Limiters with FETs

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
carlmart
Posts: 169
Joined: Mon May 19, 2014 6:11 am

Limiters with FETs

Post by carlmart »

I have found three different versions for limiters using FETs.

1) The Urei 1176, schematic and all.

2) The Rodd Elliott FET limiter

3) Marshall Leach's limiter.

I'm particularly curious about the last two, because they use opamps.

Has anyone tried any of these or built one?
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Limiters with FETs

Post by mediatechnology »

No, I haven't though of the last two Rod Elliot's circuit is something you could build and run with.
Rod's designs always seem bullet-proof, well-tested and documented.

If it were me I'd stick with a VCA in a predictable feed-forward configuration.

For a limiter FETs are going to be THD/level-challenged and Vactrols slow, power-hungry (from LED current) and expensive.

In addition to limiting you still might want to consider a diode-based clipper that would gently round the waveform in an early stage below the internal clip point.
It's smoother than a hard clip and, unlike a limiter, has zero recovery time.
Sort of like JRs diode-limited bass booster in another thread.
carlmart
Posts: 169
Joined: Mon May 19, 2014 6:11 am

Re: Limiters with FETs

Post by carlmart »

The ultimate proof is listening to all this stuff.

For instance, I never tried a diode-based clipper or know how to design one. Being gentle with the waveform and having zero recovery time sounds very promising.

Where do you place it? Before the limiter?
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Limiters with FETs

Post by mediatechnology »

For instance, I never tried a diode-based clipper or know how to design one. Being gentle with the waveform and having zero recovery time sounds very promising.

Where do you place it? Before the limiter?
The best place to put it, if possible, is in the mic preamp stage.
If not possible to put it in the preamp stage then immediately after it.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Limiters with FETs

Post by JR. »

A diode clipper is NOT gentle on the waveform... unless you think a fuzz tone guitar effect is gentle.

There are far more than three versions of FET (variable impedance) shunt limiters. (I've done more than three myself over the years).

Probably the lowest distortion I've seen was one that Cordell used in the AGC loop for a low noise/distortion sine wave oscillator, in that same article of Audio magazine when he published his distortion analyzer.

While low distortion is nice, unless the side chain controlling the gain is very smooth, there will be larger and more audible gain modulation effects.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Limiters with FETs

Post by mediatechnology »

Are we talking instantaneous overload protection or audiophile-quality limiting during capture committed to flash or disc?

A diode clipper can be made to soft-clip as it approaches the clipping point and it will be less audible than the op amp, A/D or more often than not the limiter doing it for you.

The example given was recording during pyro.
You don't need to actually record the pyro but you need to record the lower-level dialog that happened before it and just after.
You can't have an AGC or limiter taking seconds to recover.
I know of one manufacturer of cinema recording equipment who used diodes in the mic preamp feedback loop for that very reason.

It's pretty much a given that the "event" creating the overload is going to create distortion - it's the artifacts afterwards if there is a significant recovery time which hose the recording.
Its protection limiting.

I don't do this for a living but I was told this by the toolmaker who makes stuff for the people who do.
It makes sense.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Limiters with FETs

Post by JR. »

mediatechnology wrote:Are we talking instantaneous overload protection or audiophile-quality limiting during capture committed to flash or disc?

A diode clipper can be made to soft-clip as it approaches the clipping point and it will be less audible than the op amp, A/D or more often than not the limiter doing it for you.
I try not to argue about what is audible.

Older technology used to be far more sensitive to overload. An op amp overloaded could generate harsh artifacts and take significant time to recover from losing NF, so a diode clamp would limit the distortion to just the duration of the overload, Early A/D were notorious for wrapping around past full scale making very nasty noise, another argument for diode clippers, and finally broadcast pioneered much of the work in clippers.

dbx designed a soft limiter (not clipper) used in some stand alone EQs that tried to use the limiter to extending the digital dynamic range. Midas also promotes a soft limiter in their digital mixer mic preamps to buy a little extra digital dynamic range and encourage their customers to hit the limiters hard.
The example given was recording during pyro.
You don't need to actually record the pyro but you need to record the lower-level dialog that happened before it and just after.
You can't have an AGC or limiter taking seconds to recover.
I know of one manufacturer of cinema recording equipment who used diodes in the mic preamp feedback loop for that very reason.

It's pretty much a given that the "event" creating the overload is going to create distortion - it's the artifacts afterwards if there is a significant recovery time which hose the recording.
Its protection limiting.

I don't do this for a living but I was told this by the toolmaker who makes stuff for the people who do.
It makes sense.
Yes, losing NF from overload can drive DC into a NF (cap to ground) causing a significant settling time after the event to re-mnormalize the DC operating point.

Clip limiters are a mature technology in power amps for sound reinforcement.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Limiters with FETs

Post by mediatechnology »

I'm quoting Carlos from another thread.
That's not what I want, I want a tool designed for an eventual overload not ruining the recording. After it happened things get to normal quickly, as if nothing happened.
I think he's looking for a fast-recovery clip limiter.
Even if the op amps, A/D etc clip gracefully and there's no DC events to recover from, a conventional slow-to-medium release time peak limiter will punch a huge hole recovering from a loud transient.
And, at the onset due to attack time, the signal will probably clip anyway.

