jBam

SSL bus Threshold (and no doubt many others)...
« on: April 21, 2017, 03:03:58 AM »
Hi all... Has been a while!

Busy busy life at times.  I'll confess though: I've locked myself out of my house, and have headed to the pub till my partner gets home - and in a moment of "now, how can I make this productive beyond drinking beer", I've decided to pop in here and ask a question Ive been pondering:

Threshold design on compressors.

Some of u may recall that I set about designing a dynamics system from scratch without referring to a current design... A bit of a fun check of "I think this is how I'd do it".  Still haven't built my various gadgets, but: my understanding of threshold seems to differ to standard practice.

My threshold is more of a: rectify signal; Env follow / filter; extract peak off signal via more rectifying (user selected depth, aka threshold).  This means that below the threshold, there is no control signal (ie it is zero volts).

I appreciate that this approach is very much a peak detection / compression approach.

Now that I'm trying to understand RMS control signal rather than my current peak dedigns, and also trying to understand existing designs like the SSL bus, I'm getting stuck: I don't understand how the threshold separates "no compression" from "some compression", because from my basic understanding of the circuits I looks at, all the threshold seems to do is amplify the RMS or add a DC to it or something.  This means that I don't quite understand why the circuit isn't always in compression; and the threshold seems do what I'd assume ratio does (increase the amount of control signal similar to ratio).

What am I missing.  Is it a diode in reverse that gets pushed to reverse flow? After a breakdown voltage is exceeded (that'd make sense)??...

Or am I missing that it does rectify again, extracting a portion of the RMS above certain level (where threshold amplify the signal up and down past this point).

As always, sorry for any misused jargon.

An explanation of any threshold circuit would be great, or a link.

Best,

Jonny

« Last Edit: April 21, 2017, 03:10:31 AM by jBam »


gyraf

Re: SSL bus Threshold (and no doubt many others)...
« Reply #1 on: April 21, 2017, 04:26:07 AM »
The simple way to think of this is:

1) You always have a rectifier circuit that also contains the "real" part of the threshold entity - either something as simple as a diode forward voltage drop, or something as complicated as subtracting some fixed or variable voltage. This threshold sets the voltage you need to get over in order for the rectifier circuit to output any DC at all.

2) Now your Threshold function is simply the gain/attenuation you apply to the signal before rectifier(threshold)

3) And your Ratio function is the gain you apply (to the resulting DC voltage) after rectifier(threshold)

Jakob E.
« Last Edit: April 21, 2017, 04:29:27 AM by gyraf »
..note to self: don't let Harman run your company..

jBam

Re: SSL bus Threshold (and no doubt many others)...
« Reply #2 on: April 21, 2017, 04:56:57 AM »
Oh my Jakob... Thanks so much mate!

You know... I'm really pleased - That makes so much sense!!! No wonder I could make out rectifier type circuits in the threshold section of various schematics (I'm still far from an electrical engineer; just a junior enthusiast!).

That is a really great explanation... i think I actually understand it now, but I'll have a further think and be in touch if I have any mm ore questions... The fun continues!!!

Quick edit:...

So yeah... I'm seeing now (I think), that voltage drops (forward, not even reverse breakdown) across diodes is a big part of many thresholds I'm looking at... (Yeah?)... Very interesting!!
« Last Edit: April 21, 2017, 05:14:11 AM by jBam »

bluebird

Re: SSL bus Threshold (and no doubt many others)...
« Reply #3 on: April 21, 2017, 04:37:29 PM »
So yeah... I'm seeing now (I think), that voltage drops (forward, not even reverse breakdown) across diodes is a big part of many thresholds I'm looking at... (Yeah?)... Very interesting!!

The forward voltage drop across a diode (about 0.7v for silicon) is also the voltage at which the diode is passing max current.  So it can be looked at, as where the control current starts flowing. But you can bias up a diode (idle it) to 0.7v with DC so its always conducting.  Then your incoming AC signal before the rectifier doesn't have to produce the 0.7V DC to  make the sidechain current start flowing. 

jBam

Re: SSL bus Threshold (and no doubt many others)...
« Reply #4 on: April 21, 2017, 06:27:13 PM »
Yes - thanks Bluebird… I'm starting to paint up a new view of threshold + rectifier.  Originally all of my rectification efforts have been based around high precision of rectification separate to any threshold.  This has then been based on combating the voltage drop for accurate rectification (e.g. by notably amplifying the input signal pre-rectiying so that 0.7V present less of the signal), through to negating it entirely to ensure that the voltage drop doesn't effect the integrity of the rectified signal…

What a strange surprise that my new question jumps back in and the answer is: "use the voltage drop" not "damn voltage drop, how do I avoid you"!! haha…  all part of the fun of learning.

Now that I'm getting my head around this, I'm seeing how to incorporate RMS into my current design.  In fact, I'll probably keep my odd threshold design as-is.  It's based on a fairly intense rectifier I designed up using differential signals and CMOS switches to avoid diodes / voltage drops.  Likely more expensive than it  could be, but it models up as VERY accurate, which I need (or maybe just want) for my peak compression and even more-so for my super limit mode (which is actually more of a VCA based soft clipper which models very good results).

I really appreciate your comments and help guys.  Great work, and hope you're all having a great day!

Andy Peters

Re: SSL bus Threshold (and no doubt many others)...
« Reply #5 on: April 21, 2017, 06:41:01 PM »
Now that I'm getting my head around this, I'm seeing how to incorporate RMS into my current design.  In fact, I'll probably keep my odd threshold design as-is.  It's based on a fairly intense rectifier I designed up using differential signals and CMOS switches to avoid diodes / voltage drops.  Likely more expensive than it  could be, but it models up as VERY accurate, which I need (or maybe just want) for my peak compression and even more-so for my super limit mode (which is actually more of a VCA based soft clipper which models very good results).

While you're loathe to peek at prior work, this THAT Compressor Design Note (pdf) shows how the detector output is turned into the VCA control voltage.
"On the Internet, nobody can hear you mix a band"

jBam

Re: SSL bus Threshold (and no doubt many others)... New
« Reply #6 on: April 21, 2017, 06:46:19 PM »
While you're loathe to peek at prior work...

Haha… Thanks Andy…  I'll have a look through - just don't tell anyone;)

EDIT:  Actually - yes… I do recall looking at this a while back when I first started to try and understand typical threshold designs… I think this (and similar designs) is what first confused me (probably because it was very different to what I'd done… well not really I guess - just "kind of different").  Funny - because it's really sticking out to me now, and there's clearly a simple diode based rectifier design + DC voltage stage there…

*Level Up*

+1 knowledge
« Last Edit: April 21, 2017, 06:52:57 PM by jBam »