On power

[Julf]

Then you have a very different understanding of how a compressor works than I have.

All compressors have an attack and release time, so don't act instantaneously. Clipping is instantaneous, and acts on any signal that exceeds the maximum available voltage output. Clipping always increases distortion, compression and limiting doesn't.

I do know how the studio device that is referred to as a "compressor" works. Yes, that is a very spacial case of compression - I was actually thinking of adding "the exception is of course adaptive "compressors" that change the gain/volume based on the signal level, but with a time lag for recovery" to my previous posting. That is different from the compression that happens when an amp runs out of steam in peaks.

 

[anmpr1]

If you are Paul Klipsch, then what is needed is a good 5 watt amp. If you are Bob Carver then the answer is 700 watts RMS. So it's got to be somewhere in between those two figures, I'd imagine.

 

[sergeauckland]

I do know how the studio device that is referred to as a "compressor" works. Yes, that is a very spacial case of compression - I was actually thinking of adding "the exception is of course adaptive "compressors" that change the gain/volume based on the signal level, but with a time lag for recovery" to my previous posting. That is different from the compression that happens when an amp runs out of steam in peaks.

Exactly. What you're describing is not compression, it is clipping. There is no compression taking place when an amplifier clips, it's just pure clipping, i.e. running out of volts. That always means an increase in distortion, although the ear is remarkably accommodating when the clipping is short, which is why so many CDs are clipped, not just limited.

Compression involves a change in gain, and no amplifier changes its gain when it runs out of volts. An amplifier may behave oddly in clipping, with increases in output impedance as feedback fails, and/or latching when the release from clipping isn't instantaneous. None of this is compression.

As an aside, I suspect that the reason why many subjectivists think all amplifiers sound different is because they allow them to clip on audition, and what they're hearing is the difference in overload behaviour, which could well be different.

S.

 

[Julf]

Exactly. What you're describing is not compression, it is clipping. There is no compression taking place when an amplifier clips, it's just pure clipping, i.e. running out of volts. That always means an increase in distortion, although the ear is remarkably accommodating when the clipping is short, which is why so many CDs are clipped, not just limited.

Compression involves a change in gain, and no amplifier changes its gain when it runs out of volts. An amplifier may behave oddly in clipping, with increases in output impedance as feedback fails, and/or latching when the release from clipping isn't instantaneous. None of this is compression.

As an aside, I suspect that the reason why many subjectivists think all amplifiers sound different is because they allow them to clip on audition, and what they're hearing is the difference in overload behaviour, which could well be different.

S.

An amplifier doesn't suddenly go from "no distortion, fully linear" to clipping. There is a transition, where distortion increases as a result of not being able to (fully) accommodate the amplitude / power demand. That increasing distortion is a result of compression. Ever worked with tube amps?

 

[Jim777]

An amplifier doesn't suddenly go from "no distortion, fully linear" to clipping. There is a transition, where distortion increases as a result of not being able to (fully) accommodate the amplitude / power demand. That increasing distortion is a result of compression. Ever worked with tube amps?

Modern amps clip suddenly indeed, look at any measurement on this site. THD rises like a hockey stick graph...

 

[digitalfrost]

Personally, I've calibrated my system to 78dBC at 2/3 volume, and I crank it to 83dBC or 86dBC depending on material. That said, I've followed the guidelines for subwoofer setups (+10dB peaks) and I've measured the explosion at the beginning of "The end of tomorrow" or the airplanes in "Top Gun" at 107dBC.

While I listen in the near field, I use Tang Bang W5-1183SMF as subwoofers (4 chassis) and they measure a measly 75dB at 1W at 40hz, although they are rated 82dB (see here http://www.tb-speaker.com/products/w5-1138smf). It works good for me. Satellites are either ER18DXTs or KEF LS50, which are almost identical in terms of efficency.

I use AHB2 for the sats and XTZ EDGE A2-300 for the woofers. They are wired in series, so they are 8 ohms which means ~ 150W max with the XTZ.

Personally I think if you need more than 100W to reach the required volume, you need to change speakers/setup. Sure you can buy 200W/8ohm amplifiers, but really after this the air gets very thin. I used an amplifier with VU meters before the AHB2, and I've never exceeded 10W/8ohm when listening really loud. That is for the sats. Subwoofers are a different story. That said, I own an Tripath 10W (more like 8W without distortion) amplifier, and it is not enough for anything if you want proper volume. Horn loudspeakers maybe.

Hope this gives some perspective.

 

[sergeauckland]

An amplifier doesn't suddenly go from "no distortion, fully linear" to clipping. There is a transition, where distortion increases as a result of not being able to (fully) accommodate the amplitude / power demand. That increasing distortion is a result of compression. Ever worked with tube amps?

