Educate me regarding using JFETs as voltage controlled resistors please...

[frequencycentral]

Can a J201 replace an LDR for example....? How would they be set up? Downsides? Anything else I should know.......?

[Ice-9]

I'm not too sure where you want to go with this, but the obvious answer is that the J201 isn't light controlled, maybe it could be used in a similar way using a different control circuit. What lines of experiment are you thinking of here ?

[Ice-9]

a link to basic info on fet voltage control resistor


http://graffiti.virgin.net/ljmayes.mal/comp/vcr.htm

[frequencycentral]

I've been looking at some tremolo/vibrato/phaser circuits. Some use LDRs, some use JFETs. My understanding is that an LDR increases/decreases it's resistance as the control LFO cycles, influencing the brightness of the LED or lamp. I'm wondering if applying said LFO to the gate of a JFET does the same thing.

EDIT: Obviously, I'm looking at replacing the whole LED or lamp/LDR optocoupler with a FET - if that's possible.

[Ice-9]

I think it it certainly would, i have a diagram for a fet controlled tremolo using a LFO to control the resistance of the fet , i will try and find it, this is from memory of a book i had years ago by R.A. Penfold . I will have to check that it does what i think it does because as i said its a long time since i have looked at this book, so my memory of the cicuit might not be correct.
I will be in the loft in the morning to look for it.

[frequencycentral]

I think it it certainly would, i have a diagram for a fet controlled tremolo using a LFO to control the resistance of the fet , i will try and find it, this is from memory of a book i had years ago by R.A. Penfold . I will have to check that it does what i think it does because as i said its a long time since i have looked at this book, so my memory of the cicuit might not be correct.
I will be in the loft in the morning to look for it.

Thanks Mick, thats great. I'll breadboard, I just needed to know it's theoretically possible - so many times when I breadboard new ideas I feel I may be really close to success but fall at the last hurdle - you don't know that the 95% works until the 100% happens. That link is useful, if only for the R1/R2 arrangement, the math hurts though!

[Ice-9]

I have just re-read this bit of your post

"My understanding is that an LDR increases/decreases it's resistance as the control LFO cycles, influencing the brightness of the LED or lamp"

The LFO changes the brightness / frequency of flashing of the LED's which the LDR see's thus changing the resistance. ( the LED influences the LDR not the other way)

[frequencycentral]

I have just re-read this bit of your post

"My understanding is that an LDR increases/decreases it's resistance as the control LFO cycles, influencing the brightness of the LED or lamp"

The LFO changes the brightness / frequency of flashing of the LED's which the LDR see's thus changing the resistance. ( the LED influences the LDR not the other way)

Duh - oh yeah, I knew what I meant but my fingers typed different! Hey, it's late! I'm going to bed!

[cpm]

look at the "EA tremolo" for a quick example

one tricky thing is to achieve linearity, but this also applies for the LDR thing, i guess

[R.G.]

A resistor is a device where the voltage across it is proportional to the current through it, and vice versa.

A linear resistor is such a device where the constant of proportionality R, which is equal to the voltage divided by the current, does not change with either voltage or current.

A variable resistor is such a device where the constant R (=V/I) can be changed by some external operating means. A pot is a mechanically varied resistor. A Light Dependent Resistor is an optically variable resistor.

A JFET has a small region of its operation where the drain to source channel behaves as a voltage controlled resistor. The region in the unmodified JFET is that portion of the JFET's operation where the voltage between drain and source is less than the pinchoff voltage for the JFET. Get outside that region and it starts distorting (that is, "R" starts changing with the channel voltage/current, not just with the gate-to-source voltage). So the voltage across the JFET is one limitation, and this must be smallish, usually a couple of volts maximum. The current in a JFET can never be larger than Idss, and in the resistor region, it must be even less. In addition, there is a square-law distortion inside the resistive region of a few percent, worse as the signal approaches the entire resistive region.

The control voltage on a JFET is the voltage between the gate and source. If the source is wiggling around with signal voltages, then the gate must either follow the same wiggle plus the control voltage or distortion is introduced. In practice, people look for ways to use JFETs as variable resistors where the source can be tied to some reference voltage, bias voltage, etc. and then the control voltage can be used without extra worries about the signal modulating the control voltage. It is difficult to make a fully floating voltage variable resistor out of a JFET. This is why people turn to LED/LDR modules - it is incredibly easy not to let the control signal modulate the LDR; it's completely insensitive to control feedthrough, and JFETs are not.

