the memory fet

[Brian Marshall]

say you want a constant ouput voltage of some sort for what ever reason.  anyone think somthing like this would work.  I havent tried it yet, but thought of it a while back ago, and finally got arround to drawing it.

this is it's simplest formation (well i guess the drain resistor isnt really needed)  but not really a practical circuit.

basically you close the switch set the voltage via the pot, and then the cap holds a charge, and the fet (in theory) keeps a constant output voltage.

The problem is input impedance of the fet.  I think fets like a mpf102 are something like 100Meg, but eventually the cap would discharge i guess.

What does everyone think?
I'm sure this is probably pretty close to the way old old computers worked???!?!?!?!  sure it's been used before, but not sure how reliable.

[niftydog]

Old computers?!?!
You've just invented a RAM cell!  

Trouble is, you may need to be constantly "updating" the value - depending on the size of the cap and charge/discharge rates etc.

[Peter Snowberg]

Joe Davisson posted a schematic for something similar a few months back. It used individual pushbuttons to raise or lower the charge, but the concept is the same (I think it was MOSFET in his diagram).

Oddly enough, DRAM still works similarly today. There's a tiny cap that either has or doesn't have a charge. The charge only lasts milliseconds after being placed there and it gets drained just by the act of checking to see if there is a charge there, so each cap is constantly being "refreshed" by reading it and storing the result back into the cap.

I would think it would work fine for many control requirements and it has the advantage of lacking digital precision.

I don't know about the leakage of the protection diodes, but opamps like the TL072 have the input resistance speced at 10^12 ohms. That shouldn't put too much strain on a 10uF cap.

Take care,
-Peter

[Jason Stout]

Here's Joe Davisson's take on it:

http://analogalchemy.com/pedals/mosbuttons.gif

Cool circuit!

[Brian Marshall]

Joe Davisson posted a schematic for something similar a few months back. It used individual pushbuttons to raise or lower the charge, but the concept is the same (I think it was MOSFET in his diagram).

Oddly enough, DRAM still works similarly today. There's a tiny cap that either has or doesn't have a charge. The charge only lasts milliseconds after being placed there and it gets drained just by the act of checking to see if there is a charge there, so each cap is constantly being "refreshed" by reading it and storing the result back into the cap.

I would think it would work fine for many control requirements and it has the advantage of lacking digital precision.

I don't know about the leakage of the protection diodes, but opamps like the TL072 have the input resistance speced at 10^12 ohms. That shouldn't put too much strain on a 10uF cap.

Take care,
-Peter

I was sort of planning on using an opamp (if i ever do anything with it) just because the voltage swing is greater, and just seems overall more useful and reliable.

i drew the fet in the schematic just for arguments sake.

[petemoore]

If I were to apply this, it would most likely be a parameter that I would update fairly often.
 The drain rates [I guess] would probably be so slight, that the time frames between adjustments would be so small I'd never notice a change in the parameter.
 Changing the parameter over time [if you could calculate it] is cool, I like slow drops and swells.
 I had a EZ Vibe setup [accidentally'] for fading sweep, and a switch [momentary power supply cut] wired to reset the full phaze, to med phaze, to no phaze sweep ..cycle.  The effect is on one of the favored recordings of mine.
 Ive since got it to maintain sweep, but that Was cool having that effect. No dount/

[Brian Marshall]

If I were to apply this, it would most likely be a parameter that I would update fairly often.
 The drain rates [I guess] would probably be so slight, that the time frames between adjustments would be so small I'd never notice a change in the parameter.
 Changing the parameter over time [if you could calculate it] is cool, I like slow drops and swells.
 I had a EZ Vibe setup [accidentally'] for fading sweep, and a switch [momentary power supply cut] wired to reset the full phaze, to med phaze, to no phaze sweep ..cycle.  The effect is on one of the favored recordings of mine.
 Ive since got it to maintain sweep, but that Was cool having that effect. No dount/

well the way i was thinking of it working, you could use a 6 possition rotary switch, and only ONE pot instead of 6.  ofcourse this would require a bunch of other parts to make up for it, but still interesting.

Anyways,  i calculated that if you use a 10 uf Cap,  and a TL072, it would take about 2 years for it to completely discharge.  but not really sure if i did the math correctly.  Also caps discharge more quickly (push harder) when their charge is at its greatest, so you might notice a difference of say 10% pretty quicly.   That i dont know the math for, but I'm sure it involves  period(sin*.1)  maybe i dont know i'm just typing out loud.

bri

[gez]

If you want voltages lower than your supply you can simply use regulators and switch between them, or use a variable reg with a pot.

For a fuller range you can use a suitable op-amp (something like a TDA2130) and bias the thing from a regulator (+ve input to the wiper of a pot with outer lugs going from the regulator's output to ground if you want variable output, or to the junction of a divider for fixed).  With an op-amp you can have DC gain so you can wire it up in non inverting mode with a trimpot from output to ground, wiper connects to -ve input.  This allows for a wide range of voltages (above that of the regulator) but gives you a steady regulated supply (within the confines of the initial supply).  

You don't need caps to do this (unless you want the thing to ramp up gradually).

PS The above is a conventional circuit (not my idea).

PPS I don't think any 10u cap is going to retain its charge for 2 years  

[Ge_Whiz]

You've also described something like a voltage-controlled synthesizer "sample and hold" circuit.

[R.G.]

There is at least one circuit for something similar in one of the E&MM articles. I believe that the ramped capacitor voltage was buffered by a CMOS input opamp and drive a VCA. There was an "UP" switch and a "Down" switch and a rate pot to set how fast the cap moved up or down.

But yes, both DRAM computer memories and sample-holds use something similar.

It's not much help, perhaps, but the gate-source of a large power MOSFET looks like a (typically) 1000pf capacitor. If you can use cap values something like that, you have the capacitor integrated into the MOSFET. MOSFET gate capacitance is very high quality for a cap, and very low leakage.

[Brian Marshall]

If you want voltages lower than your supply you can simply use regulators and switch between them, or use a variable reg with a pot.

For a fuller range you can use a suitable op-amp (something like a TDA2130) and bias the thing from a regulator (+ve input to the wiper of a pot with outer lugs going from the regulator's output to ground if you want variable output, or to the junction of a divider for fixed).  With an op-amp you can have DC gain so you can wire it up in non inverting mode with a trimpot from output to ground, wiper connects to -ve input.  This allows for a wide range of voltages (above that of the regulator) but gives you a steady regulated supply (within the confines of the initial supply).  

You don't need caps to do this (unless you want the thing to ramp up gradually).

PS The above is a conventional circuit (not my idea).

PPS I don't think any 10u cap is going to retain its charge for 2 years  

Well the whole point is that the pot would be disconected most of the time.