Orange squeezer question

[Fancy Lime]

Hi all,

I've been pondering a suitable limiter/compressor design for integration into a bass preamp and one of the possible basic topologies is obviiusly the orange squeezer. One thing that has always puzzled me about this design is this: why is the bottom of the variable resistor JFET fed by a constant current source (the left JFET)? Why is this better than simply biasing it betweenntwo resistors? It should improve ripple rejection a little bit, but how much better is it really? For reference, R.G.'s classic schematic:



If I understand the circuit correctly, then what we want at this particular node is a stable voltage, no? This is achieved by putting the 4u7 cap between the constant current JFET and the bias pot. But wouldn't it be both better and easier to have a BJT wired as a variable voltage reference instad of the JFET? Maybe Dan Armstrong had 2N5457s to burn but I don't plan a clone but rather a new design that simply might use the same basic idea of "L-pad with JFET as variable R ground leg" as the O'Squeezer or the Urei 1176, so I'm trying to figure out what lies behind the design choices of the old classiscs: what are genious ideas that I just don't get and what are design flaws worth fixing.

Thanks and cheers,
Andy

[Eb7+9]

If you think the FET in the 1176 circuit works the same way as it does in the OS (ie., as a voltage divider) you're not analyzing the circuit right ...
the 1176 designer does a much smarter thing to handle larger signals than would the same FET in a divider ...

but that's a side point

---

you are correct in assuming that a good AC ground is required in the OS, and that the original design provides it thru a shunting cap ...
which in itself is passable for basic operation

the reason for using a jFET current source and resistor arrangement there is that it provides current invariance against a varying (dying battery) source voltage ...
it's a convenient (same FET) low part-count solution to generating a constant current

but this also sets the threshold of the limiter circuit ...
which varies from FET pair to FET pair

and in general seems to have been "chosen" for low output instruments - ie., for single-coil guitars pickups

for Bass - or any kind of generally useful larger-scale response - you'd want to make this threshold voltage variable ...
at the very least so that you can go from light spank (Limiter) to heavy (Compressor)

otherwise, you're always stuck trying to match for optimum instrument levels to the circuit ...
ie., hoping for the best, and with little to no room for adjustments

---

these design constraints can all be covered by replacing the FET-load-cap circuit by an active (variable) back-bias that provides both
the AC ground and a variable range of Threshold adjustment

I'm sure y'all can guess what that might look like ...

[Mark Hammer]

Hi J.C.,

Question:  If "the reason for using a jFET current source and resistor arrangement there is that it provides current invariance against a varying (dying battery) source voltage ...it's a convenient (same FET) low part-count solution to generating a constant current", then what is still required if the circuit is powered by the average pedalboard supply, rather than a battery?  One can certainly understand the compromises Dan Armstrong made, and had to make, when the entire series of little plug-in boxes were originally designed.  But are all those compromises and adjustments still necessary, now that the era of 9V batteries is largely behind us?  Or can changes be made that don't interfere with proper functioning?

[PRR]

If you think the FET in the 1176 circuit works the same way as it does in the OS (ie., as a voltage divider) you're not analyzing the circuit right ...

[Fancy Lime]

If you think the FET in the 1176 circuit works the same way as it does in the OS (ie., as a voltage divider) you're not analyzing the circuit right ...
the 1176 designer does a much smarter thing to handle larger signals than would the same FET in a divider ...

but that's a side point

---

you are correct in assuming that a good AC ground is required in the OS, and that the original design provides it thru a shunting cap ...
which in itself is passable for basic operation

the reason for using a jFET current source and resistor arrangement there is that it provides current invariance against a varying (dying battery) source voltage ...
it's a convenient (same FET) low part-count solution to generating a constant current

but this also sets the threshold of the limiter circuit ...
which varies from FET pair to FET pair

and in general seems to have been "chosen" for low output instruments - ie., for single-coil guitars pickups

for Bass - or any kind of generally useful larger-scale response - you'd want to make this threshold voltage variable ...
at the very least so that you can go from light spank (Limiter) to heavy (Compressor)

otherwise, you're always stuck trying to match for optimum instrument levels to the circuit ...
ie., hoping for the best, and with little to no room for adjustments

---

these design constraints can all be covered by replacing the FET-load-cap circuit by an active (variable) back-bias that provides both
the AC ground and a variable range of Threshold adjustment

I'm sure y'all can guess what that might look like ...

