Variable sweep width for the Ropez/Ross Phaser

[Mark Hammer]

The Ross Phaser bears a number of striking similarities to the first issue E-H Small Stone with 6 CA3094's.  In the old Small Stone, two CA3094's (the functional equivalent of a 13600) are also used to generate the LFO.

You will note, in particular, the 270k resistors going to V+ from pins 3 and 1 on IC4a in the Ropez.

The Small Stone has the traditional 2-position "Color" switch.  The produces a change in sweep width as well as regeneration.  The Ross/Ropez uses a variable regeneration control so we'll ignore that half.  The half of the Color switch that changes the sweep width, however, does something interesting.  The Small Stone also has a 270k resistor identical in function to the one coming from pin 3 on the Ropez.  It serves as part of a voltage divider (with the 100k resistor) to provide a bias voltage to pin 3.

Where the Ross/Ropez has a 270k resistor going to V+ from pin 1, however, the Small Stone has an interesting trick.  It has a 180k (we'll call it Ra) and 100k (we'll call it Rb) fixed resistor in series, going from the equivalent of pin 1 on the Ross to the V+ end of the 270k resistor on the voltage divider.  So, 270k tied to 280k.

Here is where it gets interesting.  On the Small Stone, the 270k resistor itself is not tied directly to V+.  Instead, the Color switch  connects V+ to either the junction Ra and Rb, OR to the junction of Rb and the 270k.  So, what we have essentially, is a change that, if implemented on the Ross, would amount to having either a pair of 270k resistors to V+, OR having a 180k resistor to V+ from pin 1, and a 370k resistance to V+ from pin 3.

What this suggests is that the sweep width of the unit could be made variable by using a 100k or maybe even 250k pot instead of the Color switch.  Imagine the following (using a 100k pot).  The fixed resistor from pin 3 of IC4a remains 270k, but the resistor to V+ from pin 1 is now 180k.  The normally V+ end of that resistor goes to one outside lug of the 100k pot.  The wiper goes to V+, and the V+ end of the other 270k resistor (from pin 3) does not go directly to V+ but instead goes to the other outside lug of the pot.

The principle is essentially that sweep width changes as the bias voltage to pin 3 is lowered while the Iabc to pin 1 is increased (by lowering the resistance).

Make sense?  I haven't tried it out yet but it ought to work.

[jrc4558]

Good to know! I used something vaguely resembling this when experimenting with a EZ Vibe, but I tought that i'm simply shifting the X axis with respect to the LFO wave...

[Mark Hammer]

The "hypertriangularity" of the LFO waveform on both the Ross/Ropez and the Small Stone would appear to be a function of the Iabc and bias voltage at that OTA stage.  My sense is that actually several things are being altered at once, including not only the amplitude of the waveform, but where it starts and what it looks/sweeps like.  The Color switch on the Small Stone also seems to have the propensity to change the speed range a little, but then again I have only actually used a 5 x CA3094 issue and not  the issue with 6 x CA3094.  Perhaps the sweep-width function works differently on that one (and identically on the Ross).

Here is a doctored drawing (thanks to Francisco for the original to doctor!).  You can take a look at the original at Tonepad to see what was changed.

What the early-issue Small Stone did was to use a fixed 100k resistor and half of a DPDT switch.  The common of the switch went to V+ and the switch selected between either the 270k/100k node or the 100k/180k node to produce two different widths.

I should point out that these were simply two preselected sweep widths and not necessarily the entire range of what the SS LFO is capable of.  What I've drawn/suggested is simply what the SS schematics says is possible, and does not go beyond that.  I leave it up to you adventurous types to tinker with the different values shown and see how far you can take it.  As I mentioned at the start of the thread, it may well be that a 250k pot and different fixed resistors (perhaps some in parallel with the pot to adjust value/taper) are called for.

At the very least, being able to have narrower sweep and low resonance for the fastest speeds (ersatz Leslie) and high-resonance/wide-sweep for slow speeds will make it a more usable unit....not that it isn't a great little phaser already.
Until now, I had been using an 8k2 fixed resistor and 100k pot in series, in place of the 10k fixed resistor way over on the right (under "LFO") in order to adjust sweep width, but the sweep always seemed a little unsatisfactory.  I'm hoping this will be an improvement.

