Full Wave Rectified Orange Squeezer, anyone?

[MrSimple]

I mean I'm not offering one, I'm asking whether anyone ever actually built one and how did it sound compared with the standard half-wave style?

I'm just irritated by half-wave rectification anyway, just on a conscientious engineering basis, audio signals being so asymmetrical as they are and especially that the control voltage is going to modulate the signal so why not make it as clean as possible?

There is at least one schematic out there and I've sketched out a couple myself.  Has anyone actually ever built one?

Simp

[Mark Hammer]

I suspect the audible difference between half-wave and full-wave rectified signal will depend on:

a) the nature of the control element - photocells will tend to smudge any differences, whereas OTAs and JFETs can reveal them more readily;

b) what part of the note lifespan we're talking about - the more obvious ripple of the envelope shows up most during the decay, rather than the initial attack;

c) what sort of recovery time one is using - fast and very long recoveries will obscure differences, medium recoveries will tend to reveal them.

So, full vs half probably won't matter for a photocell-based compressor set for fast recovery time.  The stock OS uses a JFET as the control element, which puts it at a bit of a disadvantage for "revealing" ripple.  But it also uses a fairly fast recovery, which offsets that.

Of course, if one implemented variable recovery-time so that the OS was nudged to behave a bit more like a compressor and a bit less like a limiter, I can see where full-wave rectification might come in handy there.

[midwayfair]

I breadboarded this idea not too long ago. I used the full-wave recitifier from one of Ray Ring's (circuitsalad.com) compressors. You just replace the single diode with the whole rectifier circuit (this was Ray's suggestion).

It sounds fine, but I wouldn't say it's exactly an improvement except from an engineering standpoint unless you want to also play with the decay and threshold (in which case you get a lot more control). I haven't heard an Orange Squeezer ripple when biased correctly. In return, you end up with a half dozen more parts. I'm just not sure it's worth it. I'm not the most disciminating guy when it comes to soldering something up, and I didn't bother with this, so that's something. I also don't see the point of making a super-transparent comp like the OS a parallel compressor like JHS did, but they get a lot of money for them, so what do I know?

[R O Tiree]

Ripple on an Orange Squeezer is minimal, because there's a 4µ7F cap to "store" the output from the diode and the charge dribbles away to GND through a 100k resistor. That works out at a time-constant of 3 seconds. The 1 or 2mV "top-up" on every high cycle is un-noticeable, as far as I can determine, both from sim (LTSpice) and practical, real-world observations. As Jon says, if it's biased correctly, it's clean as a whistle.

The stock OS works well as a compressor as long as the max output from your pickups is more or less the same as the "turnover voltage" of the FETs involved.  By this I mean the amplitude at which Vin = Vout. If you have some fairly feeble single coils that output a signal significantly less than this voltage, then you're already well into the "amplification" side of the curve, and it's TOO DAMN LOUD! So you turn it down. Given that its max gain (after losses) is of the order of 15 or so, and your single coil signal is already getting amplified by a factor of 4 or 5, then you're going to turn the output pot down to about 20-25%.  The max compression that you might see, therefore, is a gain of about 3 or 4. Conversely, someone with insanely hot humbuckers will be solidly on the "attenuation" side, so it will sound very much like a limiter (it isn't, actually - output signal still goes up as input rises, but it is perceived as limiting, and it's the ears that count). The one I built has a "turnover" of 1.1V. My single coil Tele sounds... OK, but the compression is very subtle (for the reason stated above). My Maverick (HSH) sounds really good, with lots of sustain. My lead guitarist's PRS sounds... pleh...

As I've said before, simple circuits like these can never be all things to all men.

[MrSimple]

"I haven't heard an Orange Squeezer ripple when biased correctly."

"As Jon says, if it's biased correctly, it's clean as a whistle."

I've been shopping Orange Squeezer posts for a while now, and I don't think that I've ever seen a specific procedure for setting the bias.  Would anyone venture to offer one?

S

[R O Tiree]

The simplest way is to start with the trimmer pot somewhere near GND and work your way upwards until it sounds good... a bit hit-and-miss perhaps.

One more accurate way (it's the one I worked out - there are probably other ways to do this) is to start with RG's FET matcher circuit. Instead of an R_set of 10k, use 1M. That matcher circuit is designed to give a reading for V_GS where R_DS = 10k. We need to find V_GSoff, and 1M will get us within a few µV (and you can only read your DMM to within a mV or so, so it's all good). Let's say that came out at -2.813V... The FET that is acting as the current source can be popped into the circuit without any testing. The one you tested goes into the other slot. Power the circuit up with no input signal at all (maybe plug your guitar in and turn the Vol pot to 0). Adjust the trimpot so the Drain of the current-source FET is sitting at plus 2.813V with respect to GND. As this is connected directly to the Source of the FET that controls the amount of leakage to GND, and also given that the Drain of that FET will sit at 0V with 0V input signal, this FET is now sitting exactly at its cut-off point. Tweak slightly to taste.

Hope this helps.