Automatic phaser sweep

[manson]

Hi there,

The EHX Electric Mistress flanger has a switch to disengage the automatic sweeping. Is there a mod to do the same on a Phase 90? I think it would be cool to have the effect on in one position all the time.

Thanks

[R.G.]

Sure - where the LFO goes to the JFET gates, disconnect the LFO and sub in a pot with DC on it. If you make a switch so you can switch between them, you have selectable auto/manual sweep.

[Mark Hammer]

For FET-based phasers like the P90, tyhe FETs are actually being governed/swept by a combination of the LFO and a DC bias voltage.  Normally, the DC bias is intended to simply place the FETs into a range where they undergo the widest and most desirable sweep when pushed by the LFO.  But as RG points out, there is no reason why you couldn't simply rely on the DC bias alone.

Case in point.  Go to the Tonepad schem of the P-90 ( http://www.tonepad.com/getFile.asp?id=42 ).  The squiggly line going to each FET (the gate) is what you are trying to vary.  The DC bias voltage set by the 250k trimpot (coiming via the 1M resistor in red), and the AC voltage coming from the LFO via the 3M9 resistor each arrive at the same point.  That is, the gates of all 4 FETs see a mix of what comes in via the 1M and the 3M9 resistors.  Lift the 3M9 resistor from that point, and the FETs only see what comes in from the bias circuit.  Make that 250k trimpot a 250k chassis mount, stuck in a toggle to break the connection between LFO and gates, and the job is done.

HOWEVER, as RG wisely notes (indirectly or tacitly), the ideal is to have a 3rd source of DC bias.  Why?  Because you'd rather have the "best" setting for using the LFO be something you can depend on, not something you have to find all over again with the 250k pot, every single time.  In which case, the ideal is to have a 3rd DC source which comes in via its own resistor, to the same tie point.

How would I do this, you ask?  Tag, you're it, RG.

[manson]

So it can be done. You guys make it all seem so easy Now let's see what R.G. has to say..

[Mark Hammer]

So it can be done. You guys make it all seem so easy Now let's see what R.G. has to say..

It's not really all that hard but it sure FEELS a lot easier when you've thought about it 500 times already.
Remember, that at a certain point, even complete morons can take a pretty good stab at the chords to a blues song they've never heard before!   

[GFR]

While you're doing the mod, you can add a jack for external control, like an expression pedal (or any other control source you can think of).

[R.G.]

While you're doing the mod, you can add a jack for external control, like an expression pedal (or any other control source you can think of).

Oh, sure... make it really fancy... 

In the P90, there is a "bias" trimmer pot on the board. It's 250K, and the hot lug connects to the bias zener diode voltage, the cold lug connects to ground, and the wiper goes through a 1M resistor to the gates of all the JFETs. The gates of the JFETs also get the LFO through a 3.3M resistor. The point where the 1M and 3.3M meet, let's call that the "LFO drive point". The junction of all four JFET gates let's call "the gates" and assume that there is one wire between the LFO drive point and the gates.

If we want to change to a manual control for the sweep, we need something to make a DC voltage somewhere between the max of the bias voltage and ground. A pot does this nicely, so we hook another pot, this one panel adjustable, between the bias zener voltage and ground, in parallel with the bias pot's outside lugs. The wiper of the new pot has a 1M resistor connected to it. The free end of the 1M resistor we'll call the "manual drive point".

So we hook up an SPDT switch, with the pole connected to the gates of the JFETs, one throw to the LFO drive point, and the other throw to the manual drive point. Ding. We're done. With the switch flipped to the LFO side, it works normally. With the switch flipped to the manual drive point, we have manual control.

Too simple?

Well,  yes, actually it is. While what I've told you works as stated, you have too much control - the sweep on the manual pot is too wide, and most of its range is not useful, as the JFETs are either cut off or saturated, not in the linear resistance region. There's only a fairly small region somewhere in the manual pot range that does the whole JFET sweep.

What we need are range restrictions. These are just two more resistors, one at the top end and one at the bottom end of the manual pot that cause its top voltage to be less than the bias voltage, and the bottom voltage to be more than ground. Cool. What value resistors?

ACK! It depends on your JFETs. Each set will be different, in both the top and bottom voltages. But once you find them out, they stay that way until you replace the JFETs. So what you do is to use THREE pots - say, about 100K each. String these in series, with the top and bottom ones set up as variable resistors of 0-100K ohms, and the middle one as a 100K voltage divider, with its wiper to the 1M for the manual drive point.