View attachment 32165

That graph isn't representative of SS amps, which do go from linear to clipped instantly. Also, the axes of that graph make little sense as audio amplifiers don't have a power gain of only 3x, they have a voltage gain of, typically 10-20x, but a power gain of thousands, given their high input impedance and low output impedance.

Nevertheless, what you show is still not compression, but non-linearity, i.e. distortion

S

 

[Julf]

That graph isn't representative of SS amps, which do go from linear to clipped instantly.

Which is why I asked "Ever worked with tube amps?". Most solid state amps don't go from linear to clipped instantly either - look at a typical power-supply-bound class A/B amp, not the latest class D amps.

Also, the axes of that graph make little sense as audio amplifiers don't have a power gain of only 3x, they have a voltage gain of, typically 10-20x, but a power gain of thousands, given their high input impedance and low output impedance.

The absolute gain has no relevance to the discussion - the effect is the same, be the gain 2 or 20000.

Nevertheless, what you show is still not compression, but non-linearity, i.e. distortion

What is your definition of compression?

By the way, I borrowed the graph from here: https://en.wikipedia.org/wiki/Gain_compression

 

[Julf]

Modern amps clip suddenly indeed, look at any measurement on this site. THD rises like a hockey stick graph...

Look at a typical power-supply-bound class A/B amp, not the latest class D amps.

 

[Willem]

The wiki article is useful. It shows the issue is worst with valve amplifiers. That is all the more serious since typically these have less power. To be honest, it was primarily those that I had in mind when I first posted.
The practical consequences are clear: do not save on watts. So when I bought a power amp for my son's birthday I bought a 2x250 watt rms at 8 Ohm Yamaha p2500s - for 300 euro (new).

 

[sergeauckland]

Look at a typical power-supply-bound class A/B amp, not the latest class D amps.

They still don't compress, just clip.

Maybe we have a different understanding of dynamic range compression. My understanding is that gain is altered dynamically such that the input and output don;t track, BUT there is no increase in distortion, as the gain is adjusted dynamically. What you're referring to, which I understand as clipping, is when the amplifier runs out of volts, so flat-tops, thus the output voltage level no longer increases to any appreciable extent, but the distortion rises very rapidly.

Two totally different processes.

As to what valve amplifiers do, that's again another issue, and in my view most valve amplifiers are not transparent (although a few can be) and thus are effects boxes, not High Fidelity amplifiers.

S.

 

[Julf]

Maybe we have a different understanding of dynamic range compression.

No, we have the same understanding of dynamic range compression as done by a compressor in the studio. I am talking about something else. I am talking about gain compression. See the link above.

 

[sergeauckland]

No, we have the same understanding of dynamic range compression as done by a compressor in the studio. I am talking about something else. I am talking about gain compression. See the link above.

But gain compression IS dynamic range compression. What you keep referring to as gain compression is just plain old clipping. There are NO gain changes, the amplifier just runs out of volts.

S

 

[Julf]

But gain compression IS dynamic range compression. What you keep referring to as gain compression is just plain old clipping. There are NO gain changes, the amplifier just runs out of volts.

Again, except for some class D amps, amps don't run out of "volts" in a binary way, but there is a transition band where the amp doesn't clip totally, but the tops get rounded off - and the output voltage is lower than it should be. Gain is the ratio of output voltage to input voltage. If the output voltage increases less steeply than the input voltage, the gain has decreased. If tops get rounded off, that is compression.

 

[sergeauckland]

Again, except for some class D amps, amps don't run out of "volts" in a binary way, but there is a transition band where the amp doesn't clip totally, but the tops get rounded off - and the output voltage is lower than it should be. Gain is the ratio of output voltage to input voltage. If the output voltage increases less steeply than the input voltage, the gain has decreased. If tops get rounded off, that is compression.

Whatever. I'm out of this discussion as we're getting nowhere.

S

 

[Julf]

But gain compression IS dynamic range compression.

Again, what does this graph tell you? Yes, you need valves or possibly FETs to get this soft a curve, but a solid state amp shows exactly the same behavior, just not as profound as this.

 

[Julf]

Whatever. I'm out of this discussion as we're getting nowhere.

"We are getting nowhere because you refuse to see things my way, and I refuse to understand your point".

 

[Willem]

But the take home is that insufficient power is bad. And my view is that in reality you may well need a lot of it.

 

[Julf]

But the take home is that insufficient power is bad.

I don't think anyone disagrees with that

And my view is that in reality you may well need a lot of it.

Depending on the (in)efficiency of your speakers, your room, and how loud you like to listen, yes. But if you need a lot of power, noise floor is usually not an issue.

 

[Willem]

With my inefficient electrostats noise has never been audible.