A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
JR.
Posts: 3732
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by JR. »

Getting a little esoteric is there any risk of making compression too clean (i.e. having no sound signature)? Back in the 80s when I was tweaking companding NR my goal was to be as transparent as possible. Modern compression customers may have preconceived notions about what compression "should" sound like. I have seen people in other areas of sound experience issues with "too clean" results...

a) with Peavey's power amp clip limiting some customers preferred the clipped amplifier sound (albeit louder due to clipping amplifier making more average output power).

b) Another market segment was Danley's low distortion bass speakers. Customers are used to a certain amount of bass loudspeaker distortion to gauge how loud a sound is.
====
I vaguely recall some discussion of making a variable capacitor to interface with THAT corp detector here... viewtopic.php?p=7560&hilit=variable+capacitor#p7560

including this hypothetical circuit scribble
Image

As I recall I never melted solder for this chicken scratch so may not even work. :?

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5744
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by mediatechnology »

I remember that drawing.

The cap multiplier I'm using is a single op amp inverting stage Av = 0 to -9 connected to the normally-grounded end of the timing cap. At a gain of 0 the timing cap's value is 2.2µF making the total range 10:1 to synthesize a 22 µF.

I just finished listening to a walking bass part with and without QPS and "the without" is noticeably brighter due to the added third harmonic.
So yes there's a color added by the ripple which, whether good or bad, is in the ear of the beholder.
Without QPS there's no option to switch off the added LF color.
At higher frequencies the ripple is much lower and there's no real increase in brightness.

I tried a 1 µF timing cap and it's just too small and anything greater than a synthesized 22 µF too long.
So the 600 dB/s to 60 dB/s range seems about right.
I found that a super-duper-long dB/s rate even with gentle compression jacked with reverb tails too much.

The beauty of the cap multiplier is that I can use a film cap for Ct for low leakage and low dielectric absorption.
User avatar
mediatechnology
Posts: 5744
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by mediatechnology »

A comparison of conventional "True Power Summing" to "Quadrature Power Summing" with a variable timing capacitor.

The test is 10 cycles of 50 Hz with about 10 dB of compression.
The first sequence is conventional True Power Summing (TPS) with the decay rate swept from 60 dB/s to 600 dB/sec and back again.
The second sequence is Quadrature Power Summing (QPS).

The sound file is here: https://proaudiodesignforum.com/content ... vs_QPS.wav

In an editor it looks like this:

Image

You can see in the first segment, which is TPS, that the rate affects overshoot at the shortest settings. As we will soon see this is not necessarily a good thing.

(The overshoot is on the positive half cycle because the burst starts on the positive half-cycle.)

QPS in the second segment has slightly more overshoot - more on that later - but it's overshoot is unaffected by the rate control.

Zooming in to the TPS overshoot we see this:

Image

The first cycle attack is distorted due to super-fast attack time.

Zooming in to the QPS overshoot we see greater waveform fidelity:

Image

The increase in waveform fidelity on the first half-cycle isn't solely due to QPS - it's due to delay in the sidechain allpass filter delaying the first cycle relative to the input.
This causes a bit more overshoot but with less distortion on the attack and succeeding waves.

An FFT of the last 9 cycles (without the first cycle overshoot) of TPS shows that it has a significant (-24dB) amount of HD-3:

Image

The FFT of the last 9 cycles of QPS shows no detectable HD-3:

Image
User avatar
JR.
Posts: 3732
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by JR. »

Back when I was seriously designing dynamics processing circuits a homemade tool that I found useful was a variable tone burst circuit. Sorry if I've shared this before, the drawing was done in 2007 and I designed this back in the 70s/80s.

The bottom section generates adjustable on time and off time gates feeding a simple JFET mute. The cmos FF insures that the burst output turns on and off at zero crossings, and always presents full cycles of sine waves.

The pan pot pans between fully dry and fully gated signals. I found this useful when tweaking side chains to replicate more dynamic scenarios.

This was designed for sine wave tone burst generation but it also works with complex music input allowing one to punch up the crest factor to make dynamic music sources even more dynamic. I found this useful for coaxing out weaknesses in dynamics processing circuits.