The location recordist might want something like this with a higher threshold and a lower ratio.
Below threshold operation is linear.
Pretend the tone burst is an explosion.
The input is on top, output on bottom.

Image

Zero attack, zero release.
No overshoot, no recovery.
No diodes, no FETs, no Vactrols.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Limiters with FETs

Post by JR. »

The important question is how does it sound? I am not smart enough to predict from just looking at the scope trace.

FWIW I have built up a bunch of stuff that "looked" like it should sound good and it didn't. :oops: The steeper region around zero crossings looks a lot like slew rate limiting, so "might" sound like that... probably better than hard clipping, but generally we can't really hear hard clipping if it is brief enough, and recovers cleanly. We really hear it when clipping steps on the envelope generally containing more LF content, and that gets folded up into the midband where our hearing is most sensitive. Another way to look at this is how much overload are we talking about? Ignoring dropped mic's and terrorist explosions, the first several dB of overload is generally only squaring off narrow transient peaks and generally not even audible to most. It is only after the first several dB of overdrive past clipping that we involve the envelope and LF content. That is clearly audible. I have seen pretty successful clip limiters that ignore the first few mSec of clipping (No not by Peavey). Using the slightly slower attack simplifies a lot of the design and some customers think it sounds better, certainly louder, a good thing for power amp customers.

I've told this story before but I've done single blind listening tests where the meat in the seats preferred clipped amps (on bass) over cleanly clip limited amps. The clipped amp was louder which was impossible to control for and nobody ever complained that an amp was too loud. :lol:

The bottom line is to build it and listen to it. As I mentioned power amp clip limiters are mature, and generally use both fast attack and fast release so they don't punch holes in the sound after transient peaks..If you can hear it working the design is not optimized. That's why they put adjustments on dynamics processors.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Limiters with FETs

Post by mediatechnology »

John -

That device is a guitar effect and as you might expect sounds a little - but not a lot - like fuzz.
That particular photo is pedal to the metal with the threshold beginning at 0 cross and the ratio near-infinite.
What appears to be slew-rate limiting is very, very high below threshold gain combined with above threshold gain reduction.
The setting shown in fact provides near-infinite sustain.

How does it sound?
It sounds far better than a hard-clipped waveform with square corners.
The demo units in beta are nearly impossible to get back because people don't want to give them up and are touring with them.
The typical comment is "I can hear it riding on top of my guitar but it's not altering the guitar under it. I'm not stuck using power chords. I can actually play Jazz chords without a wall of grunge."
My favorite one is "I use it on everything all the time to varying degrees when I play live." "I can get a pushed "opened-up" tube-like sound from a small amp in a club environment without hurting people on the front row."

But that's not what I proposed here:
The location recordist might want something like this with a higher threshold and a lower ratio.
With a threshold adjustment not available on the unit in the photo, a softer knee and some additional magic one can make a nice impossible-to-clip gizmo that not only prevents overload but adds, when wanted, a touch of color.
In Carlos' case what may be needed is overload prevention without the flat-top of clipping and with no recovery.
Colorization is not the goal here.

I'm not willing to share details of this project yet.
And though I have photos and sound files of it in action I'm not willing to share those either at this time.
But I mentioned it because it looks useful to prevent hard-overload.

Thus, some imagination and fill-in-the-gaps are needed...

When one symmetrically compresses the high-level tips of the peaks "just a little" you get that Class-A amp with the saggy power supply effect ricardo writes about or your current limited wall-wart powering the demo headphone amp.
I've spent a lot of time listening to compressive distortion.
It's a bit like tape or saturated magnetics in the way it sounds.
Lush, slightly warmer, wider and louder for a given peak level.
Just because it looks bad doesn't mean it sounds bad.
The ear is very easily fooled this way.

When you make the waveform bend itself, rather than slam it into a wall, lots of cool things can happen audibly.

One of the things Glen Clark wrote about in describing the Texar (which used Vactrols) was:
The photoconductive cells used for the VCA on the M-101 cards were not the lowest distortion VCA you could find with a meter. But the choice of the photoconductive cell for the VCAs in the Audio Prism was a way of sneaking some sound sweetening into the Prism without stepping on the Aphex patents.
EDIT: After writing this I decided to create a post in Document about the History of Broadcast Audio Processing hosted by Barry Mishkind: viewtopic.php?f=12&t=728

There's a lot of information about the Texar, Dorrough, Orban, CRL, Innovonix, Omnia and others.

One particular quote that might apply in this thread is a gain control strategy:
You see, there is a lot more magic in the Audio Prism than Glen talked about in his article last month. His control voltage side-chain is a work of art: there are four comparator “trip points,” creating five operational “level windows” for
the audio: [1] too low to do anything with, so do not do anything 'cause it is probably noise, [2] high enough to probably be useful program material, but not high enough to be competitive, so bring it up, [3] right where it needs to be, [4] too high, so bring it down, and [5] way too high, grab it quickly and YANK it down – but release it quickly when the momentary overload has passed.
Post Reply