Sitar (mostly @Mark)

[R.G.]

I got some new ragas from Ravi Shankar's back catalog, and as I listened there was a long sequence of slow single notes that picked up the individual drone strings as they were played over.

And it hit me how this works from the engineering standpoint. The drone strings are very high Q tuned circuits, with mechanical inputs, and the certainty that they can never get into oscillation because there is no active power input. And they are hard-limited by hitting the bridge mechanically at some amplitude. The drone-filters are driven by a very limited amount of energy being fed in, and so they take some time to build up to ringing, giving the delay of the drones in building up to limiting.

I speculate that one could buffer the straight signal, then run it into a very high-Q filter set below the edge of self oscillation, and limit this with some diodes. Probably one diode, as the sitar bridge only limits on one side. This could even be gated by an envelope watcher. You would have to build one high-Q filter per drone frequency, but this probably becomes one state variable filter per.

It's limited, as only a few drones are possible, but this might be a decent sitar-like adjunct to guitar, and not of the wandering-notches variety.

[artifus]

throw in some envelope pwm too?

[earthtonesaudio]

How about a chromatic scale's worth of single transistor twin-T filters?  With high gain transistors and matching you'd be right on the hairy edge of oscillation anyway, and you could intentionally mis-bias them to get your hard limiting.

[Mark Hammer]

So how do octave doublers manage to achieve something like this when you pick by the bridge?  (The jawari does too, and is a sort of wimpy Tychobrahe Octavia) I'm not disputing what you suggest (mostly because I don't know enough to be able to!).  I'm just curious about building a conceptual bridge between the two phenomena.

[artifus]

How about a chromatic scale's worth of single transistor twin-T filters?  With high gain transistors and matching you'd be right on the hairy edge of oscillation anyway, and you could intentionally mis-bias them to get your hard limiting.

if we're synthesizing would not cmos be quicker and cheaper? 4017/4046? envelope sweep thru pulse width and bobs yer aunites other half?

[digi2t]

So how do octave doublers manage to achieve something like this when you pick by the bridge?  (The jawari does too, and is a sort of wimpy Tychobrahe Octavia) I'm not disputing what you suggest (mostly because I don't know enough to be able to!).  I'm just curious about building a conceptual bridge between the two phenomena.

Thanks for kicking in that one Mark. My recent Superfuzz adventure proved worthwhile insofar as that phenomenon was concerned. I even demo`d it in my video. Very sitar like, minus the drones.

if we're synthesizing would not cmos be quicker and cheaper? 4017/4046? envelope sweep thru pulse width and bobs yer aunites other half?

Could the PLL circuit figure into this somehow?

[artifus]

i haven't really studied the pll but i'm guessing it has a divider in the 4046 loop? so yeah. i'm going on the electrax and clark synth schematics.

[WaveshapeIllusions]

So something like a drone synthesizer? I've thought about something like that before, but for some reason I didn't do it right. I was thinking something along the lines of twin-T filters sharpened with feedback. But then that would give a notch, when I just wanted to boost that part. Perhaps parallel paths, one straight and one with a series of filters?

I might be looking at this wrong, but we want the drones to react to a certain pitch? Say set one to A440 and make it so it oscillates when a note of similar pitch (another A, or a C) is played? Or do they go off anytime a note is played? The second sounds simpler.

[artifus]

i was thinking a 4040 could produce a lower octave or two audio output and that its later outputs could be control voltages to sweep 40106 pwm filters (led/ldr?) from fat to thin or vice versa the octaved (up & down) and harmonized outputs.

or something. night night x

[wavley]

I have been listening to almost exclusively Ravi Shankar for the last two weeks leading up to an art installation in which I intend to play a few (space rock interpreted) ragas.

And then the universe (diystompboxes) tells me that R.G. was possibly listening to Ravi at the same time.

groovy.

[Mark Hammer]

Thinking about it a little more now, note that a great many octave doublers will use a phase splitter and diodes to rectify the signal by lopping off one half cycle and then combining the result to produce two blips where there used to be one.

But here's the thing: those diodes place a constraint on what portions of the spectrum can pass in order to be combined.  If the blip falls below the forward voltage of the diode, there's nothing to combine.  So, the harmonic content that might be doubled is, in some respects, more "aggressively" eliminated from the spectrum at the output by those diodes.  Which sort of suggests that changing the forward voltage of the diodes (via diode type) would result in qualitatively different sitar-like sounds.  It also suggests that signal level, and deliberate tone shaping prior to rectification, would make a noticeable difference.

Hmmmm...über-jawari, anyone?

[earthtonesaudio]

How about a chromatic scale's worth of single transistor twin-T filters?  With high gain transistors and matching you'd be right on the hairy edge of oscillation anyway, and you could intentionally mis-bias them to get your hard limiting.

if we're synthesizing would not cmos be quicker and cheaper? 4017/4046? envelope sweep thru pulse width and bobs yer aunites other half?

