4 variable resistances from one potentiometer?

[uvacom]

Hey, I'm working on a design that will require me to vary 4 separate resistors. I figured the best way to do this will be to use some kind of FET in saturation, with a small Vds, varying Vgs with a pot as a voltage divider. I've gotten that far, but I've never built something like this, so I'd like to see a working circuit example that I can adapt for my needs. Also, I have no idea what kind of FET is best for this. Right now the resistances are all equal, and should vary between 50k and 3Meg, but I'm not committed to those values, they were chosen for minimal current draw.


Sooo, ideas?

[R.G.]

Hey, I'm working on a design that will require me to vary 4 separate resistors. I figured the best way to do this will be to use some kind of FET in saturation, with a small Vds, varying Vgs with a pot as a voltage divider. I've gotten that far, but I've never built something like this, so I'd like to see a working circuit example that I can adapt for my needs. Also, I have no idea what kind of FET is best for this. Right now the resistances are all equal, and should vary between 50k and 3Meg, but I'm not committed to those values, they were chosen for minimal current draw.

Do what the IC guys do - use a high frequency PWM clock and switch the duty cycle of fixed resistors or capacitors to make duty-cycle variable resistors. You can do that with either resistors or caps. Probably inductors too, but Rs and Cs are so much easier.

[Phostenix]

Isn't the Boss PH-1 an example of 4 FETs as variable resistors?

[R.G.]

Isn't the Boss PH-1 an example of 4 FETs as variable resistors?

It is, as are all JFET based phasers. But -

JFETs are highly, highly variable even within the same part number, even within the same manufacturer's selection group. In fact, manufacturers' selection groups were invented to help take a little of the variability out of thing.

They work OK if the signal you're variable-resisting is significantly (about ten to one) smaller than the Vgsoff of the JFET and the currents never get very large; and if you can stand a resistance that's variable but not too predictably variable between rdson and infinity; and if what you want is several resistors to vary in the same direction at the same time and aproximately the same amount.

From the poster's suggestion of different resistors I inferred that he maybe had different amounts and different directions in mind. Don't know exactly what he's wanting, though.

[Phostenix]

JFETs are highly, highly variable even within the same part number, even within the same manufacturer's selection group.

And all of this reminds me that I need to look up your JFET matching jig & go through my 2SK30as so that I can fix that PH-1 in the "to be fixed" box....

[Steve Conner]

This is one of the classic tricky things in electronics, IMO.

Mesa use an arrangement of dual tracking LDRs in their Triaxis preamp. This has been reverse engineered on this forum.

There's the PWM approach mentioned by RG, though the problem with this is that you're now sampling the signal and can generate alias frequencies. So to keep those out of the audible range, the PWM frequency has to be very high, and then it's difficult to get a wide variation in duty cycle.

Probably the easiest way is to redesign the circuit so that, instead of variable resistances, it's controlled by variable-gain amplifiers. You can then use OTAs (aka "Norton amplifiers") and control as many of them as you want from a single pot, because their gain is varied by a DC control current. The VCAs and filters in many classic analog synths worked this way.

Look at the LM13700 datasheet for ideas. In particular, look at the state-variable filter example, and compare it to the circuit of a state-variable filter using ordinary op-amps.

[uvacom]

This is a good start. R.G., that information is very helpful. In fact, I do want several resistors to vary in the same direction at the same time by the same amount. They don't necessarily need to be the same exact value (same order of magnitude is probably workable), I'm doing a one-off so I can hand-match the JFETs anyway.

Basically I'm trying to build a tow-thomas biquad lowpass filter, a pretty standard textbook active filter design that has 2 resistors R that control the cutoff frequency, a resistor QR that controls the Q by the function (take a guess) Q*R, and a resistor "R2" which controls gain by the function R/(dc gain). Basically, there are 4 resistors that depend on the value "R". To make things simple, I've selected a Q of 1 and a DC gain of 1. But it's not critical, it's actually more important that they vary in the same direction by the same amount at the same time. When I said 50k to 3Meg, that's the range of voltage-controlled resistance in which I'd like the FETs to operate (more accurately, I'd like them to operate across about 6 octaves of frequency range - the absolute values are less critical).

