2ndhannes

Hi all (especially Max ,

did anyone besides me notice that the slope of the crossover plugins
(both 2nd and 4th order) seems to be imperfect and to vary with
cutoff frequency?

I am trying to use the lower half of a "Crossover-4"-plugin (from the ufx
download pack, I think I have 3538k) set up a subwoofer lowpass crossover
with cutoff frequency at 100.2 Hz, and with this parameter I get a slope that
falls with a bit more than 6db/oct instead of the 24db/oct I expected.

With a higher cutoff frequency, the slope gets closer to the intended
slope, but the response seems to be shelving at something like -40db
at the top end of the spectrum.

Am I doing something wrong / is my measurement software broken?

Bye,
Hannes

Max M.

the slope should not depend upon frequency. so either this is a bug within plugin or a problem with a measurement (50/50 ;). well, actually some drift in slope is expected at low frequencies (due to non-infinite precision) but not so drastic.
could you probide some more details on what/how (which software) you test it?
btw. note that FFT algorithm (typically used for analisys) itself can give very poor/coarse results at low frequencies so be sure to set the best possible options for the measurement soft, if it has any (especually 'n of fft points').

and, sure, i'll doublecheck the crossover itself too.

Max M.

i've tested. the crossover appears to be ok.

2ndhannes

Hi,

thanks for the quick answer... I am using the "Hobbybox" software for measurement,
it allows measuremet of the frequency response via MLS or impulses. I've tried both
and arrived at similar results, so maybe there is a flaw in the measuring setup.

I will do another round of measuring tomorrow and post my findings here.

And while I have your attention Max: thanks alot for the ufx bundle and the crossover
plugins, I am very glad to have these useful tools and I am using (& tweaking) them
almost daily, I have been using my A2zs as an active crossover for various loudspeaker
projects for quite a while now and very much like the flexibility and the sound quality
this gives me!

Bye,
Hannes

2ndhannes

Hi again,

I've just had another look and it seems indeed that the measurement software has some
troubles: it fails to correctly record the whole impulse response: for long rise- /fall-times
(as they occur with lowpass filtering), it misses the beginning and also a bit of the decay
part of the signal, thus giving me a quite different response curve. That's what happens in
the MLS case, with impulse measurement I get a saturated "step response", I wonder what
happens there.

I also tried RMAA55, but that fails to find it's sync markers in the lowpass-filtered response
and crashes

So I take everything back and feel a bit stupid - just wondering how I could teach the
software to do decent measurement - this might have to do with the kX buffer settings,
which I fail to understand properly

Bye + thanks for your time,
Hannes

Max M.

it's possible to fool RMMA if you set left channel to highpass part of the xover and the right channel to the low-pass one.
see:
http://kxdev.narod.ru/xover/setup.png
http://kxdev.narod.ru/xover/result.png

bzdang

I have been using measurement software from www.audiotester.de with very good results. Here is a test where I connected (in kxproject dsp window) fxbus 0/1 to crossover-4 and then outputs from crossover-4 to left and right rec inputs of epilog. Ran a frequency response sinus sweep and captured the screen display to a jpg image. Where the curves go off the bottom of the graph the 'pens' will just travel along the bottom of the frame, is a matter of choosing range to display.
http://img159.imageshack.us/img159/2...ssover4yy9.jpg