Voltage to resistance converter

[David Hess]

a pair of matched JFETs

Very difficult to find. Realistically you have to use lots of source resistance to degenerate the pair to the point where they are much closer to a fixed resistor than a voltage-controlled resistor.

There are two options here.  Dual JFETs made by Linear Integrated Systems are not that difficult to procure:

http://www.trendsetter.com/SearchResults.asp?Search=23000+23103+23503&Search.x=1&Search.y=1&sort=11

Alternatively grading your own and thermally mating them will yield pretty good results but of course they will not work as well as the monolithic devices at the above link.  A Vos of 50 millivolts and 100 microvolts per degree C is feasible in this way which is about 2 to 10 times worse than a monolithic pair.

[atferrari]

As another option an OTA with 100% feedback also can be used as a controlled resistance.
However, in both cases the controlled resistance is grounded.

Hola Lutz,

There is an application of the 13700 IIRC, that mimics a resistor not referred to ground. Not sure if it could be of use here.

[GonzoTheGreat]

If you can decipher the text marked in red, this might be a solution to build Voltage-Controlled-Resistor (VCR) for high signal levels with two enhancement mode MOSFETs ( ...or JFETs if a negative control voltage is acceptable to you ).


"ENGINEER'S NEWSLETTER

When a single field-effect transistor is used as a voltage-controlled resistor or a current-controlled resistor, its resistance characteristic unfortunately becomes highly nonlinear at high signal levels, notes D. S. Gibbs, an application engineer at Ferranti Ltd. Oldham, England.  But, by using two FETs - either p-type or n-type devices, one inverted and the other noninverted - you can cancel out the nonlinearity of each FET and obtain an extremely linear resistance characteristic.

It's done with a potentiometer.  The input control signal is applied to the wiper of the potentiometer whose resistance element is connected between the source of one FET and the drain of the other.  The gate of the first is then grounded, as is the source of the second.  Then the drain of the first and the gate of the second are tied together, with the output taken between these two terminals.  The potentiometer can now be used to balance the two FETs so that signal distortion can be minimized."


If anyone has any other schematics related to a Voltage-Controlled-Resistor for high signal levels, please post them in this thread.

   

[GonzoTheGreat]

Does Gibb's article describe complementary P-ch and N-ch FETs or non-complementary e.g. N-ch & N-ch FETs ?

[Rerouter]

if you where not opposed to the microcontroller approach, you can use a R2R dac and an op amp to make a constant resistance load,

You divide the load input voltage down to cover you maximum voltage range, feed this voltage into a buffered Vref pin, and use the output of the dac as your op amp set point, as the load voltage reduces so does the op amp set point, maintaining a constant resistance.

leaving the micro to read in your set point and update the dac to the new resistance, with no fancy math in the middle.

[BrianHG]

If you want to modulate sound, why not use a VCA, (voltage controlled amplifier) or much more precise and better analog 4 quadrant multiplying mixing amplifier.  1 input has your audio signal.  The second input comes from a DAC.  With input #1 with a +/-1v AC audio signal, input 2 at 0v, output = 0v.  Input #2 at 0.5v, output = input +/-1v X input +0.5v = output +/-0.5v.  Or in input 2 = 2v, output will = +/-2vac.

http://www.analog.com/en/products/linear-products/analog-multipliers-dividers/ad633.html#product-overview

You can also invert the output by feeding a negative voltage into input #2.

It's a little pricey, but, you can make fantastically accurate audio modulators, volume limiters/mixers.  I would only use this in really high quality apps like analog studio audio mixers.  I'm sure cheaper equivalents exist from third party manufactures.

[danmcb]

+1 for VCAs.

Back in the day the LM13700 TCA was the standard thing in guitar pedals and so on, but it is pretty noisy and generally sounds  like crap.

Another neat approach worth investigating is a lamp (these days white LED) packaged with an LDR. Not very linear but absolutely a current-controlled resistor. I have used this with some success in compressor designs, with a 3D printed LED/LDR mount.

Otherwise, there are very good VCA chips available which can do volume controls  (v/dB law, very linear and also low noise) and be used in digital filters. They typically have to work at the virtual earth point of an inverting opamp stage. the classic one was the dBx2150, which is now made by THAT corp. Analogue Devices also had some great chips - SSM2020 - with dual VCA and log rectifiers which are excellent for all kinds of audio building blocks. The data sheets and app notes are worth a read even if you don't use that chip.

 

[GonzoTheGreat]

if you where not opposed to the microcontroller approach, you can use a R2R dac and an op amp to make a constant resistance load,

Not only such solution is expensive, unwieldy, slow and limited by DAC quantization  but complex as well, e.g. for 10-bit DAC it uses 1000x more transistors than an analog solution.

