"Hyper triangular wave"

[rx5]

any body here have a working hypertriangular wave generator for chorus-flanger?? preferably using descrete components?  minimum output swing  is 2.4V and maximum ouput swing is 5.8V.....

thanks for any info....

btw, how do you post pics here?? the pic is in my(this) computer..[/url]

[Mark Hammer]

Go to my site (http://hammer.ampage.org) and you'll find one if you scroll down a bit.  Alternatively, go to http://hammer.ampage.org/files and you'll see the GIF.

[rx5]

you mean hypertriangleclock.gif by: James F. Mc Conkey??? got a copy of it too.... have you tried making it?? the ramp gen doesnt work.... polarity of the 2nd op-amp is reversed.... but the jfet circuit works (like) a half wave rectifier...and other op-amp as a DC level shifter...... tried simulating it in EWBv5...

any info on this? change of some parts??

does it sound like or function like the CEM3340 ??

[R.G.]

The point of "hypertriangularity" is that you generate a sweep that is exponential in character. There have been a couple of versions of how to do this, including a full wave rectified signal.

There is an easier way. If you construct anything that is controlled by the current through a silicon junction (like, say, a transistor base-emitter) and drive that junction with a *voltage* signal, the current into the junction is an exponential function of the applied voltage.

For instance: I have been playing with a number of LM13700 controlled circuits. These things are controlled by the current into the Iabc pin, which is itself equal to two silicon junctions to the negative supply. If you drive this pin with a low-impedance triagular voltage waveform, you get the appropriate exponential control current into the device. The size of the sweep waveform needs to be less than about 1.4V peak to peak, and you may want it significantly less, but with a DC offset added to it to give a bias point plus a smaller sweep range.

My circuits have been pointed at generating a triangular sweep around 4.5V and then resistively dividing that down by 9V/1.4V to get the right size and using an opamp to drive the current input point.

This works with anything you can drive with a current, because you can arrange the current to be supplied by a current mirror if it's not current driven anyway. The Iabc pin on OTA's is the input of a current mirror, for instance.

In working with OTAs, you have to be certain that the input current doesn't get above 1ma for the 3080 and 2ma for the 13700 or the device will die instantly. It helps if you make the voltage driving circuit be current limited to less than the death current.

This is one instance where you cannot make the simplifying assumption that a diode is "off" until you pass a voltage threshold. The current that flows in a diode actually starts increasing from 0 Volts across it. However, that current is so small that it takes several "decades" of input voltage to get it to where we normally think of things as conducting. At the low end of the scale, say at a few millivolts, the junctions conducts maybe as well as a chunk of wood - negligible but still measurable.

LEDs work this way too, by the way.

Interestingly, if you think of it this way, the classic two-opamp, two-transistor exponential converter also uses the voltage-driven technique. Consider - a simple two-transistor current mirror where you drive the input side base-emitter with a 0-0.7V triangle voltage results in an exponential current reflected out the output collector. If you use a resistor to some + supply voltage, you get an exponential triangle across the resistor. If you put that same current into the inverting input of an opamp with a resistor feedback, you get the same exponential triangle at the output of the opamp, but buffered. From there it's a really short step to the classical expo converter.

Of course, it's also simple to use an audio taper digitally variable pot chip with a DC voltage across it and drive it with a linearly varying bunch of settings from a $1.25 PIC microcontroller (Think GEO's ASMOP pot-turner). This also generates the exponential directly and cheaply, but as a voltage in this case. Expo weighted resistors in a johnson/ring counter probably does the same thing(see GEO's Psuedorandom LFO and extentions for the sine wave output version), but I haven't looked into how to weight the resistors.

[Mark Hammer]

The 13600-based LFO in the 13600-based Ross phaser also produces a hypertriangular waveform.

[rx5]

thanks for the explaination R.G.    ill try to understand it....

