Available compressor circuits?

[merlinb]

now, the idea of a soft knee shouldn't be mistaken with the profile of low ratio ALC curves // them's two different things ...
technically speaking

Technically yes, but I doubt anyone could really hear the difference between them, especially on a single instrument

[fryingpan]

I'm a recording engineer myself. Still, I dont know what you mean with "the limited output would rise at high input level". I think that's what the idea is about blending dry and wet: instead of having a limited output, transients can pop over the threshold. Parallel compression (as I said). I'm not a great fan of parallel compression, I prefer using compressors 100% wet and set them right/choose the right kind, but the good thing with parallel compression is, essentially, that you can put any bad compressor to good use, just overcompress and add in just a smidge over the dry signal, and there you get a nice fat yet dynamic track (folks do this a lot on drums).

If you blend some clean through it basically tilts the gain curve upward and reduces the CR at higher input levels.

Yes, I know that. I know how parallel compression works and how this is different (and sounds different) to "regular" compression.

Professional limiter/compressors have a true CR control but for simple compressors the shape of the CR profile is largely fixed.  The compression adjustment is more like moving the CR profile left and right (largely variable threshold).

The way I see it (it applies to the Flatline, to the ET, to basically all optical compressors with the gain control element in the opamp's feedback) the "ratio" control generally controls the gain boost the opamp applies when the circuit is not compressing.

Also, the ultimate compression ratio for higher levels is fairly high and is pre-determined by the simple circuit.  By sliding the CR profile left/right you kind of tune the effective CR since only the high level signal makes it into the high CR zone.

You really control the threshold. Actually, the threshold is fixed so by applying the gain you are determining how much of the signal goes over the threshold. But essentially, that's what you're doing. Also, depending on the compressor's knee and response over the threshold, you may be increasing the CR (or not: sometimes the compressors stops compressing for sufficiently high peaks, which means that the transient gets a bit lowered in volume but is otherwise unchanged in its envelope, which makes for a very "dynamic" compressor).

With these types of compressors a clean blend lets you get more control of the CR as the high level signals are no longer subject to the inherent high CR.

Which is why the parallel compression trick is generally used with extreme amounts of compression...

One advantage of using clean blend to reduce the CR is under high levels the clean path stays clean so even if the compression path clips under high inputs some clean always gets through.   Studio compressors use high supply rails so it's not such a big deal, although you can overload OTAs, but small 9V pedals have limited headroom.

...and distortion. Lots of studio compressors really distort quite a bit. Not much in the grand scheme of things, certainly less than a pedal, but still they make things "fat", "warm", "smooth" etc. which is the whole point of parallel compression: take the warmth, fatness, etc. of a pushed compressor for the body of the sound, and use the clean signal for the transients.

The Engineer's Thumb, according to its creator, like all other (OTA?) compressors is really a limiter and there is no real way to reduce the ratio. I'm not familiar with OTAs (they are a somewhat specialist part, they're noisy, they have very little range due to the fact that they distort following an arctan curve for input differences over 50mV or so).

The Engineer's Thumb actually has an implicit blend due to the non-inverting gain configuration.  No matter how much feedback is applied the input signal must leak through with a gain of unity.

As PRR mentioned the flip trick on the Engineers thumb can help reduce the perceived noise.   Modern OTAs are pretty much at the limit of semiconductor physics for noise.

FYI: this is a good paper on compressors

https://ajoliveira.com/ajoliveira/gen/pdf/preprints/paris88.pdf

It was also published in the main AES journal, in a neater format.

I take issue with the characterisation that "there is an inherent clean blend with a non-inverting gain configuration". Yes, all opamps thus configured (in a distortion, in a compressor, whatever) become sort of "buffers" when the non-linearity kicks in (a diode, an OTA, whatever), but really, the fact that the gain function shows a 1 on the right does not mean that you can think of this as blending the clean signal with the effect. Or maybe you can, but looking at the signal overall the fact that the gain goes from, say, 25 to 1 (think of a Tubescreamer, or a compressor set with the "ratio" knob so that the opamp gain is 25) makes little difference to hard clipping. In fact, if when compression is off the gain is 25 and when compression is on the gain is approx. 1, that is the definition of a 25:1 ratio. (Although in feedback compressors you practically never get to such a high ratio).