Finally, it is possible to use a couple of resistors to feed back about half the drain voltage to the gate of a JFET. This helps with linearizing the variable resistance; it about doubles the signal which it can carry without distorting a lot, and helps with control signal rejection. It also reduces the sensitivity to the control signal by about half as well, so the control voltage needs to be bigger.

Changing to JFETs from LED/LDRs is tricky electronically because you must get the source and drain voltages accurately set up, as well as getting both the DC level and AC size of the control signal set up properly (and reliably!) to make this work well without introducing ticks, thumps and swishes from the control voltage itself into the audio path.

[frequencycentral]

Thanks R.G.

So given that I am a monkey-with-a-machine-gun E.E., should I try another approach? Are LEDs/LDRs the way to go, or should I consider configuring a LM13700 and a VCR too?

EDIT: 13700 floating VCRs uses both the chips OTA's, so to make two would need two 13700 - that seems overkill.

[nelson]

One word - OptoFet.

[R.G.]

So given that I am a monkey-with-a-machine-gun E.E., should I try another approach? Are LEDs/LDRs the way to go, or should I consider configuring a LM13700 and a VCR too?
EDIT: 13700 floating VCRs uses both the chips OTA's, so to make two would need two 13700 - that seems overkill.

Monkey with a machine gun? Sounds like fun! 

The approach depends on what you're trying to do and how much time and effort you're willing to invest in getting it to be how good. There is no "the way to go". There are only tradeoffs. LDRs have built-in distortion that changes with light level, but not many people know that. JFETs are limited for signal level and suffer from feedthrough. MOSFETs are good, but suffer from the body diode, which limits signal level even further. OTAs are great, but complicated to set up for floating resistors and have noise problems. Photo-Fets like the H11F1 series get rid of the control feedthrough at the expense of small signal handling and non-linear response to LED current; but they're great as signal switches.

As you might guess, the Existence Theorm applies - if a "The Best" existed, you wouldn't need to ask the question, as "The Best" would already be everywhere and you'd already know it. Just like the stock market. No one can predict it. We know this because if you COULD predict it accurately, you'd start accumulating money, and eventually you'd have it all. No one has it all, nor even most of it, so the method cannot exist.

So what are you trying to do and what are your limits on what you're willing to invest in terms of money, time, new design, and what will you accept in terms of distortion, max signal, feedthrough, etc.? Maybe we can figure it out.

[Ice-9]

The circuit i had mentioned in an earlier post uses a 555 timer to generate a square wave to drive the fet, It looks more like this will switch on/off the fet rather than controling it like a vr. With some mods you should be able to drive the fet over its normal operating range (hope i've worded this correctly)
If you still want a look at the schematic i will post it up .

[frequencycentral]

So what are you trying to do and what are your limits on what you're willing to invest in terms of money, time, new design, and what will you accept in terms of distortion, max signal, feedthrough, etc.? Maybe we can figure it out.

I've set my self a number of challenges, all based around low voltage submini tubes. I've already had some success and a great deal of learning - all of which I share on this forum. I'm not a manufacturer, I'm a 'gentleman scientist', whatever I develop is just for me, and anyone else who cares to build from the schematics I share.

My current goals are:

- low voltage tube vibrato: as you know I've already done some work on this, but its the phase splitting of the PSO which is stumping me, so I thought I'd try a different approach: Vibrato Idea

- low voltage tube voltage controlled filter: I've played about with a few ideas, this will be my next attempt, based on Rene Schmitz's design: VCF Idea

- low voltage tube phaser: this idea I haven't played about with yet, but the control method will be influenced by whatever comes out of the two above.


I will use LDRs if I have to, for some reason though I would prefer not to.

[frequencycentral]

The circuit i had mentioned in an earlier post uses a 555 timer to generate a square wave to drive the fet, It looks more like this will switch on/off the fet rather than controling it like a vr. With some mods you should be able to drive the fet over its normal operating range (hope i've worded this correctly)
If you still want a look at the schematic i will post it up .

