JFET Gain Stage on Orange Squeezer

[Rayman]

I tried to use a JFET gain stage like on the orange squeezer in lieu of the opamp but can't get it to work.  I matched the gain and even tried using two JFETs so the signal isn't inverted, but it doesn't seem to work.  Does anyone have any ideas why the opamp (JRC4558) gain stage works but the JFET gain stage doesn't?  Theoretically I can't think of why it wouldn't, but I'm probably missing something???

[R.G.]

A couple of things come to mind.

You say "JFET gain stage like on the Orange Squeezer". There is no JFET gain stage in the Orange Squeezer. One JFET is a variable resistor to cut the signal level and the other is a constant current source to bias the first one.

If you meant "tried to substitute in a JFET gain stage from somewhere else to replace the opamp" then you have design issues. That opamp does not just do voltage gain. It also supplies the current to fill the 4.7uF envelope detector cap through that 1.5K resistor.

It is possible to replace the opamp with a different amplifier, but it's more complex than just using a JFET, or even two JFETs for noninverting gain. But what you have to do is replace both the 23x voltage gain and the output current drive capability so the 1.5K and rectifier load does not reduce the gain. That means at least an emitter follower output on your voltage amplifier if not a two- or three-transistor gain/follower setup. I don't know of any good way to do that job with a single transistor, although I know a few two-transistor (not two-JFET) circuits that might.

[Rayman]

Once again, thanks R.G.  I knew I was missing something.    I realize an opamp would be easier, I'm just trying to learn everyhting about how the orange squeezer works.   

Do you have two transistor follower/gain circuits you could point me to?

[R.G.]

Do you have two transistor follower/gain circuits you could point me to?

Here are a few:
-The SRPP Mu-amp, especially if made from MOSFETs and used in the way which produces low output impedance as in the article at GEO.
http://geofex.com/Article_Folders/modmuamp/modmuamp.htm
-http://sound.westhost.com/no-opamps.htm
-http://www-personal.engin.umd.umich.edu/~fmeral/ELECTRONICS%20II/05%25c3-Discrete%20BJT/DiscreteAmplifierProject.pdf
-http://mysite.du.edu/~etuttle/electron/elect8.htm

[Rayman]

Excellent, this is exactly what I was trying to learn.  Lot's of good reading material.  Thanks.

[rankot]

You say "JFET gain stage like on the Orange Squeezer". There is no JFET gain stage in the Orange Squeezer. One JFET is a variable resistor to cut the signal level and the other is a constant current source to bias the first one.

I know it's an old thread, but however, I was just trying to figure out few more things about OS, and one of them is - is it possible to use MOSFET instead of a JFET for current source? I'm asking that because JFETs are getting more and more scarce, and using more available alternative could be good. I tried to use 2N7000 and 10k resistor instead of 2k4, and it shows almost the same behavior in LTspice, but did anyone actually tried to build this that (or similar) way?

Also, would using JFET with bigger V_gs,off for Q2 (the one which is used as a variable resistor) give any benefit?

[merlinb]

is it possible to use MOSFET instead of a JFET for current source?

Kind of, but most MOSFETs are enhancement devices, unlike JFETs, so the configuration would be different. You can make a current source more easily from a regular BJT, so no biggie.  In fact, the OS doesn't really need a current source at all. A resistor from a zener-stabilised voltage would do just as well, e.g. see the SSL talkback compressor.

Also, would using JFET with bigger V_gs,off for Q2 (the one which is used as a variable resistor) give any benefit?

A bigger Vgs(off) means you needs more voltage swing to control it -something we don't have the luxury of in a 9V environment.

[rankot]

So actually I have to provide some 1.6V for Q2 source, right?

[merlinb]

So actually I have to provide some 1.6V for Q2 source, right?

You have to provide a voltage equal to the cut-off voltage of the other (gain control) JFET, so when there is no audio signal the JFET is basically 'off'. This won't always be 1.6V, it depends on your exact JFET. They vary a lot, which is why there needs to be a trimpot.

Assuming the opamp can swing about 3Vpk, and you need to overcome the side chain rectifier diode drop, that leaves about 2.4Vpeak of control voltage from no signal to max output. So ideally you need a JFET with a cut-off voltage no greater than 2.4V, otherwise you'll lose some range of compression.

[rankot]

Thanks Merlin, you are very helpful, as always! So if I use 1N5819 for rectification, it will be 2.8V. I have some J310 JFETs with ~3V V_gs,off, so I can try with them. Just wanted to know if there is any benefit of bigger V_gs,off.