@ Mark Hammer
  since u posted the GIF image of it, have you tried doing it?


what would be the ideal "hypertraingular" look like? the "pointed" going up ?? or "pointed" going down??  the hypertriangular.GIF shows the "points" going down while Craig Anderton's Hyperflanger PDF, the "points" goin up... im confused.... sorry for the "pointed" thingyy ..dont know how to call it....

still looking for a much simpler circuit.... :cry:

[rx5]

im back...

cant seem to find in my bin a transistor current mirror schema.... anyone got one?

i think Lm13600 isnt available here.... is lm3080 the same function with lm13600? can you please include pin to pin substitution from lm13600 to lm3080....?

thanks

[Paul Perry (Frostwave)]

LM13700 will be OK to replace LM13600.
Or if you are in Europe, NE5517.

[brett]

The small bear (Steve Daniels) sells LM13600s for a very reasonable price (cheap enough to be worth paying the freight to Australia if you're buying more than 1!)

[Mike I.]

As an alternative to the typical exponential V/oct response a 1/x response is refreshingly different sounding... but still familiar. An OTA-based phaser or a flanger with the typical expo current source has a very even sounding sweep - much like a synth VCF - moving at a steady number of octaves per second in response to a triangle LFO waveform. In contrast, something like the EH Electric Mistress or the A/DA Flanger has an approximately 1/x response, with the number of octaves traversed per second steadily increasing as the delay shortens - this gives an "accelerating" sound  not unlike the what is heard when listening to two slightly detuned sawtooth waves beating together - the tonality changes very rapidly when the two waves become  nearly in-phase (equivalent to a very short time delay that is approaching zero). Same thing with tape flanging. Perhaps a hypertriangle waveform could be used to give some "acceleration" to a V/oct phaser rather than to just make a linear phaser simulate a V/oct phaser.
Regards, Mike

[Mark Hammer]

Hi Mike and rx5,

Actually, Thomas Henry's "Smooth Phaser" project that appeared in EM used a hypertriangular LFO exactly as you describe, Mike.  While the LFO section is easy to build (and adheres very closely to the Boileau/McConkey circuit in POLYPHONY), the phaser section uses the whooping-crane-like SSM2040 so it's a bit of a non-starter unless your daddy is Bill Gates or Dave Tarnowski.

rx...I have a Tone-pad Ross phaser board etched, populated, and wired up but not functioning properly.  Seems to work when I put my thumb on the copper side of the LFO corner of the board but not when I lift it off.  So, unfortunately, I have not heard a properly functioning hypertriangular LFO.  I am assured by Osamu Hoshuyama that the LFO in the Ross is of the type described.

The  13600 *can* be functionally substituted for with either 3094's or 3080/MPSA13 combinations but I can't speak to whether component values remainthe same or not.

[rx5]

i saw the diagram for the LM13600... consisting of an op-amp and transistor.... can it be emulated(LM13600) by common op-amp and a transistor??

or how about any links for a "hypertriangular" gen..?? from a single 9V supply... im trying to 'upgrade' john hollis ultra flanger...gonna put an anti-aliasing filter at input of BBD, add pre-emphasis on the input...an output filter(7KHz) after the BBD output...de-emphasis at the output....

is adding compressor/compander to the input/output respectively to the BBD enhance the sound of the flanger??

the first 'upgrade' would be just easy because parts are available.. dunnno bout the  compressor/compander... havent heard of that component yet.... lastly , im having hard time about this clock cause i cant seem to find good enough info bout this "hypertriangulating" a sawtooth ramp......

[Paul Perry (Frostwave)]

rx5, if you look closeley you will see the LM13600 isn't an ordinary op amp & transistor. See the National Semiconductor datasheet for details (it is actually an 'operational transistor amplifier' and a buffer. Twice.

[rx5]

yes , i know...just wondering if it could be done through descrete components...

guys, pls post links to where i could possibly have more info about "hpertriangularity"....

or send it to my e-mail : beamrx5@yahoo.com

i really badly need a diagram bout it.... cant sleep thinkin bout it....hope someone helps...... :cry:

thanks...

[rx5]

to R.G.

tried the idea of shunting a diode... i tried simulating it on EWB.... i get square waves at first.... i then seriesd it with 12K..... and in between the 12K and diode , is 15K as the output to inverter/gain/buffer op-amp... nice output.... but top waveform is not steep enough.... i recently tried a transistor as replacement fr the doide... much better ,"smoother"waveform bottom than the diode....
hope you could send me more info about exponential circuits....

thanks