Now we tune. Set the phaser to running, set all three 100K pots to max resistance. Turn the top pot down until you reach the edge of good-sounding phasing on the top end. Turn the middle pot to minimum and turn down the bottom pot until you reach the edge of good sounding phasing. Turn the middle pot back to full up and go readjust the top limiter pot, because top and bottom interact a bit. Now readjust the bottom. Now the top... after a few iterations you have the perfect settings. Now measure the top and bottom pot resistance settings and substitute in the nearest standard value fixed resistor. Done.

You could also just use one pot, measure the wiper voltage, and note the voltage at the top and bottom of the range for good sounding phasing, then calculate the top and bottom fixed resistors. I detailed the set and measure pots stuff because no one can remember ohm's law well enough to do the calculations.

As to a plugin external source - yeah, great idea except that both ends of the expression pot are floating, so you either have to have fancy floating setups, or you have to do some opamps at the expression jack to convert the expression pot voltage to the proper scale and level.

[Mark Hammer]

As to a plugin external source - yeah, great idea except that both ends of the expression pot are floating, so you either have to have fancy floating setups, or you have to do some opamps at the expression jack to convert the expression pot voltage to the proper scale and level.

Which is, in fact, exactly what modular synths have for mixing different control-voltage sources (modulation, keyboard tracking, expression pedal, breath controller, etc.).  Of course, at this point, our simple little P90 stompbox starts to turn into something much more ambitious.  Not impossible, but a step up in terms of planning out and integrating control sources.

[manson]

So I guess it's something like this, neat huh:



But where do those two limiter resistors go exactly?

[R.G.]

So I guess it's something like this, neat huh:

Yes. That's a good way to do it.

But where do those two limiter resistors go exactly?

The new pot marked "???K" has one lug connected to the bias zener voltage, and one to ground. Put one resistor (which I'll call 'r1' for reasons which will be clear later) between the pot lug and the bias zener voltage, and a second resistor(which I'll call 'r2'))  between the other pot lug and ground.

What we are doing here is restricting the range of voltages that can come out of the pot wiper. In this setup, the resistors are now connected
Bias zener voltage -> r1 -> top of ???K ->bottom of ???K -> r2 -> ground. This ignores the wiper on ???K.

The voltage at the wiper of the ???K pot is going to be determined by the zener voltage divided through all of the resistance above the wiper and all of the resistance below the wiper. So we can get ANY range of voltage on the pot wiper from 0-Zener or any restricted range between by diddling with the resistor values.

Let's do an example. Say the JFETs have a "good resistance range" of Vgs of -0.5 to -1.5V. Further assume that the bias zener is 4.00000V. So the voltage we want to feed to the gates is 3.5V down to 2.5V. We start with 100K for the ???K pot, and 100K for each of r1 and r2. In what follows, we don't know any of the numbers, only what sounds good. I've included the numbers to show what is happening.

Following the easteregging instructions, we turn the wiper up to the top of the pot range. We don't know this, but that puts the wiper at 4.0*(200K/300K) = 2.667V. This is in the good range, but low.

We adjust r1 by ear to get to the top of the good range, or 3.5V. Again, we don't know the voltage, only that it is the biggest voltage that sounds good. So r1 is actually r1= (200K)*(4.0V-3.5V)/3.5V = 28.57K.

Now we turn the pot wiper down to the bottom of the range. The voltage on the wiper is 1.75V. That's low, so we increase r2 (by ear, so it still sounds good). The value that comes up is 214.285K.

But that changed the resistance at the top. We go tweak by ear again.

After tweaking r1 and r2 back and forth a few times, we home in on r1 = 49K and r2 = 249K. Of course, we don't know the numbers, but we know that the 100K manual pot has the whole good sounding range in its sweep.
(I checked - the sequence converges to these values in eight iterations in excell.)

We could also have done it the voltmeter way. Picking 100K for the pot, and leaving r1=0 and r2=0 we measure the voltages where the JFETs sound good as 2.5V and 3.5V. Then we calculate the resistors r1 and r2 to give us 2.5V at the bottom of the manual pot and 3.5V at the top.

This is done by knowing that the 100K pot will have 3.5V-2.5V=1.0V across it, and is 100K, so it must have 10uA going through it when we get r1 and r2 right. r1 and r2 have the same 10uA, so the value of r1 must be (4.0-3.5)/10uA = 50K and r2 = 2.5V/10uA = 250K. Amazing isn't it - that's right where the interative adjustment was converging towards.

Your JFETs will of course not have the same Vgs as I picked out of the air for this example, but it's going to be similar.