[Eb7+9]

I take issue with the characterisation that "there is an inherent clean blend with a non-inverting gain configuration" ...

um, yeah ... wt?

indeed, I don't think this forum is the place to discuss AGC circuit design ... or even electronics in general

what you mainly get here as far as expertise is concerned are technicians quoting papers they don't fully understand and then
waving off-handedly about circuits they haven't breadboarded ... let alone played

sounds familiar ??

---

last summer someone asked me to check into THUMBS a bit more

the first thing I did was build one as a means of hunting down the crackle some builders were experiencing
I did that ... even my full-circuit SPICE sims showed it, and it's easy to see how it can happen

but ...

there are fundamental (structural) issues with this thing, including how its creator is interpreting what his thing actually "is"
which bears a heavy influence on the overall implementation of his design effort

first off, let's be clear here ... THUMBS is being treated by its author as, and compared to, an audio-Limiter
which it ain't ... wrong species altogether

the point here, there's a bunch of controls that don't belong on this type of AGC loop and one in this particular case direly missing
indeed, we've seen THUMBS with eight knobs on them - when in reality there should only be two on a proper/correct version of this kind of AGC loop
(see the datasheet circuit examples for the 57x, THAT Corp chip sets ... etc etc)

that's telling in itself ...

and then, the fact that there is a variable time constant at play in this design shows the author believing his circuit to be a Limiter
when, again, it ain't.

Walter Jung was very clear in his writings on this kind of AGC loop about the "local averaging" role of the charge capacitor following the rectifier stage ...
this is totally different than the fast-on/slow-off time constants seen in practical Limiter designs, where it becomes reasonable (and useful) to make the discharge rate adjustable

I could go on as there is more that needs addressing here ...

---

at this point you might be wondering, how do we tell what kind of AGC animal we are dealing with here ...
well, simply, the steady-state transfer curves are ultimately what tells us this ...

the so-called AGC pudding

---

so, I spent a few weeks last summer trying to "fix" these mistakes until I had to drop it by realizing THUMBS' design path had headed
in a un-correctable direction and wasn't practically rescue-able

that in itself would possibly make an interesting read, since studying pathological designs can be as educational,
even direction provoking, as studying flawless designs ... it's in answering the why's and how's that a design is fully understood

---

in the process of doing all this I returned to where I had first started, with the SE570 compander chip set ...

I eventually figured out what I consider to be other "practical" flaws in the implementation of that IC design as well
(notice the vehement refusal by wizz man to accept an obvious structural link between THUMBS and the 570 ALC loop, as I pointed out several times already)

In the 570's case certain design choices were made to help limit the number of off-chip components and also
the number of access points (which translates into number of pins on the package and therefore circuit density)

==> specifically, these include the internal 1.8v referenced hard wired to the op-amp non-inverting input,
the DC-shifting 30k to ground on the inverting input, and the current limiting resistors going into the rectifier and variable-gain blocks ...

of course, this is completely understandable and reasonable from a (then) marketing point of view

but, ...

I'm saying this in terms of what we would want this "loop" to do in a general sense as musicians using low-output sources
the pro version of this idea (eg., Dbx one-knob squeezers) were typically only ever used on Bass outside of std studio use ... not guitar
there's a very good reason for this

the original chipset, even including the 572, all contain these design choices that preclude full generalized use

I've fixed that now by building a discrete/modified equivalent of sorts, and have been playing it ...
it took at least four revs for me to get there ... usually a new design takes at most two for me

bottom line, it's very easy to assume a basic principle will work here when in fact some secondary artifact lies in waiting to bite us ...
at least at the end of it I now know why certain design constraints need to be respected and why a certain basic approach was taken in the 570
all I had to do here was de-couple the basic functional blocks and also remove some of the "limiting" components

to give an idea, it took me about a year to reverse the original 570 design using SPICE ...
and then maybe another six months to fully suss out a generalized alternative

the idea now works as theoretically predicted, both on low output instruments to super hot ...
that is, from weak Tele pickups to honking Bass output

again, it's something the 570/571/572 compander chipsets where not designed to, and could not, do ... as is
in the process, these liberating changes extend the operating range of the 57x systems when configured as compressor/ALC loops

for those interested, go have a look at my "NOT-A-LIMITER" and hear both compressor/ALC modes in action
you tell me if it behaves like THUMBS in character or not ... though, much of it has to do with setting up the loop properly against the source
which THUMBS cannot accomplish by design (oops, I'm letting the cat out now ...)

https://viva-analog.com/synopsis-status-artis-jcmaillet-2022/

I think this is where my 570 re-design needs to be, as the transfer graphs also suggest ...

