phaser LFO

[Ice-9]

I have been reading some posts here about triangle waves to to control the fets in a phaser, (the usual way it is done) I am starting work on a phaser this week which i am going to use 2 * 4007ub using the n channel fets as the vr's. What i was wondering is would it be beneficial to use a sine wave for the LFO on the gates rather than a triangle . Has this been tried by anyone ?

Regarding the LFO circuits i have been looking at, there are a few different ways to do this.

1. My first thought was to use a 555 timer as an astable circuit to generate a square wave then use a capacitor to filter the waveform to more like a triangle.
2. Then i checked some schematics and found most circuits use 2 op amps, which i think is a schmitt trigger and comparitor. which also give a triangle wave.
3. The mxr phase 90 uses 1 op amp, which i think is a simple square wave generator, but the use of the 15uf cap charging and discharging should give a triangle wave again. (I think thats how it works)
4. but then i thought why not use a sine wave ??

Is there any info on designing suitable L.F.O's available. An easy tutorial would be handy, would also benefit from some maths on this as well !
Thanks

[R.G.]

I have been reading some posts here about triangle waves to to control the fets in a phaser, (the usual way it is done) I am starting work on a phaser this week which i am going to use 2 * 4007ub using the n channel fets as the vr's.

Watch the substrate diodes. Be aware that diodes connect the N-type diodes to V-  and the P-type devices to V+ on all CMOS chips, and that the gates have diodes to both of them that are normallyl reverse biased. It may be tricky to get the DC voltages set up correctly.

What i was wondering is would it be beneficial to use a sine wave for the LFO on the gates rather than a triangle . Has this been tried by anyone ?

The human ear likes sine waves. It doesn't matter the MOS devices or phaser at all. It's been done.

1. My first thought was to use a 555 timer as an astable circuit to generate a square wave then use a capacitor to filter the waveform to more like a triangle.

You could buffer the timing cap and get a triangle vrom 1/3 to 2/3 of V+ directly.

2. Then i checked some schematics and found most circuits use 2 op amps, which i think is a schmitt trigger and comparitor. which also give a triangle wave.

Essentially the same as a 555 does except that the positive and negative trip points are adjustable and the integrator gives you a true triangle, not a shark's -fin wave like a filtered rectangle.

3. The mxr phase 90 uses 1 op amp, which i think is a simple square wave generator, but the use of the 15uf cap charging and discharging should give a triangle wave again. (I think thats how it works)

See one of my posts here recently. The cap in the P90 is serving as a crude integrator.

4. but then i thought why not use a sine wave ??

Historically it's been because sine waves are hard to do well and triangles are very easy and cheap.

Is there any info on designing suitable L.F.O's available. An easy tutorial would be handy, would also benefit from some maths on this as well !

Old National Semiconductor app notes and The Art of Electronics should work well.

[Mark Hammer]

The preferred waveform can vary with speed.  For very slow sweeps (couple of seconds per sweep), so-called "hypertriangular" (sine at the low end, triangle at the top) sweeps are preferred since they deccellerate as they go lower and spend more time in the region where you notice the notches.  At faster "chewy" speeds, triangle is good, while at fastest "bubbly" speeds sinusoidal on both ends of the sweep (top and bottom "turnarounds") is probably most pleasing.

One of the tricks that EHX used in the first issue Small Stone was a cap to ground between the LFO output and the OTAs being modulated.  The cap value was selected such that it formed a lowpass filter that added some discernible softening of the peaks as the LFO started to sweep faster than around 0.5hz.  In that way, the hypertriangular waveform that the LFO provided for slow sweeps became closer to sinusoidal at fastest sweeps.  Clever idea.  The Ross/Ropez phaser (see project at Tonepad) uses the identical LFO circuit as the original SS, but omits the cap in question.  If you want to nail the SS speed/shape thing, you add a 33-47uf cap to ground on the OTA side of the 10k resistor coming from the LFO.

[Ice-9]

Thanks for all the info, I am going to check out all the points made and also have a look at the L.F.O generator with mods on the ross phaser mentioned. I have had a quick look on ebay for The Art of Electronic by Paul Horowitz & Winfield Hill (is this the correct book ?) Wow at £70 UK pounds on ebay, about $120 us dollars i want to make sure its the correct book. It will be available cheaper if i have a good look around.

When i get a chance to start this project i will post the diagrams of the circuit blocks i will be working with. Time to get my maths revision books out i think.
Thanks

[Paul Perry (Frostwave)]

If anyone wants to make a hypertriangle from an ordinary triangle, i think you could do it by amplifying the triangle by an OTA (a 3080 or half a LM13700) and doing the following:
1. have the signal amplitude high enough to flatten the triangle to a sine, and also
2. bias the other input to the OTA so that the triangle is squashed on the top but not the bottom.
A bit of fiddling around to get it right, but I'm sure it would work.

[slacker]

I've been trying out your overdriven OTA idea recently Paul, using the same setup as in Ray Wison's "Cool New LFO" that you've mentioned before. I've only simmed it so far but while messing about I did get something like a hypertriangle out of it.
It's a very cool little circuit and fiddling with the values lets you get triangle, sine and trapezoid  waves at different settings of the "sine shape" pot. Using the other half of the LM13700 and an opamp you can make a nice simple multi shape voltage controlled LFO.

Another way to get a rough hypertriangle is to clip the triangle using diodes like in John Hollis's Easyvibe, but only clip the bottom half of the waveform. This looks alright but at slow speeds the clipped side might be too flat.

[Mark Hammer]

Look here: http://hammer.ampage.org/files/hypertriangleclock.gif
...and here: http://www5b.biglobe.ne.jp/~houshu/synth/RossPhLfo0510.gif
Somewhere on his wonderful site, I think, Osamu Hoshuyama has some other LM3600/13700-based LFO circuits. http://www5b.biglobe.ne.jp/~houshu/synth/

[Ice-9]

Look here: http://hammer.ampage.org/files/hypertriangleclock.gif

Very interesting article . Would i be correct in thinking that the phased locked loop and flipflop part of the circuit which makes two square clocks, one inverted to the other are only for the control of BBD chips clock,so wont be needed in the phaser LFO ?

OOPS. just re-read the article and it does say they can be left out ! Sorry sometimes you need to read things a few times.

Thanks Mark