Incidentally, for those who want to optimize it for Univibing, where the range of usable speeds is much smaller, I suggest the following:
- replace the 3.3uf cap with 2.2uf
- replace the 4.7k resistor on pin 9 with 6.8k
- replace the 500k pot with 250k or even 100k, wired up the same way

The 3.3uf cap sets the speed range.  The 4.7k resistor sets the upper limit on that range.  The pot value reduces the speed below that upper limit.  So what we've done is shifted the whole range upwards (with the 2.2uf cap), indicated that we want the upper speed limit to be a bit below what that wholesale range-shift might otherwise permit (using 6k8 instead of 4k7), and carved out only the upper 2/3 or so of the remaining range for variation with our reduced pot value.  This will give a little more speed at fastest speeds, but skip the very slow sweeps that a phase-shifter makes use of in favour of having more dial-able throb from the Univibe.

[LP Hovercraft]

That's a really nice trick!  I just built a 12 stage Ross (scaled back from 24 stages because it wasn't sounding all that great)  and I noticed that the regeneration knob will cause audible oscillations at the bottom of the sweep when set too high.  Sounds like your mod could take care of that problem I've been having.  Thanks Mark!

[jimbob]

I just built a 12 stage Ross (scaled back from 24 stages because it wasn't sounding all that great)

How does that sound? Compared to the stock version? nIm thinking about doing the same.

[LP Hovercraft]

As you add more stages to a phaser, of course there are more notches sweeping across the audio coming into the unit.  I'd describe it as a bit closer to a Flanger sound.  The sweep takes on a subjectively higher pitch and sounds a bit more "sharp" and "lasery" as opposed to the stock "doppleresque" tone.   I'll try to get a sound sample up.  I'm really happy with the sound of the thing at 6, 8, and 10 stages.

[Mark Hammer]

Okay, I did some experimenting and it's actually easier than I had originally depicted.  You can leave the 270k resistor tied to pin 3 exactly where it is.  All that is really required is to substitute a 500k pot and 180k resistor series resistor for the 270k resistor on pin 1.  Smaller combined resistance values there result in wider sweep.  So, you go from a bit wider sweep than stock to much narrower.  Faster, bubbly speeds can sound nice and shimmering.  The Small Stone version still works, but it does not produce as broad a change in sweep width.  Thismod is not only variable, but improves upon the performance of the SS.

There is still a bit of interaction between speed and width the way there is on the Small Stone (speed changes a bit as you alter width so you'll need to re-adjust speed), but it works like a charm.  It is still possible that some tailoring of the 270k resistance off pin 3 would compensate for the speed change in some manner, but I'll just enjoy what I have now and leave the further experimenting for some other time.  Incidentally, I imagine this would do a nice job when converting a Ropez to a Univibe.  Univibes like a gentler, shallower sweep.

Like Jimbob, I had some "woof" issues with high resonance sweeps.  The solution lay in reducing the 1uf caps ahead and after the 500k regen pot.  Dropping them both down to .22 or so will retain the sharpness at the peak of the sweep but eradicate the woofiness at the bottom of the sweep.  Your speakers are now officially safe!

[Mark Hammer]

Here's the simplified, revised version so you can see.

So, all that is required is to snag yourself a 500k linear pot, a 180k (or 220k) fixed resistor.  Solder the resistor to one lug of the pot, and run the two leads from the pot to the pads where the 270k pin-1 resistor used to go.  That's it.  At some point, I'll need to take a look at the LFO output on a scope and see how the waveform is changed.  My understanding is that the Iabc is important for achieving the hypertriangular sweep, so I gather it is probably altered with the variation in resistance.

Almost forgot to mention one more thing "borrowed" from the Small Stone.  Run an electrolytic cap from the point indicated by "LFO" to ground.  Values of 10-22uf ought to be fine.  Plus side goes to the 10k resistor.  These wil take some of the spikiness away from fast sweeps.