Standardising Rds of FET VCR at a particular Control voltage

[Vivek]

For a compressor, I need the Rds of a FET Voltage controlled resistor to be a specific value at a specific control voltage V3

This will help to use many different types of FET, many FET with same part number but widely differing Vgsoff (-0.7 to -1.5V) , but still get some level of standard compressor response from all of them. It will help to standardise if I make multiple pedals.

I need the Rds locus curve to pass through the arbitrarily chosen standardization point (Control voltage V3  = -1.977V, Rds is 8.285K)  (or very close to that point) which is my standardization point for the different FET.

R106 and R107 are linearisation resistors to cancel the Vds/2 term in the Rds calculations, however there is a trimmer to make small adjustments for different Vgsoff. They help to fine tune the relationship between control voltage V3 and Rds, i suppose.

I have been trying to understand the constraints and interrelationships on Vgsoff, Idss, R106, R107 etc that will guarantee that a solution will always be found within the range of variable resistor VR2


In practice on the breadboard, I found that in some cases like low Vgsoff, I needed to reduce R106 and R107 to be able to get a solution, maybe to 1 Meg + 2 Meg pot + 1 Meg

while with higher Vgsoff, I needed to increase to maybe 10M + 2 M + 10 M resistors


Why should the ability of FET to have Rds 8.285K at control voltage V3 of -1.977V depend on absolute value of linearisation resistors R106 and R107 and not only on their ratio, when all the literature only say that "linearisation resistors should be large, much greater than load resistor and should be equal to each other"


Please help me to understand the constraints of the FET specifications within which an solution will always be found.

I hope we can mathematically show the system constraints and bounds like "Any FET with Vgsoff between X and Y and Idss between U and V will always work to the desired specification by adjusting the trimmer as long as R106 and R017 are above S and below T ohms"

Thanks !!!

[R.G.]

If there is an answer to what you're asking, it's very well hidden. The electronics industry is well past needing voltage controlled resistors, but back in the 1970s and 1980s when this might have been a good solution in analog designs, no one could come up with anything better than sorting and binning JFETs. JFETs were essentially left behind by the industry as alternative solutions were found. Sorted JFETs with drain feedback linearization are about as good as JFET VVRs ever got (as far as I know) and not for lack of smart people trying to make them better.

I suggest that you look at other means of getting a predictable voltage variable resistance. Yes, the circuitry is more complicated, but you might find an approach that can be made to be predictable and reliable.

Other means of setting a voltage variable resistance include
> PWM with switched resistor and/or switched capacitors
> using a center tapped LDR with one section providing feedback for servo-setting the active half
> using an OTA or two as a variable resistor.

[Vivek]

Thanks Sir R.G.

That snippet is from the compressor section of the Rockman X100.

So I feel validated when I felt selecting specific FET that fall into very tight range is the only way.

But what flummoxed me is that those FET that did not work with 5M + 2M trimmer + 5 Meg started to work with 1M + 2M trimmer + 1 Meg

and I never saw any discussion on effect of total resistance of linearisation resistors in the literature except that they be 10 times larger than load. They only spoke about ratio of the two resistors.

[iainpunk]

[off topic?]

Other means of setting a voltage variable resistance include

thermally variable resistance, using heat to lower resistance.
read that in an old text book, but they warned that measures should be taken to isolate the resistor and heater from outside influences, and that the current through the resistor shouldn't be larger than 1% of the heater current.

cheers

[Rob Strand]

You can also use a straight out voltage divider between the gate and source to scale-up/set the effective Vgs.

Best done in moderation.    If you took a very small VP JFET and scaled-up VP with a divider the thing that's not
equivalent is the distortion.  A low VP JFET will have more distortion than a high VP JFET for the same
ac signal on Vds.

The voltage divider will present a load to the circuit driving it.   In most cases you can allow for this loading
or incorporate the divider into the circuit driving the gate - for example turn the release resistor into a
divider.