also - with only two knobs, it's the proper way for this kind of architecture  // just like the big boys have it
(SE-570/571/572, Dbx, That-Corp, SSM ... etc)

this is not a matter of anybody's personal choice but rather the product of the inherent natural response of the control systems they embody
again, THUMBS ain't no limiter - and neither is the classic circuit it's being compared to in the graph above

probably not what many people would like to hear

---

now, one big question remains for the would-be user ...

does one really want to play with this kind of loop or not ??!

first, I would recommend that the player gain experience with proven Limiter designs ...
and understand why these are often referred to as comp-Limiters - there's a useful story there btw

LA-2A and 1176 are two important and proven examples of this, there are a several others

at the end of the day full-servo based AGC loops have entirely different behavior to partial-servo Limiters
and may not really be what many or most players are looking for ...

of course, there's no point going "there" before a proper setup of the loop is performed
(imo that's the main thing missing with THUMBS and many one-knob squeezers ...)

I think it's a good idea to experience all four types and then realize that their responses are different form one another
to be clear, that's four types of AGC loops right there // not including expanders, gates and duckers ...

---

I agree with the notion that providing a parallel dry path (Barber's idea ?!) was meant to act as a fix to a terrible Limiter design
personally I think it's a bit of a cop-out although it can obviously be of practical use

and besides, who's got the design chops do design a properly functioning side-chain ... ?!
(as even MXR got this wrong when they tried to fix it ...)

of course, the parallel path idea is a bit of a no-brainer in this sense

but more importantly, in terms of the final performance "feel" is that really such a great approach to take ??!

As a circuit designer I find it useful to first identify what the flaw might consist of ...
and then seeing if we can fix the flaw with careful re-design

In the end, once the famously flawed design was "fixed" I discovered a way more enjoyable response than the cop-out approach,
maybe that's just my mileage ...

---

as a final note, notice for example how the LA-2A and 1176 don't need this fudge ... and we should ask "why not ?!"
there is MUCH to uncover here, I could write a small book about all this (and almost have here today)

be well ...

[PRR]

Thanks for your opinions.

[ElectricDruid]

Instead of writing a small book, why don't you cut to chase and tell us what you think is wrong with Merlin's design?

You've done this before, hinting that there's something fundementally wrong with it, but you never say what. It's starting to get a bit tired. This time you've hinted that it can't be fixed, but again, you won't say why.

All the time you do this, you imply that those of us who can't already see it are simply technical illiterates that aren't worth bothering with. I could probably post some crazy assembly language stuff that you wouldn't understand, but it doesn't make me cleverer than you - we just learned different things is all. Many people here have specialist areas. So rather than hoarding your knowledge, why don't you share it? Otherwise, what exactly are you here for?

[Rob Strand]

I take issue with the characterisation that "there is an inherent clean blend with a non-inverting gain configuration". Yes, all opamps thus configured (in a distortion, in a compressor, whatever) become sort of "buffers" when the non-linearity kicks in (a diode, an OTA, whatever), but really, the fact that the gain function shows a 1 on the right does not mean that you can think of this as blending the clean signal with the effect. Or maybe you can, but looking at the signal overall the fact that the gain goes from, say, 25 to 1 (think of a Tubescreamer, or a compressor set with the "ratio" knob so that the opamp gain is 25) makes little difference to hard clipping. In fact, if when compression is off the gain is 25 and when compression is on the gain is approx. 1, that is the definition of a 25:1 ratio. (Although in feedback compressors you practically never get to such a high ratio).

If you think about a unity gain limiter (infinite CR) with 1V output, then blending a clean signal at a level of 1/10 (-20dB) the input: when the input signal level reaches 10V the clean and limited signal are the same level.   That reduces the CR from infinity:1 to about 2:1 at 10V.   In between 1V and 10V the CR tapers off between the two values.   The point I'm making is for high CR limiting/compression you don't need to add much clean to affect the CR.  The fact it does do that is why you can use blending to knock back 10:1 compression to 3:1 with some delicate adjustments.

(FWIW, the definition of compression *ratio* is at the start of that Oliveira article I posted.    When you consider different input/output gains, soft knees vs hard knees, you need a definition that handles all the cases.   Your CR definition isn't correct.  CR is defined on a dB scale.  Your definition is the *amount of compression" or "amount of reduction", which is usually stated in dB, but you can state it on a linear scale.)