2N7000 variable resistance

[Matthew Sanford]

For posterity

Been thinking on this awhile, finally set down to try it. After all, vactrols eat current in ways DMOS doesn't. So I set a couple source to drain, other drain to 5v, other source to ground. At first 10ks in parallel to each FET and 10k trims as dividers for the LFO & complement (to achieve lower gate voltages, they bring 0.25 - 4.75 from MCP6002 post StompLFO 5v wave) only allowed a one volt range. Weird was with the trimpot taken straight to ground the voltage in the output still changed. So after switching to highest resistance on hand (10M & 2M trims) I was able to tune it to do a 3 volt range. To get that, the voltage side trimpot is 2M to ground (less lowers top voltage) and ground side trimpot is 2M 1M5 in series, 500k to ground (less flatlines the start then becomes 1.65v bottom, more increases the bottom voltage in).

I did swap the FETs around as they weren't matched, but everything worked the same. I had thought the 4.75 volts would make it top out so there would be flat spots during the ramp up wave, but it doesn't seem so. So I don't get it, but am happy with the result. (I guess datasheet does show 0.8 - 3 volt min-max gate threshold) Perhaps increasing the resistance in the trimpot could increase the range as 10k to 2M gave a 200% increase? Still, happy to test it in something.

Not sure how to, or if possible to, increase range but may not need it. I think I'll set up Christine on the breadboard to have this mess the pots, she is very happy to glitch, sending howls for an oscillation change, bass drum thumps for gain and power starve in opposition, so should demonstrate well.

[R.G.]

MOSFETs make great variable resistors as long as the signal across them is notably smaller than one diode drop. There is an inherent reverse diode across the source to drain as a result of the semiconductor fabrication process. This can be extended a bit with feedback tricks, up to a volt or two. The transition to not-a-resistor happens somewhat abruptly when the drain I-V curve flattens.I suspect you're running into some of this as the voltage get larger.

[Matthew Sanford]

MOSFETs make great variable resistors as long as the signal across them is notably smaller than one diode drop. There is an inherent reverse diode across the source to drain as a result of the semiconductor fabrication process. This can be extended a bit with feedback tricks, up to a volt or two. The transition to not-a-resistor happens somewhat abruptly when the drain I-V curve flattens.I suspect you're running into some of this as the voltage get larger.

So I am supposing the large resistor in parallel, source to drain, is the feedback trick? Prior to was one volt, after was 3. Good to know...and essentially there will be no perfect transition to 0 or full like a pot. Had a thought to do power supply decoupling with say 2 100uF in series, take the hash from the middle and possibly amplify it to run the gates for a random wiggle, so maybe I'll try it too. Christine seems happiest with pots wiggling

[antonis]

So I am supposing the large resistor in parallel, source to drain, is the feedback trick?

I'm not sure if 10M resistors could serve as feedback configuration 'cause MOSFET's output resistance (r_o = 1/g_os) is much smaller than those resistors values..

Of course, MOSFETs here don't work in the quadratic region of Drain current, but..

[R.G.]

So I am supposing the large resistor in parallel, source to drain, is the feedback trick?

No, it's a resistor in series with the gate and feedback from the drain to gate. This linearizes the variable resistance region of the drain curves and enlarges the amount of signal usable because the actual voltage at the gate is half the size of the signal at the drain.
Resistor from drain to source softens the transition from on to off as the MOSFET off is hugely many megohms.

[Rob Strand]

I went through the better topologies in this post,

https://www.diystompboxes.com/smfforum/index.php?topic=52905.msg1142151#msg1142151

https://postimg.cc/kDPkthpC


MOSFETs have a fairly low on resistance.  Typically somewhat lower than JFETs.  It's a good idea to keep the circuit impedances low so the circuit operates with a lower valued variable resistance.   That way the MOSFET does not need to operate so close to the turn-off voltage.

[Matthew Sanford]

This all explains a lot. I will try the two with sources tied and large drain/gate resistor, perhaps a different circuit's pots to try it on as 6k max is well under the 100k/1M on Christine.

I am a bit confused as to why, as-is, it was refecting the ramp up control voltage LFO (and compliment) from the middle; there was some gate resistance in but no drain/gate connection...albeit excessive gate voltage.

[Rob Strand]

This all explains a lot. I will try the two with sources tied and large drain/gate resistor, perhaps a different circuit's pots to try it on as 6k max is well under the 100k/1M on Christine.

I am a bit confused as to why, as-is, it was refecting the ramp up control voltage LFO (and compliment) from the middle; there was some gate resistance in but no drain/gate connection...albeit excessive gate voltage.

If I understand what you are doing correct there's a good chance the top MOSFET is operating as a source follower and just tracking the input voltage.

When the voltage across the MOSFET is high it acts like a voltage controlled current source.   If you had a resistor from the lower MOSFETs drain to +V and vary the gate voltage you can vary the drain voltage.   However, all you are seeing is Vout = Vsupply - Rdrain * Idrain.    The MOSFET doesn't need to operate as a variable resistor to do that, just a voltage controlled current source.

In order to see the MOSFET operating as a variable resistor the voltage across the drain and source needs to be low - as per RG's comments.

When you use JFET as an ac resistance use can use a single device for an AC controlled resistance and it works over a reasonable range of AC swings.   A single MOSFET will also work but MOSFETs have a built in body diode which will conduct in one of the polarities of the AC signal unless the voltage across the MOSFET is quite small.   The circuit(s) I gave try to make the controlled resistor circuit operate over a
wider range of voltage (and be low distortion).

The icing on the cake is the RC linearization networks, which can also be used on JFETs.

[Matthew Sanford]

This all explains a lot. I will try the two with sources tied and large drain/gate resistor, perhaps a different circuit's pots to try it on as 6k max is well under the 100k/1M on Christine.

I am a bit confused as to why, as-is, it was refecting the ramp up control voltage LFO (and compliment) from the middle; there was some gate resistance in but no drain/gate connection...albeit excessive gate voltage.

If I understand what you are doing correct there's a good chance the top MOSFET is operating as a source follower and just tracking the input voltage.

So then it follows the top gate voltage, would that mean then the lower would also be a source follower, albeit following a lower (though high for V_th) gate voltage yet as inverted (drain to output) which maintains the similar linear appearance once mixed out?

When the voltage across the MOSFET is high it acts like a voltage controlled current source.   If you had a resistor from the lower MOSFETs drain to +V and vary the gate voltage you can vary the drain voltage.   However, all you are seeing is Vout = Vsupply - Rdrain * Idrain.    The MOSFET doesn't need to operate as a variable resistor to do that, just a voltage controlled current source.

In order to see the MOSFET operating as a variable resistor the voltage across the drain and source needs to be low - as per RG's comments.

When you use JFET as an ac resistance use can use a single device for an AC controlled resistance and it works over a reasonable range of AC swings.   A single MOSFET will also work but MOSFETs have a built in body diode which will conduct in one of the polarities of the AC signal unless the voltage across the MOSFET is quite small.   The circuit(s) I gave try to make the controlled resistor circuit operate over a
wider range of voltage (and be low distortion).

The icing on the cake is the RC linearization networks, which can also be used on JFETs.

Hmmm. Much to read and think on. I will try the ways you'd suggested, and put it on something where a variable 2k would work.
Thanks again!

[Rob Strand]

So then it follows the top gate voltage, would that mean then the lower would also be a source follower, albeit following a lower (though high for V_th) gate voltage yet as inverted (drain to output) which maintains the similar linear appearance once mixed out?

Yes, I'm sure that is at least part of what you are seeing.