2 waveform controlled switches-how could I make these work?

[ExpAnonColin]

I don't know my FETs much at all except the 5 or so pages I read on them in one electronics book, and this is just a guess as to how and whether or not it would work, but I'm a tiny bit more confident about the H11F3.  Anyone care to enlighten me?

Thanks,
Colin

[Brian Marshall]

first let me say that fets are a pain in the ass to use, at least if you want anything more than a buffer.  I have been working on a compressor that uses a fet as a variable resistor to ground as a volume control in a non inverting opamp circuit.  the biggest problem is the input at the gat has to be really really hot, and if you use a fet with the same part number you basically have to have a trim pot some where because off voltage is all over the place.

Brian

[ExpAnonColin]

first let me say that fets are a pain in the ass to use, at least if you want anything more than a buffer.  I have been working on a compressor that uses a fet as a variable resistor to ground as a volume control in a non inverting opamp circuit.  the biggest problem is the input at the gat has to be really really hot, and if you use a fet with the same part number you basically have to have a trim pot some where because off voltage is all over the place.

Brian

I figured that sort of thing would make FETs unnattractive-I mean, you don't see them in VCA hardly at all, and you'd think if they were easier to work with they'd be super common.

-Colin

[Brian Marshall]

how about a voltage controlled opamp...  

I know they make them, but not sure how useable they are

[ExpAnonColin]

how about a voltage controlled opamp...  

I know they make them, but not sure how useable they are

How about an operational transconductance op amp?  They do what I want, and I do use them, but I'm curious about FETs.

-Colin

[niftydog]

If you're simply after a modulator, try this device.

As for you're curiousity about FETs, have a look here.

[ExpAnonColin]

If you're simply after a modulator, try this device.

As for you're curiousity about FETs, have a look here.

Ah, the 1496, I've got a design with one of those going on my breadbord right now...

Not looking for a modulator, I'm looking to learn about FETs as voltage controlled switches. Thanks for the link, I'll read it up.

-Colin

[gez]

FETs aren't a pain in the anything.  You'll find them in plenty of schematics and they can work extreemly well as quick & easy variable resistors.  Chips like the 4007 and 4049 are a godsend.

So long as the LFO's waveform isn't too large, distortion can be kept to a min.  I usually aim for .75V - 1V (peak-to-peak) on the gate, depending on the FET(s) I'm using.

Here's one way I do this.  The values shown are back of an envelope calculations - haven't tried this (but should be fine):



They can be used in the signal path too.  I usually use a p-channel MOSFET with the source connected to the inverting input of an op-amp (virtual ground, so is kept stable by negative feedback).  I bias the op-amp from a 5V regulator (making the top resistor of the bias divider 10X smaller than the lower one gets you around 4.5V bias from a 9V supply) and bias the FET using a trim connected from 5V to ground (similar arrangement to above).  Audio is fed into the drain (from an op-amp follower, and can usually be directly connected if the following amp is unity/low gain ).

[Transmogrifox]

FETs can be a little tricky to use as a bilateral variable resistor.  There is an x^2 term in the on/off characteristic that will make it impossible to get linear resistance change from a modulating signal.  

Something I recommend doing for a linearly controlled variable resistor is to pulse-width modulate the FET or a bilateral CMOS switch (CD4066, CD4016) at 100kHz.   This way you get a linearly controlled resistance that depends on the pulse-width.   The most wide pulse approaches the resistance of the CMOS switch, and the narrow pulses increase the resistance linearly to the resistance of the switch's "off" resistance.

Check out MXR envelope filter for ideas.  Lancaster's Active Filter Cookbook also mentions this technique, though Lancaster does not give any ideas about how to generate a pulse-width linearly from a modulating signal.  The MXR envelope filter schematic will provide a solution to this problem.

On a different thread, RG had a good idea about using a multivibrator as a frequency - to - resistance converter.  All you need with that is a voltage-controlled oscillator to modulate its frequency.  Set your one-shot pulse width at about 5 uS  (200 kHz modulating frequency would produce a 50% duty cycle here), then use a VCO to range from maybe 50 kHz to 400 kHz...or 100 kHz to 400 kHz or so...

[ExpAnonColin]

Thanks for the tips, guys.. I guess I'll just await my order of them and play around.  I wouldn't have guessed about the 1V Peak to peak.

-Colin

[puretube]

"Elektor" magazine covers several pwm CMOS-switch operated filter/lfo/phaser schemo`s/projects.
Got all issues, and copied all "important" pages for personal use...
A nice collection over the past 33 years!
Today I got the bill for the 30-est annual subscription...
(happily gonna pay tamorra...)

[Brian Marshall]

Would the PWM controlled fet work the same way as say a class D amplifier... i did a little reading on those... seems like some cools stuff, but never really thought about a practical use for it.

[Transmogrifox]

From what I understand about the class D amp, yes, this is a very similar concept.  If used in a filter that is low-pass or bandpass, your filter takes the high-frequency garbage out for you, and I would argue that if you clock it at 100 kHz+, you will lose no more signal information than you would by using a TL082 instead of an OP27.  It would be interesting to do some tests and measure harmonic distortion on this PWM thing.