To be honest, it doesn't have to work extremely well, but I have to put some effort into trying, even if the result is impractical (actually, that's a desirable outcome). I'm doing an independent study course for my EE degree on active filter design, and this is the preliminary work that should highlight the need for the OTA-based and ladder filter networks that typify virtually every audio VCF topology around. The whole point is that the designs already given to me in previous coursework aren't adequate and that I need to do this research to understand how to build a transistor ladder filter that I can vary with control voltages driven via pots on my telecaster. It's going to be a really cool project, but they'll only let me do it if the material doesn't overlap with existing structured courses which are offered at the university (which it doesn't).

The desired outcome is to develop some cool designs that should be quite applicable to the modular guitar effects concept I've seen discussed elsewhere on the forum, primarily an outboard VCF/VCA unit and a method of replacing standard tone/volume controls on a guitar with control voltage outputs.

[uvacom]

They work OK if the signal you're variable-resisting is significantly (about ten to one) smaller than the Vgsoff of the JFET and the currents never get very large; and if you can stand a resistance that's variable but not too predictably variable between rdson and infinity; and if what you want is several resistors to vary in the same direction at the same time and aproximately the same amount.

That doesn't sound too bad. As a start, can I just choose a JFET with a large Vgsoff, a smallish Rdson, and use a series resistor to set the minimum resistor?

[Steve Conner]

Wait a minute, you're saying that you already know about OTAs, ladder filters and so on, but for some reason you want to reinvent the voltage-controlled filter?

Again, the LM13700 datasheet shows how to make a biquad-type filter with voltage-controlled cutoff frequency. I've made it myself and it works fine.

If you follow the JFET idea, I guess you'll have a JFET ladder filter. Nothing wrong with that.

[uvacom]

Wait a minute, you're saying that you already know about OTAs, ladder filters and so on, but for some reason you want to reinvent the voltage-controlled filter?

Again, the LM13700 datasheet shows how to make a biquad-type filter with voltage-controlled cutoff frequency. I've made it myself and it works fine.

Yeah, I know. It seems silly. But it's an educational thing. It's not practical, but it demonstrates the need for more exotic components like OTAs in the first place. Besides, if I can shoehorn these variable resistors into working reasonably well in this application, I should have a very solid foundation for implementing voltage control to a great many parameters in any analog audio signal processing application (e.g. guitar effects & synthesizers).

If you follow the JFET idea, I guess you'll have a JFET ladder filter. Nothing wrong with that.

That's an interesting idea, but I don't *think* that's what I'm doing (unless I'm missing something). A tow-thomas biquad looks like this:



The only difference is that I'm injecting the input signal at the input of the second op-amp instead of the first, which makes the output of the first opamp a lowpass function.

[Steve Conner]

Oh, well I hadn't heard of the Tow-Thomas, but the schematic you posted is more or less what I know as the state-variable filter. And my earlier comment stands, the easiest way to voltage control it is by replacing the regular op-amps with OTAs. You can get voltage controlled Q too, by adding a third OTA. Roland used this topology in some of their early synths.

This will work better than JFET "resistors". For a start, how do you make one that doesn't have one end grounded? (I've used the H11F1 JFET optoisolator for this.)

[uvacom]

Okay, I've attached a pdf of the first rev of the filter design schematic. It's not totally cleaned up, sorry about that. The jfets used are j113s, selected for a balance of Rdson and Vgsoff. Quite possibly there is another device which will work better. I haven't built it yet, but it seems to do the trick in PSPICE, with a cutoff frequency of about 500Hz to 12KHz or so, a solid -12dB/octave slope, and fairly linear operation with input signals of up to 200mV p-p or so, probably usable up to about 500mV p-p. Obviously matching jfets and their corresponding linearizing resistors will be fairly important.

Incidentally, what kind of peak voltages could I expect from a telecaster? I can barely detect any voltage on my digital meter and tiny jumps of the needle on my old analog meter, but I don't have a scope at home to get a better idea.

Steve: You don't have to convince me that this is not really a good way to do things. In fact, building and analyzing an OTA SVF will be one of the main objectives of the course. I wish you could meet the professor who administers the undergraduate independent study program. He won't approve the project if it seems like something we should have learned in a previous course. These kinds of biquad filters are the most advanced we study in any course. I have to show, not just tell him that a good engineer's best effort with this sort of topology would not be adequate for the application. And like I said, this is a learning experience for me.