The Waveulator: Two VCA-based Clipper Saturators

[mediatechnology]

I'm about one op amp away from having a working stereo Waveulator.
I did have a chance to audition a single mono channel and can say that this is a completely different animal than what I had before.
Under large amounts of compressive peak limiting it's way smoother.

I looked at the compression response three different ways:

1) Waveform mode
2) X-Y transfer function
3) K meter

Gone are the hard corners both coming out of zero crossing at low thresholds and in the transfer function at higher thresholds. It simply sounds smoother.
I'm naming the detector the "lazy rectifier."

In X-Y mode I can see smooth S-curve transfer functions with low threshold and high ratio settings.
With a high threshold setting I can draw Z-like shapes. What's different about this Waveulator is that the rotation point at high thresholds is more "over-easy" and rounded than it's predecessors which had harder corners.

The K meter results are interesting and achieve what I originally set out to do with the original Waveulator.
On most complex material with high input crest factor I listened to I could easily reduce peak levels 4-8 dB without the RMS level being affected and introducing harshness.
With peak levels reduced 4-8 dB and the output peak level normalized to the input with makeup gain increases in overall signal power can be obtained.

On material that's been previously brickwalled or highly clipped applying the same settings it gets pretty nasty-sounding fast. It's "double-dip" distortion.

On tone I can clearly hear the change in color but on music it's more of a compressive effect.
The fact that the "limiter" has zero-attack and zero-release, ignoring signal history, causes it to get in and get out without stepping on RMS level.
The Waveulator, operating as a limiter, provides crest factor control.

In addition to the final op amp stage for the second channel I need to wire up proper Threshold and Gain Make-up pots and a bypass switch to control the relay.

I'll eventually look at skewing the negative detector gain to introduce an even-order component.
I can safely say that the ear can tolerate far more odd-order distortion than even-order on complex material where a little bit of even goes a long way.
Even is more of a brightness/darkness control where brightness or darkness depends on the polarity.

I hope to get more work done today.

[mediatechnology]

I had a chance to record a demo featuring Robert Cray's performance of "Strong Persuader."

This Waveulator demo is a protection limiting example being used ahead of the A/D recording it and features three cuts.

The first is the unprocessed sound file which produces about +3 dB of A/D overdrive. The Waveulator is in bypass.

The following two cuts, recorded at the same input level with the Waveulator on provide limiting at -2 dBFS and -5 dBFS.
There is no makeup gain applied.
The ratio is set high for a hard limit.

The -2 dBFS cut doesn't significantly reduce the RMS level; the - 5dB cut reduces it about 0.6 dB.
RMS is the unadulterated portion of the waveform in the middle.

Sound File with approximately +3 dB Overdrive: https://proaudiodesignforum.com/content ... _Overs.wav


Robert Cray Strong Persuader With Overs Statistics

Sound File with limiting at -2 dBFS: https://proaudiodesignforum.com/content ... -2dBFS.wav
Total gain reduction is about 5 dB.


Robert Cray Strong Persuader With Limiting At -2dBFS Statistics


Sound File with limiting at -5 dBFS: https://proaudiodesignforum.com/content ... -5dBFS.wav
Total gain reduction is about 8 dB.


Robert Cray Strong Persuader With Limiting At-5dBFS Statistics

The following is a persistent transfer function display comparing no processing to -5 dBFS limiting.
The rotation point was ±0.5V peak. The traces were captured over the full duration of the song. The total amount of limiting is about 8 dB.

Robert Cray Strong Persuader Transfer Function 8dB Limiting vs No Limiting

[TheeAldeen]

thanks for doing this wayne

[emrr]

Yes!

[georgeonthenet]

That's some impressive limiting. I took a careful look and listen to the -5db file and it's pretty obvious in CEP what's happening to the peaks but you would not know by listening. When I have time it would be fun to compare other limiting techniques using the same track.

Robert Cray's rather clean recordings are a good choice. I saw him live last summer and he was an impressive performer to say the least but... not the live sound. Oh well.

[mediatechnology]

Thanks everyone for your comments.
As I was looking through my playlist I realized the song title is actually "Right Next Door."

THRESHOLD

I took a look yesterday at optimizing the threshold range.
For a mastering unit I think the best control scaling for threshold is to present the rotation point in -dBFS with the 0 dBFS calibration adjustable to the peak overload point of the user's converter.

We think of most absolute level measurements in dBu, dBV or dBm which are based on defined RMS measurements of voltage or, in the case of the latter, power which, by definition, is RMS.

This thing responds to peaks and there is no dBpk.
So when I speak of the threshold adjustment range in dBu units I'm actually referring a corresponding peak level based on an RMS measurement of a sinewave.
So if the measurement is 0 dBu the actual peak threshold rotation point is 1.096V. (0.775 * √2)
Confusing?
Yes, that's why dBFS makes sense.

On the prototype I optimized the threshold rotation to go from about -2 dBu at full CCW to +24 dBu.
I'm not sure what range is needed for the entire universe of A/Ds out there but I frequently see +8 dBu (2V RMS), +14, +18 and +24 values for 0 dBFS.
Do I need to accommodate a +27 dBu overload point?

The 0 dBFS reference could be set by a jumper, a trim pot or an off-board switch.
The threshold control would then span from -26 dBFS to 0 dBFS or, for more precise control, over a limited range of -11 dBFS to 0 dBFS on a 12 or 24 step switch.
In the case of a stepped switch the step size and the number of steps could be determined by the builder.

MAKEUP GAIN

I haven't built this in yet but it's super-simple.

METERING

Looks like the best metering point for gain reduction is the wiper of the Ratio control which is scaled at 30 mV/dB. (600 mV/20dB)
I need to add a meter buffer with a gain of 10.

EVEN HARMONICS

I played around with it and the best place to do it is by skewing the gain of the negative polarity amp driving the lazy rectifier.
This allows asymmetric compression of positive or negative peaks.
I may make this an option with a jumper to an off-board pot.
It's more useful in mixing IMHO for individual tracks.
On complex material it doesn't have the wow-factor I'm hearing with odd-order only and either darkens or adds harshness.
This is something I need to listen to more.

[TheeAldeen]

this is really turning out to be a very neat unit & awesome addition to the ka-module list.

[emrr]

Agreed, glad to see this resurrected. I can’t think of a +27 converter input offhand. So many now have input reference scaling too. Not many things would drive that well anyway.

I’ll add the Serpent SB4001 compressor has a button to increase 2nd harmonics which I’ve resisted using as i track through a lot of SE tube preamps already, but on a few mixes lately it has sounded good. It’s an on/off with no panel adjustment.

Given how many unimaginable ways various folks work, I really can’t predict where people would put this in their mix or master chain, it might surprise you how early it gets placed, in those cases the knob markings will be ignored anyway. Food for thought.

My most immediate usage, unknown cal, between a live console DA output and a webcast camera input for livestreams. Though it’s remote from my location so I’d need a camera on the meter so I can direct remote adjustments based on the loopback I receive through a video monitor!

[mediatechnology]

Saturation for peak level management really needs to be the last thing in the chain prior to the converter. Any EQ or phase shift placed after it "unwraps" the limiting. For color it can go just about anywhere.

For live or location recording it can be used to make an overload-proof converter.

This is really working out to be a really simple circuit.
I need to add a couple of features and move some stages on the Protoboard but before I do I'm going to shoot some 'scope photos.

[emrr]

Saturation for peak level management really needs to be the last thing in the chain prior to the converter. Any EQ or phase shift placed after it "unwraps" the limiting. For color it can go just about anywhere.

Agreed, but the unwashed masses do crazy things. More frequently than not.