Yes, please post it, any examples will help.

[frequencycentral]

.................and if I could find existant designs I would just build them instead of attemping to design my own! Then again - I learn so much from breadboarding ideas - and there is a certain satisfaction/kudos to be gained from originating and sharing!

[DrBoogey]

Hi, maybe thats somthing what you are looking for. An Op Amp And Two JFETs Form A Voltage-Controlled Amplifier.
Schematic:
http://electronicdesign.com/Files/29/10116/Figure_01.gif
And how it works:
http://electronicdesign.com/Articles/Index.cfm?AD=1&ArticleID=10116

[R.G.]

I've set my self a number of challenges, all based around low voltage submini tubes. I've already had some success and a great deal of learning - all of which I share on this forum. I'm not a manufacturer, I'm a 'gentleman scientist', whatever I develop is just for me, and anyone else who cares to build from the schematics I share.

The gentleman scientist tradition is an honorable one. That's how Ohm got to be recognized.

- low voltage tube vibrato: as you know I've already done some work on this, but its the phase splitting of the PSO which is stumping me, so I thought I'd try a different approach: Vibrato Idea

Two things come to mind.
First, you're finding out first hand that LFOs are not as simple as they appear. One has to get the static DC level correct for the device being controlled, and it has to stay there solidly. Then on top of that, one has to get the right amount of the AC component from the LFO imposed on top of the DC without changing the DC. This isn't too hard for one LFO waveform, but when you have to get one plus its inverse correct, it gets sticky. In your case, you have to add a derived second LFO on top of a DC level for the tubes to work, and it's difficult by definition to couple a very low frequency signal (the LFO) in such situations. Opamps which can independently set an output DC level from two input DC levels make this easier, but it can be done discretely.  Sorry. Pun is accidental.  

Using JFETs for VCRs, the ideal condition is that the drain and source are at the same DC level. Then the gate's offset from that level is easier to manage, and because of the very high gate impedance, you can often AC couple in the LFO signal. For N-channel devices, the source and drain have to be a few volts higher than the gate, so it's most convenient to hold the source and drain at a positive voltage with respect to a negative ground, and operate the gate LFO at a few volts below that. The signal voltage is then arranged to sit at the same DC level as the drain and source. The MXR phase 90 does most of this.

In the P90, the source is tied to a +3V bias voltage. The same voltage is used to apply bias to opamps through a high resistor, so the drain and source are at the same DC voltage. The signal is set to the same DC voltage and kept there for the whole chain. The gate drive is generated in a way that it oscillates from nearly ground to less than the bias voltage. This ensures that it goes from allowing the lowest resistance (Vgate-to-source is nearly zero)  smoothly to the highest resistance (Vgate-to-source is several volts).

You can do much the same thing with P-channel JFETs, but they have the advantage that voltages positive with respect to the channel turn them off. So a suitable P-channel JFET can be run at a DC voltage of ground and all the control voltage will be positive with respect to that. However, one major reason P-channel devices don't own this application is that they are so expensive and hard to find in correct values.

But the key for JFET variable resistors is (a) get the source and drain at the same DC voltage and get the signal to ride on top of that DC level, perhaps by AC coupling the signal into and out of the JFET and (b) get the control voltage sitting on a DC level a couple of volts below the channel, and add the LFO signal to that so the LFO runs the gate from cutoff up to nearly zero Vgs.

- low voltage tube voltage controlled filter: I've played about with a few ideas, this will be my next attempt, based on Rene Schmitz's design: VCF Idea

In this one, you see the 1M pulling the JFET section to 0Vdc right after the input follower? That 1M could go to some reference DC voltage as easily as ground; all the signal paths are DC -blocked by capacitors. Then the LFO can be applied to the "CV input".

- low voltage tube phaser: this idea I haven't played about with yet, but the control method will be influenced by whatever comes out of the two above.

I think the same trick as the VCF applies, except that the DC voltage on the previous cathode can be used as your bias voltage if you can get the several cathodes you'll be using for multiple stages to be at the same DC level.

[frequencycentral]

Thanks again R.G. I'll digest your reply. It reminds me of just how much I don't know though!