undefined | Better HN

0 pointsKaiserPro12y ago0 comments

Thats what limiters are for. The whole point of hard clipping is that +256 or -256 (or the 16bit equivalent) is below the point at which your speakers, amps, cabling or anything else starts to melt.

Speakers/amps should have DC protection builtin.

0 comments

lutusp12y ago

> Thats what limiters are for.

Limiters cannot either distinguish or protect against the technical difference between a sine wave and a square wave:

http://i.imgur.com/oE5NFZ9.png

The red trace is a normal audio signal that doesn't exceed the dynamic range of the amplifier. The green trace is an example of the kind of signal one gets by turning the volume up way above normal.

In a hypothetical speaker of one ohm and a peak voltage of one volt, the red trace has a time-averaged power level of 1/2 watt. The green trace has a power level of one watt -- twice as high, and possibly too high for the speakers to tolerate.

> Speakers/amps should have DC protection builtin.

This example should show that this isn't possible without taking the temperature of the speaker's activation coil -- and even that might not work.

userbinator12y ago

DC protection = a <$0.01 capacitor. The bigger the value, the lower the frequency cutoff. A square wave passing through a high-pass filter gets cut off and the power is reduced.

IMHO everyone who is saying "this should be fixed by software" is doing it wrong. How much would it cost to write, test, debug, etc. the appropriate firmware/drivers? And it would still be subject to "warranty is void because you didn't use the right software". That cost could buy a lot of capacitors... and then if the values were chosen correctly, whatever the software does, it would not be able to put more power through the speaker than designed.

caf12y ago

An analogue bandpass filter could be used. Passing a square wave through such a filter will remove much of the power present at inaudible frequencies. Unfortunately, they also require bulky inductors/capacitors.

tmzt12y ago

Couldn't he codec chip do it by averaging the amplitude? I'm not sure if they are more than a DAY though or whether they have a microcontroller and firmware.

lutusp12y ago

> Couldn't he codec chip do it by averaging the amplitude?

Yes in principle, but it would be rather complicated -- it would need to have its own microprocessor just to save the speaker. D/A ICs with included amplifiers are already pretty complex. Also, if a user was listening to music that had an occasional high level, the special limiter might kick in and spoil the sound of the audio for a transient that wouldn't actually jeopardize the speakers.

j / k navigate · click thread line to collapse

0 comments

lutusp12y ago

> Thats what limiters are for.

Limiters cannot either distinguish or protect against the technical difference between a sine wave and a square wave:

http://i.imgur.com/oE5NFZ9.png

The red trace is a normal audio signal that doesn't exceed the dynamic range of the amplifier. The green trace is an example of the kind of signal one gets by turning the volume up way above normal.

> Speakers/amps should have DC protection builtin.

This example should show that this isn't possible without taking the temperature of the speaker's activation coil -- and even that might not work.

userbinator12y ago

DC protection = a <$0.01 capacitor. The bigger the value, the lower the frequency cutoff. A square wave passing through a high-pass filter gets cut off and the power is reduced.

caf12y ago

tmzt12y ago

Couldn't he codec chip do it by averaging the amplitude? I'm not sure if they are more than a DAY though or whether they have a microcontroller and firmware.

lutusp12y ago

> Couldn't he codec chip do it by averaging the amplitude?

j / k navigate · click thread line to collapse