[gez]

Something I recommend doing for a linearly controlled variable resistor is to pulse-width modulate the FET or a bilateral CMOS switch (CD4066, CD4016) at 100kHz.   This way you get a linearly controlled resistance that depends on the pulse-width.   The most wide pulse approaches the resistance of the CMOS switch, and the narrow pulses increase the resistance linearly to the resistance of the switch's "off" resistance"

Wow, did this thread take me on a trip!  Periodically I've seen this technique used but never got round to trying it out until now.  The past week I've been trying out various ways of doing this and results have been disappointing.  Linearity is only as good as the circuit doing the PWM.  Many simple circuits just aren't that linear (well, only within a small window).  As well as changing the mark/space ratio, at some point they tend to alter frequency too, resulting in a pretty non-linear response.  Fine in envelope filters, not so great if you want your 'voltage controlled resistor' to accurately track a sine/triangle wave.  

Having said that, how things would pan out in practise is another matter, sometimes non-linearity can be put to good use, but this method isn't necessarily better than using FETs though.

Check out MXR envelope filter for ideas.  Lancaster's Active Filter Cookbook also mentions this technique, though Lancaster does not give any ideas about how to generate a pulse-width linearly from a modulating signal.  The MXR envelope filter schematic will provide a solution to this problem.

This circuit is just as non-linear as all the others I tried (some standard, some cobbled together by myself).  With most of the circuits I tested there is an area where voltage to resistance is pretty linear, but resistance isn't high enough to get much of a workable range.  A lot of this is due to a 50:50 duty cycle not providing enough 'off time' to produce a resonably high resistance, and also the finite 'on resistance' of a 4016 switch.

On a different thread, RG had a good idea about using a multivibrator as a frequency - to - resistance converter.  All you need with that is a voltage-controlled oscillator to modulate its frequency.  Set your one-shot pulse width at about 5 uS  (200 kHz modulating frequency would produce a 50% duty cycle here), then use a VCO to range from maybe 50 kHz to 400 kHz...or 100 kHz to 400 kHz or so...

Again, only as good as your circuit.  Many voltage to frequency converters just aren't that linear (over a decent range that is).  Even with ones that are (I got good linearity with a 4046) there's that problem with the 50:50 duty cycle I mentioned.  It seems that, almost regardless of frequency, the resistance of a 4016 switch isn't that great with this ratio, you'd have to lower frequency well into audio to get it resonably high.  Lowering frequency does increase resistance, but not by as much as you'd think.  Yes you get more 'off time', but you also get more 'on time'.  

I think to get this to work (over the frequency ranges you quoted) you'd need to alter the mark/space ratio significantly in order to increase resistance and get a workable range (something I was unable to do with the most linear voltage to frequency circuit I tried - the 4046 has a 50:50 output and I found it difficult to change this without loosing its linear response).

I hope I don't sound like I'm having a go with this reply, I'm not.  I'd really like to get this to work in LFO driven circuits so that linearity is better than FETs.  Hopefully this will stimulate some debate.

[Transmogrifox]

Ever try this as a LFO-to-PulseWidth-Converter?



I haven't tried it yet, but if you have a fast CMOS switch or JFET, the theory is solid and dependable

Uggghhhh...it displays in the preview, but not after I submit!

here's the url:

http://www.geocities.com/transmogrifox/RFPWM.JPG

[gez]

T-fox, the link doesn't work can you re-post it?

[puretube]

hey gez: it`s the old copy-link-from-browser-title-into-new-browser-window yahoo type problem...

[zachary vex]

if you're set on using a fet or experimenting with using a fet as a sort of audio modulation device, why not just modify the dod true-bypass circuit available at hammer.ampage.org?  you'll have to go forward a few pages to find it (go to the bottom of each page to advance to the next one.)  that's got all of the bugs worked out for jfet switching using a J113, and it has a switch circuit that can easily be modded into an oscillator... in fact, there's one in this thread that you could snip out and insert in there... and what's more, there are two paths for the audio... one turns on when the other turns off, so you could make a stereo trem with it easily.  all the kinks are worked out for low distortion and no clicking, too.

i went to hammer.ampage.org and found the link:

http://ampage.org/hammer/files/dodswitch.gif

[gez]

Thanks for the diagram T-fox (and thanks PT for enabling me to view it!).  

Yes, I did come across this approach but (stupidly) dismissed it due to the higher parts count (than the circuits I eventually ended up testing) and promptly forgot about it.  You're right, the principal is sound and this should work so I'll just bite the bullet and try it out this weekend.

I'll let you know the results.  Thanks again.

[ExpAnonColin]

if you're set on using a fet or experimenting with using a fet as a sort of audio modulation device, why not just modify the dod true-bypass circuit available at hammer.ampage.org?  you'll have to go forward a few pages to find it (go to the bottom of each page to advance to the next one.)  that's got all of the bugs worked out for jfet switching using a J113, and it has a switch circuit that can easily be modded into an oscillator... in fact, there's one in this thread that you could snip out and insert in there... and what's more, there are two paths for the audio... one turns on when the other turns off, so you could make a stereo trem with it easily.  all the kinks are worked out for low distortion and no clicking, too.

i went to hammer.ampage.org and found the link:

http://ampage.org/hammer/files/dodswitch.gif

Not a bad idea...  I'll look into this one for sure.

-Colin