Maybe, but if you want simultaneous droning of different frequencies you either have to do complex synthesis or have parallel oscillators.  I was thinking parallel oscillators, like...



The "buffer" above could also be a mixer, and accept feedback from the output mixer (regen/dwell).
Note: obviously the R and C values above are placeholders, not real values.

[artifus]

i was thinking post oscillator - like tim's pwm but x 3 or however many lines are required. sweeping thru the pwm filter effect x 3 ovtaves/harmonies to replicate the acoustic sweep of a sitar. just thinking out loud again...

[R.G.]

I've messed with oscillator drones and such before.

The thing that hit me was the energy-boundary argument. Lots of engineering problems are approached by bounding how much energy it takes to do X, or how much power, or momentum, or whatever. The drones start a noticeable fraction of a second after the sounding string that excites them starts. They are poor in harmonics for the first of that time, then the movement builds up enough to get them to hit the bridge, and they get all harmonically enriched, and probably shift in frequency a bit as hitting the bridge modulates their effective length. They also carry on and fade out after the note stops.

All that comes from a model of modest energy fed into a high-Q filter, building it up at the resonance, then running down. Continuously running oscillators could do this, but you'd have to mess with the envelopes and some clipping to do it. It gets complicated.

I was also after modularity; some thing like two ICs = one drone, and all are alike except for the value of the tuning caps/pot.

This is, of course, all speculation right now.

[artifus]

^ in my pathetic and amateurish messing with cmos i found you can only add so much gain and filter off so much hf before it doesn't matter or hinders your purpose further. is that what you're saying?

*crap* that doesn't make sense - hang on a sec and i'll edit...

...nope sorry - too tired. maybe tomorrow. night night x

[pinkjimiphoton]

guys, couldn't you split the signal several ways, and assign each split to a different phase locked loop so you could get polyphonic drones happening that would be dependent on whatever frequencies you put into them? or would the tracking/latency suck too much?

closest i've gotten to anything sounding like a sitar is using a very short ddl with the regen cranked so that it's always on the edge of runaway. when ya play a note, you can make the drone come and go, and can "tune" it to whatever resonant frequency you want.

sorry.. this is way above my paygrade, but man, i'd love to see what you guys come up with.

[Jdansti]

>The drones start a noticeable fraction of a second after the sounding string that excites them starts.

Something similar happens with gongs. The proper technique is to "warm" the gong up with a few small taps (adding a small amount of energy) to get it vibrating before whacking it, as opposed to just striking it cold.

[Morocotopo]

Yes! I want a sitar pedal, so hurry up guys!

Seriously, Jimi´s idea doesn´t seem so bad... delay IS part of the mechanism, as stated before... so a delay and a few almost resonant filters after tuned to the various notes?

[pinkjimiphoton]

ariel, like, if you had 12 super short delays with a lot of regeneration, so that the "excitement" tone can be "tuned" by the delay time...

you could "excite" the delay into self oscillation any time you played a note at that pitch or some number of it's fundamentals.. i'd imagine you could get a drone that may be controllable, AND poly noted... and as you bend and slur notes between where each delay (hmm... is it a delay at that point, or an excited combfilter/phase shifter?) you could make the drones start or stop.

i can do it with the delay in my boss be5, albeit you have to "tune" it for each key it's not perfect or anything... but wondering if that's where a Phase Locked Loop could come in, if my understanding is right... those will take an input signal, and that input will synthesize a tone at the same pitch with the internal vco, right?

i will try and shoot a stupid pedal trick with the delay in a little bit, to show ya kinda what i'm talking about.

i THIMK (spelling intentional) that if you have the delays set just before they self trigger, all ya gotta do is play to get them to resonate.

i know i'm not describing this right, it's like a caveman that orders mango salsa asking neil armstrong how to get to the moon.

but, like... IS something like this possible to do? with delays that short, do you even need bbd's? couldn't you do it capacitively or something?

sorry for all the questions...

[R.G.]

One important thing (I think... ) is that the drones are frequency selective. The E-drone string (if there is one of those) would only respond to E's and the harmonic resonances of that metal string.

Hmmm. Just hit me. One might be able to build a drone-string box.

Imagine something like a rack of strings or an mbira on a sound board inside a box. A speaker under the sound board provides the necessary acoustic input to drive the sound board and hence the strings/leaves. A magnetic pickup across the strings grabs the string movement, which can then be amplified and mixed back to the straight guitar sound. It would be a tricky build, being a bit less complicated than building a guitar, as well as vibration sensitive (ever whack a reverb pan?) and requiring tuning.  But it could get a more-authentic emulation of sitar drones, being closer to the means of sound production.

It's been bothering me that while it's easy to get mechanical Q's of thousands without ever going into oscillation, it's difficult to do this with electronic means because of the ease of slipping over into a loop gain greater than unity with a powered drive.