THAT (was Fet) Compressors

[QUEEF BAG]

for a fet comp i thought the distressor brought a few nice goodies
to the table.  a few cool features.

sonically very good for a fet, but not the beautiful fem-FETal that
other designs can have. 

this thread made me pop the top off one of the el-8 units here...
shit. a couple attempts with a swab would reveal no numbers on the
chips.  oh well.

 

[mad.ax]

FETures request anyone?

Axel

 

[QUEEF BAG]

magic smoke and a current mirror in the detector.

serious about the mirror.

good smoke is fine too.

 

[SSLtech]

I'm getting FET-up waiting...

;D

 

[QUEEF BAG]

i don't mind the wait, i need the time to
get to my tailor and get FETted for a new
pair of huggies.
i might need them when this thing makes an appearance.

 

[skipwave]

Been posted here before, but germane to discussion, which may now proceed unfettered.


http://users.ece.gatech.edu/mleach/papers/limiter.pdf

 

[JohnRoberts]

Been posted here before, but germane to discussion, which may now proceed unfettered.


http://users.ece.gatech.edu/mleach/papers/limiter.pdf

Yup, I'd forgotten about that, another example of linearizing the FET channel resistance by adding in some of the drain signal. Kind of a conflicted design from a commercial perspective since FETs are usually used for simple low cost limiters and that uses a bunch of opamps, but it clearly demonstrates the relationships.

Marshall (RIP) was an educator not a working design engineer.  He is missed.

JR 

 

[SSLtech]

Thanks for FETching that for us, Skip.

We should archive that, in case it's deleted from the GA Tech server, or some other terrible FET befalls it.

 

[volker]

So, did this ever leave the FETal stage and see the light of day?

 

[shabtek]

this could be of benefet to all

 

[Kingston]

give the guy a break, it takes a liFETime to master these things.

 

[Andy Peters]

Been posted here before, but germane to discussion, which may now proceed unfettered.

http://users.ece.gatech.edu/mleach/papers/limiter.pdf

That looks highly-similar to a limiter called the Pandora that a friend in Minneapolis happened up in a pawn shop and bought for not a lot of money. It sounds pretty interesting. Not something you'd use on every input or the stereo mix, but for some things it's cool.

The Pandora uses a 741 as its gain element, which I guess dates the box.

=a

 

[JohnRoberts]

Dead thread walking....  ;D

JR

 

[volker]

Nah, it's not dead, it just smells funny. Call it fetid if you feel like doing so.

 

[tv]

 

[okgb]

give the guy a break, it takes a liFETime to master these things.

truly clever , where is the O.P ?  and what is the thing that squeezes my bank account ?

 

[andreapetucco]

Does anyone  of you tried to built the Marshall Leach limiter?
I suppose it deserve a try, the tesign is very smart and well thought, using a proper op Amp it could sound very good.
https://leachlegacy.ece.gatech.edu/papers/limiter.pdf

 

[JohnRoberts]

Does anyone  of you tried to built the Marshall Leach limiter?
I suppose it deserve a try, the tesign is very smart and well thought, using a proper op Amp it could sound very good.
https://leachlegacy.ece.gatech.edu/papers/limiter.pdf

This is a fairly old topic (so a forum search should reveal more discussion) but for the newcomers, dynamic processor "sound" is dominated by the side chain (attack, release, hold, smoothing, etc.) that modulate the gain. The gain element used only contributes to the sound quality impression "when" it is bad enough to generate  audible distortion or artifacts.

FETs used as variable gain elements have been widely used because they are cheap, but generally not in premium paths that can justify more expensive gain elements. I've used them in oscillator AGC circuits, and even inexpensive noise gates, never as a compressor/limiter and I'm cheap.  8)

The Leach (RIP) design is instructive because it teaches the linearization technique, but is odd to see that many op amps, and current mirrors, and whatever used in a typically cheap gain control path...  Many a bass guitar compressor has been done with a single op amp and maybe a diode and resistor for the complete side chain.  :-X

JR

PS: FWIW the benefit of the linearization technique is indeed audible. I used JFET in an inexpensive noise gate while at Peavey and distortion caused by an uncorrected JFET was audible for some combinations of threshold and release (slow).  With the simple linearization circuit the distortion was not very apparent for gating off complex waveforms.

 

[PRR]

> The Leach ... odd to see that many op amps, and current mirrors

Everything has a context.

This Leach paper mentions "broadcast". So not live sound, not 30ips tape or 24-bit. Ultimate performance does not have to be impeccable. But it should be predictable and reliable.

"Many opamps" is relative. Two opamps for a full-wave rectifier is about par. (Half-wave would misbehave on some speech/music waveforms.) One to buff-up CV and one to buff audio does not seem like overkill.

This basic-limiter paper should be read as a sub-part of the Big Design:
A Four-Stage FM Broadcast Audio Peak Limiter.

This includes an exhaustive derivation of FM preemphasis as pertains to pre-limiting before the final preemphasis in the transmitter (exciter).

Reading between lines: Georgia Tech had a student-run FM radio station. Student radio does not over-indulge in "loudness wars". But radio always needs a limiter (overmodulation is too disruptive, and not-much undermodulation is disappointing). It is entirely reasonable that a tech-school student-staff would DIY the station limiter.

And turn to Dr Leach. Who would plan the lesson.

In that milieu, it would be favorable to use ONLY parts deeply stocked in the EE Lab. Dual/quad opamps, BJTs, diodes, caps, and FETs. Vacuum tubes, no way. LDRs, perhaps not. dBx/THAT chips, not stocked and the best use is to copy the Application Notes, nothing to learn there.

In all it uses _26_ opamps, twice for stereo, 52! So the full FM limiter may be a major dent in the EE Lab inventory. But the base limiter is just four opamps, two TL072 or one TL074, you can have every student build this part and report on it.

Note that if you build the base 4-opamp limiter, then add the input/output sections from the larger paper to get "balanced" interfaces, you double the opamp count. (He uses the invert/invert diff-input, not the 1-opamp diff in.)

The UA 1176 is an FET limiter, well regarded, and runs the equivalent of about four opamps (plus a meter driver, not done in Leach's). And transformers to get balanced interfacing.
https://media.uaudio.com/assetlibrary/1/1/1176ln_manual.pdf

 

[PRR]

Leach references a 1968 paper from RCA. This is a fine summary of audio compression theory (except he dismisses feedforward compression too quick), and advocates MOS-FET as the vari-gain element.

http://www.americanradiohistory.com/Archive-RCA-Engineer/1966-02-03.pdf
page 63 (p64 of PDF)