The 2 FET analog solution from Gibbs is better in all respects.

[GonzoTheGreat]

If you want to modulate sound, why not use a VCA, (voltage controlled amplifier)

Because this is a solution only when the VCA is used in a volume control like in Fig.1.  It is also kind of funny because VCAs are implemented with VCRs internally 

However if the VCR is not a part of a a variable attenuator (or amplifier), then it is a not a good candidate to be substituted by a VCA.
The CdS + LED/bulb combo is a nice try but it is bulky and not functional for >1MHz signals.  Also it ages pretty quickly.

[danmcb]

how can a Voltage Controlled Amplifier not be part of an amplifier?

and are you sure that a (modern) LDR is not functional at HF? and that it ages? the old ones were cadmium,  but the ones you buy now are not. How do you know they have ageing issues?

[exoticelectron]

Ahh the infamous electrically controlled variable resistor..the object of many a young bachelors fantasy. And yet...basically useless. Operating transistors in their linear range is almost never the best solution, especially with something in SOT23 or TO92. If you absolutely need an electrically controlled load, then prepare to make a very good control loop or you will end up in oscillation land. Otherwise, to see where electrically variable resistors are typically used, look to linear regulators. Thats about it.

[Rerouter]

if you where not opposed to the microcontroller approach, you can use a R2R dac and an op amp to make a constant resistance load,

Not only such solution is expensive, unwieldy, slow and limited by DAC quantization  but complex as well, e.g. for 10-bit DAC it uses 1000x more transistors than an analog solution.

The 2 FET analog solution from Gibbs is better in all respects.

I think you mis-understood me, its an analog control loop, just controlled by a digital set point, it has a loop bandwidth of 100Khz through the DAC, though it would need better compensation to work at such a bandwidth on the op amp side.

[BrianHG]

However if the VCA is not a part of a a variable attenuator (or amplifier), then it is a not a substitute for a VCR.
The CdS + LED/bulb combo is a nice try but it is bulky and not functional for >1MHz signals.  Also it ages pretty quickly.

3 terminal CdS + LED, with the LED driven by an op-amp whose - input is sensing one of the 3 pins on the CdS has perfect linearization and perfect age compensation and even temperature compensation.  IE you put in a specified voltage, and you will always get the exact same resistance out.  I used to use them on my audio pre-amps.

Sadly, since many misunderstood the purpose of the 3 pins on these dual element CdS and how to use them, they vanished and I don't believe they are manufactured anymore.

If the OP want a perfect, repeatable, true voltage to resistance control to a GND, these would do the trick with an additional op-amp to drive the LED input, however, the ones which may be still available will cost more than the 4 quadrant analog multiplier I linked to above, and with that chip, you can do a lot more, like simulate the distortion of a vacuum tube pre-amplifier almost perfectly, or create log, square root, sine phase shifting, frequency multipliers, and many more all at much higher frequency.  All these analog functions are illustrated in the data sheet and the IC is just under 7$.

[GonzoTheGreat]

how can a Voltage Controlled Amplifier not be part of an amplifier?

...as in a treble/bass controls.
Indeed, this was badly phrased.

and are you sure that a (modern) LDR is not functional at HF?

Yes, I found them to be quite inductive.

[GonzoTheGreat]

3 terminal Vactrols, CdS + LED, with the LED driven by an op-amp whose - input is sensing one of the 3 pins on the CdS has perfect linearization and perfect age compensation and even temperature compensation.

Funny that you should mention these Vactrols.
That is exactly what I am trying to replace in a large audio mixer (most of them failed ~4 years ago).

[macona]

Its done all the time in audio control boards, its called a motor driven slide pot!

Just set up a servo loop with voltage control and you have a nice voltage controlled resistor.

[GonzoTheGreat]

Just set up a servo loop with voltage control and you have a nice voltage controlled resistor.

You must be joking ! - I might as well build a new mixer...and a bigger one at that.

@Guys
Why can't we get the Gibbs' circuit working?

[stj]

lol
ask a room full of wanabee scientists and you get a complex answer.

just stick a small lamp bulb and a photo-sensitive resistor like an "orp12" into a small section of black tubing!

[mikerj]

lol
ask a room full of wanabee scientists and you get a complex answer.

just stick a small lamp bulb and a photo-sensitive resistor like an "orp12" into a small section of black tubing!

You could read the thread and see this has already been suggested by your so called "wanabe scientists" and rejected by the OP...

[T3sl4co1l]

@Guys
Why can't we get the Gibbs' circuit working?

If you want a working circuit, that'll cost extra...  Just sayin'.... 

Tim

[Bernhard]

@Guys
Why can't we get the Gibbs' circuit working?

The part in red is indeed puzzling. Do you have the complete article ?

Thanks!