Ah, I see, that makes sense. I had not considered the effects if a waning battery on the bias/threshold. The constant current source really is a very smart solution for that if we want to set the bias once via a trimpot and then forget about it. However, it becomes pointless if we put the bias pot on the outside to use as a threshold control. At least if we can assume a half-decent quality power source.

My current plan is to feed the bias at the shunt cap via a BJT buffer, which has its base tied to the wiper of a 100k pot between ground and 9V. This should then be able to provide a very stable voltage, which can be set by the user to control the threshold.

1176: You are right, I wasn't analyzing that one right. I had only ever seen incomplete and blurry schematics and simply assumed it to be L-pad based. My mistake. I have now looked at a better looking schematic and see that it works by hampering the amplification of a transistor. The nice thing here  is that there is "no" (or at least not much) AC across the FET, so no distortion from that side. Smart. However, I am currently unable to see, if this design allows attenuation of the input signal below its original amplitude (which is what I want) or merely a reduction of amplification (no use to me in this case, plus there are easier ways to skin this particular cat, at least for the purposes of a guitar/bass pedal). I might need to do some spicing when I get around to it. In the meantime, I am not averse to being schooled on the matter by my betters...

Thanks and cheers,
Andy


Edit: Wait, now I'm confused. While I typed, Paul posted a snippet from the 1176, which looks nothing like the ones I found and indeed does look like a regular old voltage divider to ground. So I might have been remembering cirrectly after all. Is there a chance that the many revisions of the 1176 don't all fundamentally work the same? For reference, I had been looking at this one:

[printer2]

---

these design constraints can all be covered by replacing the FET-load-cap circuit by an active (variable) back-bias that provides both
the AC ground and a variable range of Threshold adjustment

I'm sure y'all can guess what that might look like ...

You might be assuming too much from some of us.

[ElectricDruid]

You might be assuming too much from some of us.

I agree...

This, for example, I find "concise" to the point of unintelligibility! Sorry Paul! But I don't know from these diagrams alone what the point you're trying to make is.
It looks a bit like he's re-drawn the OS so it looks more like the 1176, but then I'm not sure what we're supposed to take away from that. That they're *still* not the same? Or that after that "Hey look! They're basically the same!"? They're clearly not *identical* but whether the differences are important is the bit that he's not telling us, and that's the critical part.

[PRR]

Paul posted a snippet from the 1176, which looks nothing like the ones I found

So show ones you found.

[Fancy Lime]

Paul posted a snippet from the 1176, which looks nothing like the ones I found

So show ones you found.

I have. Revision K is at the bottom of post #4. Does it not show? Here is revision D:

[Rob Strand]

Does it not show? Here is revision D:

I couldn't see it - maybe the lines of the GIF are thin and it shows a white box.

Rev K:
-----
[Click box to expand.]

------

Here's Paul's (Rev A?) and Rev F.  (technically *Suffix* A, D, F, K models)
http://www225.pair.com/audio/waltzingbear/Schematics/Urei/

The Rev A uses the old school method for reducing distortion by modulating the gate, whereas Rev D (and the later versions) modulate the source.

This site had some more info and also some other schematics (rev H),
https://www.masonaudio.org/diy/comp1176

[Fancy Lime]

Does it not show? Here is revision D:

I couldn't see it - maybe the lines of the GIF are thin and it shows a white box.

Rev K:
-----
[Click box to expand.]

------

Here's Paul's (Rev A?) and Rev F.  (technically *Suffix* A, D, F, K models)
http://www225.pair.com/audio/waltzingbear/Schematics/Urei/

The Rev A uses the old school method for reducing distortion by modulating the gate, whereas Rev D (and the later versions) modulate the source.

This site had some more info and also some other schematics (rev H),
https://www.masonaudio.org/diy/comp1176

Thanks Rob! That clears it up a lot. Interesting that the different 1176 models are so different. I think for the purpose of a guitar/bass limiter/compressor, the "old school method" is probably preferable. Some of that typical FET compressor distortion sounds really nice on instruments, at least to my ear. For mastering or on vocals that would be a different story.

[ElectricDruid]

This site had some more info and also some other schematics (rev H),
https://www.masonaudio.org/diy/comp1176

That's a great site, thanks Rob. A very interesting read.

