what region of the frequency spectrum are we talking about? and how much?
my secret weapon on the lows is McDSP ML4000 when its a tiny amount.
if that does not cut it I would grab fabfilter pro q for snappy upward compression.(technically a dynamic eq because instead of crossovers It uses shelf or a bell), and TDR Nova for al other tasks.

 

Okay, I'll clarify my statement:
A compressor where you can set negative ratios can be used for expansion. But a negative ratio does not expand, yes, you are right.
If the ratio of a downward compressor is set between 0.1:1 and 0.9:1, it expands the material upwards when the threshold is exceeded.

No, this is not correct. Assuming the used plugin can handle ratios lower than 1:1, the following applies:

• you can expand upwards with a downward compressor
• you can expand downwards with an upward compressor
• you can compress downwards with an upward expander
• you can compress upwards with a downward expander

This is due to the threshold or rather the gain computing.
A downward compressor cannot compress upwards because the compressor starts working only after the signal is above the threshold.
Just like an upward expander.
But in order to use upward compression, the device must start working when the signal falls below the threshold.
So like a gate or downward expander.

You can match transfer curves all you want, a downward compressor will not work the same as an upward compressor.
The makeup will boost low level signal, yes, but the results will be very different sonically. Here I'll show you:


Here I loaded Flux Solera as an upward compressor and a downward compressor respectively and matched the transfer curves.

Upward settings:

• Threshold: -32dB
• Ratio: 2:1
• Range: 12dB
• Attack: 20ms
• Release 170ms
• Make-up: 0dB

Downward settings:

• Threshold: -56dB
• Ratio: 2:1
• Range: 12dB
• Attack: 20ms
• Release 170ms
• Make-up: 12dB

So both plugin instances should do the same thing, right? Wrong! You forgot to include the threshold, gain computing and the time constants in your consideration.



What you see here is the test signal on track 1 ( 1kHz sine wave with gain automation -90dBFS , -12dBFS , -50dBFS ),
on track 2 the upward compressor and on track 3 the downward compressor.

Do you notice anything?
First of all, what is not visible on the picture, but is evident from the parameter values:
Both compressors start working at different levels. It doesn't matter what the transfer curve shows.

And of course, the direction in which both compressors work does not match. This is evident from the time constants.
In the example above you can see this especially in the release behavior. While the release of track 2 shows up when the signal level is at -12dBFS (because the level is above the threshold and the compressor relaxes), the release of track 3 shows up as usual for a downward compressor: when the signal level drops below the threshold. This would be even more obvious if the level after the -12dBFS region dropped below -56dBFS instead of -50dBFS (which is still above the threshold).

I think this shows very clearly why upward compression cannot be done with a downward compressor. You can boost low level signals by raising the makeup or by parallel compressing,
but you are not doing upward compression, you are doing downward compression!

Cheers!

 

Last edited: Nov 12, 2022

I made another small demo (with a non linear sweep) to show the difference (don't mind the steps, that's just the way it's displayed). All waves go from -inf to 0.0dB and back to -inf:

Spoiler

 

Thanks for clarifying! I knew that the threshold is completely different between upwards and downwards compression, but I didn't think about the envelope being that different. In hindsight, that's pretty obvious though

One thing I reacted to: AFAIK a negative ratio is anything below 0:1 (so -1:1, -4:1 etc). A ratio of 0.1:1 is still a positive ratio, no?

 

Sure. 0.1 is a positive number and a ratio of 0.x:1 leads to an upward expansion (for a downward compressor).