Presswerk user manual question (feedback detection)

[jackson881]

I'm sure I'm just overlooking something obvious but I'm confused by this part of the Presswerk manual.

Feedback detection does have its drawbacks, however. Not only are the threshold and ratio set- tings fairly arbitrary and program-dependent (i.e. governed by characteristics of the material being processed), but the compression ratio is also limited to values between 2:1 and 3:1.

For Presswerk, we chose not to allow more than about 2.5:1 ratio in FB mode

I'm able to set the ratio knob up to a 20:1 setting using Feedback Detection in Presswerk, so why does the manual states that FB mode "doesn't allow more than about a 2.5:1 ratio"?

This has had me scratching my head for a while now. Any help would be appreciated!

[sascha]

With feedback loops, laws of system theory and control engineering apply. Such a loop is a self-adjusting system, you can compare it it to your domestic heating system: there's a regulator on the wall sensing the current temperature, and if that is below a set threshold, it triggers the heating pump. This goes on until the temperature is achieved (almost completely). Thing is, the greater the deviation between wanted and actual temperature, the longer the process, likewise the more energy must be expended.
With control loops in a dynamics compressor, this corresponds to gain applied to the loop. But this also lowers the threshold, thus rendering the process useless at some point. With an amplification factor of 1, the resulting ratio will be around 2:1. That's why some processors include nonlinear amplification in the control loop. We do it, to some extent, but since feedback compression is dependent on the material (reacts different on transients as opposed to more sustaining audio), high amounts of nonlinear gain lead to something that sounds like sucking transients out of the signal; it becomes more unnatural. Therefore we limit the process in a way that we find to be a sweet spot between efficiency and musicality.

[EDIT: in order to obtain a large/imiting ratio in FB mode, a negative theoretical ratio (-n:1) has to be applied to the loop. That's what makes the 'Non Lin' parameter on Presswerk quite handy, just bend it down to have a slightly higher overall ratio.]

Question might remain why this is not reflected on the GUI. Answer is simple: feedback loops are unpredictable, the result is program dependent. You can make the test plotting the response curve (using something like VST Plugin Analyzer) in FF and FB until they both match, but chances are FB sounds like squeezing things more but at the same time riding the gain quite efficiently.

Not an easy topic, I know... I hope it's becoming clearer now.

[jackson881]

Thanks Sascha! I really appreciate you taking the time to respond in such detail. With a feedback design like the 1176, is it "nonlinear amplification" that allows it to achieve such high ratios?

Excuse my ignorance. I'm looking forward to studying more on this topic. Thank you for giving me a starting point!

[sascha]

With a feedback design like the 1176, is it "nonlinear amplification" that allows it to achieve such high ratios?

Exactly. The nonlinearity there is mainly due to the FET acting as part of the voltage resistor network, and the way it is driven.