[Rob Strand]

Thanks Rob! That clears it up a lot. Interesting that the different 1176 models are so different. I think for the purpose of a guitar/bass limiter/compressor, the "old school method" is probably preferable. Some of that typical FET compressor distortion sounds really nice on instruments, at least to my ear. For mastering or on vocals that would be a different story.

It a bit like "I'm Spartacus"... "I'm also Spartacus" ... Which Spartacus do you want?    I've noticed some of the plug-ins specify which version they are modeled from.

The main technical beef of the "old school method" is the control signal feeds into the audio path via the gate resistors.   Alternatives usually work to avoid that, hence the change to modulating the source on later models.

As a heads up you might want to look at the JFET compressors in the SWR amps and the Eden amps.

That's a great site, thanks Rob. A very interesting read.

Yes, there's people out there who have put a lot of time into it.  (I stumbled on that site at some point and saved the link.)

[Fancy Lime]

Hi Rob,
thanks fir the hint about SWR. Found a few schematics of preamps and even an extracted limiter schematic:



Interesting design but it looks to me like there is a considerable portion of the full output swing across the 5457 at high sensitivity settings, no? Curious choice, if the limiter is supposed to be power amp protection rather than a tone shaping device (which may or may not be the case, I just always assume that when I read "limiter" rather than "compressor").

[Rob Strand]

Interesting design but it looks to me like there is a considerable portion of the full output swing across the 5457 at high sensitivity settings, no? Curious choice, if the limiter is supposed to be power amp protection rather than a tone shaping device (which may or may not be the case, I just always assume that when I read "limiter" rather than "compressor").

Most designs suffer from that in some way.

For the SWR when the sensitivity is low it raises the threshold but at the same time reduces the AC voltage at the JFET.   When the sensitivity is high it lowers the threshold and increases the AC voltage at the JFET.   *However*, the rectifier is sensing the voltage at the JFET so at any Sensitivity setting it tries to prevent (by design) the AC voltage at the JFET getting too high.    During transients and before the attack time is met it can reach higher voltages.   This same process/issue is present on the Squeezer/1176 designs.

The main trick of the SWR is how the control signal on the gate feeds the control current down the low impedance path via the 100 ohm.   That reduces the control feed through.  (The way the user threshold is variable using a fixed threshold across the JFET is also a good feature.)

To be honest for gain control based compressor/limiters I don't really distinguish between compressor and limiter.   A limiter tends to have a fast attack but you can set a compressor for fast attack!   In theory you can set the output limit of a limiter and it doesn't exceed that level but that means infinite compression.   In most cases we want the compression curve to have a knee and most circuits have a knee hence there's a fine line between compressor and limiter.

[Rob Strand]

FYI, here's the Eden.  You can see a divider in the feedback loop to lower the maximum JFET drain-source voltage.

[PRR]

> a snippet from the 1176, which looks nothing like the ones I found

Essentially the same. Obviously they are not posting correctly.


The opamps in front and back are mostly interface. The variable loss is the tens-K resistor and the JFET. For fine work we mirror half the signal to the Gate so it tracks the average channel potential; guitar may even prefer NOT doing that (let the grit rise on peaks).

[Eb7+9]

I have now looked at a better looking schematic and see that it works by hampering the amplification of a transistor. The nice thing here  is that there is "no" (or at least not much) AC across the FET, so no distortion from that side. Smart. However, I am currently unable to see, if this design allows attenuation of the input signal below its original amplitude (which is what I want) or merely a reduction of amplification

re; 1176 attenuator FET

you're actually quite close ... SPICE it and you'll see the "no AC across the FET" is actually a variation in Base-Emitter voltage, a logarithmic voltage ... the FET, via signal caps, is providing an AC current shunt - effectively robbing base current, and effectively killing (npn) transistor action ... the FET is acting as a variable AC current divider, across a logarithmic voltage source

how's that for fun ?!

[Eb7+9]

You might be assuming too much from some of us.

a gift then ... my unbaked cake du jour

===

warning: THE CIRCUIT IS 100% UN-TESTED
(schematic provides mere suggestions, YMMV)



you'll need a rail-rail op-amp for the back bias source ... pref. low-noise
and a Vgs(off)-matched pair of jFETs if you want the (optional) LED add-on circuit to work as intended
I use the same LED technique in my 1176 clones ... 

wish I had the time to check this back bias business out myself ... just too busy with another cake
good luck //