Crossover: IIR or FIR?

[thumb5]

It seems to me that Devialet had a choice of how to present this capability in the configurator: they could have provided a "crossover" black box with a couple of parameters (cross-over frequency, filter order, ...) which would have been easier to use. Instead they chose to expose the underlying filters, which gives more flexibility for those who want or need it - the disadvantage being that it requires a bit more knowledge from the user. IIRC the default filter configuration is second order in which case it naturally "does the right thing", doesn't it?

[Waiski]

Yes, second order does work correctly. On the others you need to use the calculation I posted earlier.

[zdenes]

If your sub has 2nd order crossover, then you can leave sub out to be full range. And speaker output 2nd order 80hz.

Thanks! This thread is very informative.


Sent from my iPad using Tapatalk

[maxijazz]

Im general, in electronics, cut off frequency is the one where there is 3dB fall (or raise) in signal power.
It would mean that 1-st order filter is properly implemented (-3dB fall at 250Hz). The other filters no so well. I have to admit that different types of filters could have different definitions of cut-off

Not -3 dB.

Crossovers are -6 dB/per per order / per octave.  So 1st order is -6 dB per octave....4th order is -24 dB per octave.

Here's what I don't get:

Why would Devialet implement this in such a non-intuitative way?

I am right and you are right. It is misunderstanding. 
I am talking about definition of cut-off frequency. It is at - 3dB. 
You are talking about filters' slopes (roll-off/damping rates). 

Cheers.

[maxijazz]

I don't get it. They don't even inform us how to use the crossover...

I never found out how filters of Tone Controls work.

[Music or sound]

Okey, so some data.

Under you see the measurement taken from pre out of D250 Pro. We did check and also speaker output crossover does work the same way.
Measured is fullrange and 250hz 1-4 order crossovers, high pass and low pass.

You can see from the measurements that 1st order crossover does give -3db on crossover freq, 2nd order -6db, 3rd order -9db and 4th order -12db.
So basically Devialet does multiply 1st order crossover 1-4 times. This means that you can use same filter order for high pass and low pass only for 2nd order crossover. Other crossover filters need to be build with calculation I sent earlier.

Order
LP=>-6dB ... -6dB=>HP
HP=>-6dB ... -6dB=>LP
1st 1.733 ... 0.577
2nd 1.000 ... 1.000
3rd 0.765 ... 1.307
4th 0.643 ... 1.556

So for 2nd order crossover you can use this without adjustments. 1000hz is 1000hz crossover point.
But for example 1000hz 1st order crossover point are... low pass is 577hz and high pass is 1733hz.

One of the reasons I got a Devialet was the crossover as I am wanted a digital crossover for my diy loudspeakers. I am using the amp as a for my bending wave design (mids and high) high pass at 400hz and the pre-out for a digital amp driving my woofers. I never measured directly the crossover curves at the amp or pre-out but if I measure my speakers with and without the crossover and look at the difference I get very similar curves except early on the curves were flatter than they should at any crossover slope.
To get a perfect even summ of low pass and high pass they should be -6dB at the crossover point and as the curves show that is only the case in 2nd order. As one can choose independently the crossover frequency for high and low pass one can adjust for a -6dB crossover point at any slope.
Crossover points and slopes are only working anyhow for completely flat frequency response (and than why on would need a crossover).
Because of the slope and rinse of my bending wave driver I get the best results if I an using a setting at 800Hz 2nd order and 300Hz 3rd order for the woofers to get a crossover which sums flat at 400Hz. 
I think to use a crossover correctly measurements are quite necessary.

[Waiski]

Okey, so some data.

Under you see the measurement taken from pre out of D250 Pro. We did check and also speaker output crossover does work the same way.
Measured is fullrange and 250hz 1-4 order crossovers, high pass and low pass.

You can see from the measurements that 1st order crossover does give -3db on crossover freq, 2nd order -6db, 3rd order -9db and 4th order -12db.
So basically Devialet does multiply 1st order crossover 1-4 times. This means that you can use same filter order for high pass and low pass only for 2nd order crossover. Other crossover filters need to be build with calculation I sent earlier.

Order
LP=>-6dB ... -6dB=>HP
HP=>-6dB ... -6dB=>LP
1st 1.733 ... 0.577
2nd 1.000 ... 1.000
3rd 0.765 ... 1.307
4th 0.643 ... 1.556

So for 2nd order crossover you can use this without adjustments. 1000hz is 1000hz crossover point.
But for example 1000hz 1st order crossover point are... low pass is 577hz and high pass is 1733hz.

One of the reasons I got a Devialet was the crossover as I am wanted a digital crossover for my diy loudspeakers. I am using the amp as a for my bending wave design (mids and high) high pass at 400hz and the pre-out for a digital amp driving my woofers. I never measured directly the crossover curves at the amp or pre-out but if I measure my speakers with and without the crossover and look at the difference I get very similar curves except early on the curves were flatter than they should at any crossover slope.
To get a perfect even summ of low pass and high pass they should be -6dB at the crossover point and as the curves show that is only the case in 2nd order. As one can choose independently the crossover frequency for high and low pass one can adjust for a -6dB crossover point at any slope.
Crossover points and slopes are only working anyhow for completely flat frequency response (and than why on would need a crossover).
Because of the slope and rinse of my bending wave driver I get the best results if I an using a setting at 800Hz 2nd order and 300Hz 3rd order for the woofers to get a crossover which sums flat at 400Hz. 
I think to use a crossover correctly measurements are quite necessary.

No need to measure. You can use the calculation I sent. Then you get flat crossover summing. You can use this also to your 3rd/2nd crossover. If you want to have flat crossover at 400hz you put 2nd order high pass to 400hz and 3rd order low pass to 523hz.

[Music or sound]

No need to measure. You can use the calculation I sent. Then you get flat crossover summing. You can use this also to your 3rd/2nd crossover. If you want to have flat crossover at 400hz you put 2nd order high pass to 400hz and 3rd order low pass to 523hz.

Only if one assumes a flat frequency response.

[Waiski]

No need to measure. You can use the calculation I sent. Then you get flat crossover summing. You can use this also to your 3rd/2nd crossover. If you want to have flat crossover at 400hz you put 2nd order high pass to 400hz and 3rd order low pass to 523hz.

Only if one assumes a flat frequency response.

If you mean speaker elements then you are absolutely correct.