Whats worse, 1db error in response curve or a phase error of 6?

[primalsea]

Thats my question really!

My Circuit Maker simulator shows that a 2nd order XO has 1db peaks either side of the XO frequency. I says peaks, they gradually decline back to 0db after a octave or 2.

I can get rid of these with a circuit that has a very smooth summed responce but at the expense of having the drivers 174? out of phase instead of 180?.

So whats worse, 6? phase error or a 1db responce error???


Any help any one


Cheers
Paul

[IanHarvey]

Originally posted by primalsea
Thats my question really!


So whats worse, 6? phase error or a 1db responce error???


Any help any one


Cheers
Paul

You're not saying what frequency you're talking about, but here's some maths: the speed of sound is something like 300m/s, so a 1KHz signal has a wavelength of .3m. A phase error of 6 degrees is like delaying the signal 6/360 = 1/60th of a wavelength, i.e. like moving your driver by 5mm.

If you XO frequency is higher, reduce this distance proportionally.

1dB level differences are generally considered audible, but I'm sure I don't keep my head positioned within 5mm of the ideal position when I'm listening. I'd therefore guess that the frequency response rather than the phase shift are more audible.

Cheers
IH

[qwad]

Most speakers have a bigger error than 6 degrees, i doubt if you will hear it but you will the peaks even if they are only 1 db, so if it were me i would rather smooth the peaks off best as i can, 'cause the range a couple of oct either side of the x/over point is rather prominent and in this area you are in the ears most sensitive range..... Cheers tjb

[Jiiim]

Personally, I always try to linearise the phase rather than the amplitude response.

The human ear (and brain) is incredibly sensitive to phase changes in sound.
(when you hear a sound and know it comes from a particular direction, that is phase sensitivity at work)

Just about all speakers have incredibly bad phase responses,
as this is an area that often gets ignored.
It does pay large and audible dividends to try to linearise them.
It's a very dificult effect to describe in words, but once you have heard it, a lot of speakers suddenly sound very naff indeed!


Jim.

[primalsea]

Jiim,

How about physical phase alignment.Do you think this is a real way of solving phase problems or does it just lead to more issues???

[taufern]

i think i would look at phase linear crossovers if you're that inclined...i usually just mess with 1st order to preserve the waveform...but 1db??? come on please be serious...in room response will be nothing like a computer modeling and 1 db is pretty damn impossible to tell...thats why even the most expensive speakers in the world only shoot for +- 2.5 or so db...i would worry about phase response first and realize the advantages of having +- 3 db at 30 degrees off axis versus +-.5 on axis...how many times do you really just sit in the sweet spot for hours on end...if you're like me you probably enjoy designing your next audio project while you're listening to your speakers...my last set of speakers had +- 3 db at 60 degrees off axis and i could say i was much happier with a very broad sweet spot than i was with prior results...