaker with six amplifiers. Can anyone shed some light on this topic? Any conventional bridged configuration will result in the amplifiers fighting each other. I need to somehow decouple the signals so that each amplifier can only apply work/effort to the driver and not each other.

Just sum them through resistors (~=>10K) into a single output amplifier. With the resistors you could get by without an output buffer but it probably helps overcome capacitive loading in cables and possibly the power amp.