JR
Attachments
burst_gate.gif
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5744
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by mediatechnology »

I recall you posting that in this thread from bruno2000 (RIP) but the schematic disappeared: https://proaudiodesignforum.com/forum/p ... php?p=3096

My toneburst tests were a "0-60" pedal to the floor test.

I decided to use bruno's test tone to compare it to his findings and built one in an editor. I call it the "bruno burst."

I'll try to run some tests this afternoon.
User avatar
JR.
Posts: 3732
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by JR. »

Over the years, often working with inferior gain elements, at least part of the effort was to mitigate sudden gain change artifacts.

The pan between dry and full gating helped coax out subtle flaws. The problem was flaws heard with complex (music) waveforms that weren't heard with standard test signals.

You are likely dealing with a more robust gain change path, than I was.

JR

PS: Yes I couldn't find the schematic anywhere on the WWW, that I apparently shared years ago. I found it on my home 'puter.
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5744
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by mediatechnology »

Bruno2000 wrote: Mon Jul 06, 2009 9:30 am I did some tests comparing the Attack and Release signatures of some of the compressors that we have that are favorites for voice. Here's the info on these tests: The test tone was generated by an Audio Precision System One, and recorded to Pro Tools at 96kHz through a 192 interface. 0 crossing for the burst was verified via Pro Tools waveform to be correct. The steady state 1kHz tone is at -6dBr (0dBr=1.228V RMS) for 1k5 cycles, then +4dBr for 500 cycles (10dB "burst"). All comp thresholds were set for below compression threshold with the lower level tone, and about 4dB compression (according to their meter) during the burst. Ratio on DBX165 was set at 6:1 (really about 4:1 according to my tests) and the Pico was set at 4:1 (also, according to my tests) for both the RMS and NLC (Auto) settings. The Manley's A/R settings were set for what most of the freelance engineers here use for vocals, and although no one here uses the Manley "Limit" setting for vocals, I ran the test anyway.

Enjoy!
Bruno2000

Editor: In order of appearance, input (blue) dbx 165 (green), Pico RMS (red), Pico Auto-NLC (dark blue), Manley Comp (dark green), Manley Limit (rust)...
Image
Bruno burst at 120 dB/sec looks identical to the Pico and dBx 165:
Image

Bruno burst at 60 dB/sec looks similar to the Manley:
Image

Bruno burst at 600 dB/sec has the attack signature of the Pico with NLC but without the hiccup in the recovery:
Image

Note that the steps in the recovery are due to the editor's quantization of the displayed peak file.

Next up: Some sound files to compare...
User avatar
mediatechnology
Posts: 5744
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by mediatechnology »

One thing I'm going to change is the gain control of the capacitance multiplier.
The pot is too cramped at slow rate settings and most people would want stepped controls anyway.
Switching input resistors on the inverting stage to use a 6 position rotary switch with 600, 480, 360, 240, 120 and 60 dB/second settings seems to be the way to go.
With these settings the voltage gain ranges from "0" at 600 dB/second to Av=9 for 60 dB/sec for an effective capacitance range of 2.2 to 22µF.
User avatar
JR.
Posts: 3732
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by JR. »

mediatechnology wrote: Mon Feb 19, 2024 8:53 am One thing I'm going to change is the gain control of the capacitance multiplier.
The pot is too cramped at slow rate settings and most people would want stepped controls anyway.
Switching input resistors on the inverting stage to use a 6 position rotary switch with 600, 480, 360, 240, 120 and 60 dB/second settings seems to be the way to go.
With these settings the voltage gain ranges from "0" at 600 dB/second to Av=9 for 60 dB/sec for an effective capacitance range of 2.2 to 22µF.
I couldn't find your cap multiplier schematic, if possible use a real cap for the minimum setting (film preferable to avoid DA effects). A real cap could minimize switch clicks from switching ranges.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5744
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Quadrature Power-Summed Compressor Sidechain For "The Uno Compressor"

Post by mediatechnology »

It's in the original schematic at the end of the 2u2 film cap. When Av is 0 it's at AC ground and Ceffective is 2.2. When Av is -9 Ceffective is 2.2 uF + 19.8 uF.

Edited sign of Av. Ct is "anti-bootstrapped."
Post Reply