7 replies to this topic

[Shai]

Hi Guys,

This ones for your SQ gurus out there!

I have now finished my boot install and are running my splits from some quick (very rough) boxes I have made. I'll be using these boxes to find the right angle and direction for the mids and tweets before the door pods are constructed.

Anyhoo, Im setting gains (have read through the CAA and many other tutes) and it makes relative sense. My question has more to do with the slope of LP and HP filtering.

The way I see it - a possible configuration is High Pass the splits @ say 80Hz, Low Pass the sub at say 70Hz. Now from my understanding the 'slope' refers to the fact that at 80Hz and 70Hz are not 'rigid' figures. Subbass might start to be cut off from say 68Hz through to 72Hz and midbass from 78Hz to 82Hz...

My CDA9815 "apparently" lets me choose my slope? Does changing slope mean that (in the above example HP) 78Hz - 82Hz can be changed to say 76Hz - 84Hz? Is this known as flattening or steepening the slope?

Say I changed it from 12dB/oct to 8db/oct would this make the slope steeper or flatter?

In terms of SQ, would it not be the right thing to have a flatter slope for both HP and LP that overlap eachother so there is a smooth transition from the subbase to the midbass? If this is the case, why do many actually have a gap? Ie LP at 60 and HP at 100 (maybe exergerated)...would this not leave a 'hole' in the spectrum?

Im really sorry about the long post, but Im yet to find a decent write up on slope, what it does, what it means and how to best utilise it!

Thanks for your answers, they will most definately be appriciated.

Shai

[>WAYCON<]

Good answers and points are made about overlapping rolloffs here:
http://www.caraudioaustralia.com/forums/sh...t=cabin+loading

As for the roll off that you can choose, 6db/oct is less than 12db/oct etc. In that the signal is cut 6dB per octave in the first case and 12dB per octave in the second. So in other words 12dB is a steeper slope/cutoff/rolloff than 6

[dazdillinger]

"The way I see it - a possible configuration is High Pass the splits @ say 80Hz, Low Pass the sub at say 70Hz. Now from my understanding the 'slope' refers to the fact that at 80Hz and 70Hz are not 'rigid' figures. Subbass might start to be cut off from say 68Hz through to 72Hz and midbass from 78Hz to 82Hz...

My CDA9815 "apparently" lets me choose my slope? Does changing slope mean that (in the above example HP) 78Hz - 82Hz can be changed to say 76Hz - 84Hz? Is this known as flattening or steepening the slope?"

sort of yes and sort of no. If the lowpass is set at 70Hz itll start filtering at 70Hz, no matter what the slope is (correct me if im wrong). The difference is that filters dont act like a "brickwall" and prevent anything from over 70Hz passing through, they attenuate it ie if the slope is 6db/octave then for every octave over 70Hz itll attenuate 6db and then another 6db for the next octave and so on. If its 24db/octave tehn for every octave over 70Hz itll attenuate 24db. The 24db/octave slope is known as a "sleeper slope", if i could put some graphs up here you could see why. Basically the sleeper the slope the more "rapidly" it will attenuate the desired frequencies to be filtered. This gives advantages as it enables us to set x-over frequencies higher or lower to the extremes as we know it will filter it off sharply.

"If this is the case, why do many actually have a gap? Ie LP at 60 and HP at 100 (maybe exergerated)...would this not leave a 'hole' in the spectrum?"

As i wrote above, the crossovers dont completely block all frequencies above or below the crossover point, they attenuate ie. Hence if someone has a lowpass of 60Hz and a high pass of 100Hz there are still frequencies within that range that "pass through" the filter and will still be audible

"In terms of SQ, would it not be the right thing to have a flatter slope for both HP and LP that overlap eachother so there is a smooth transition from the subbase to the midbass"

In terms of SQ i dont think any general statements like "flat slopes are better" will work as there are too many variables between what components you have, what your install is like and what the car interior is like, compared to someone else. Its all about experimentation. Just gotta fiddle around with diff crossover points and slopes to see what sounds best to you and perhaps what measures well.

Hope i cleared some stuff up, wait for shiny to give a better explanation!

[jas]

no you can not do that

here are the settings

0,6,12,18,24 db/oct highpass or lowpass x-over slopes

125,100,80,63,50,40,31.5 Hz x-over points

if you are using 6.5inch midbass/mids then 80,63,50hz are the only real choices. I would not recommend any other setting

what you most likely will do is settle on a common x-over point for both, say 80hz and then choose different slope values to create over or underlapping roll-offs if desired. It all depends on your setup and how much bass and also rear bass you like.

i set mine to 50hz @ 18db/oct for high and low pass in my honda civic. It was the only setting that i liked , hey it only took a week of changing, listening, changin..etc 50hz was chosen due to the fact that i could say that the bass was "up front".

so have fun and pplay around with it

oh yeah forgot to mention that the x-over point may not be -3db down at the point chosen due to the fact that these are digital x-overs, hopefully they do follow the analogue standard for naming slopes. For example a high pass 80@ 24db is actually -3db @ 80hz in an analogue x-over. One octive below this would be -27db @ 40hz.

[Cyberpunky]

This is hard to explain with out diagrams but here goes

lets call our filter point A and one octave away B if we use a 6 db Xover then our signal will drop 6 db at B while at A its at full strentgh. If we have a point C that is another octave away, the signal will have dropped another 6 db from point B so will be 12 db below point A. this as a shallow slope. If we use 24 DB filters at point B the signal will have dropped 24 db and 48 db at point C so this is a steep slope.

if we use 72 db slopes at point C it would be 144 db below the signal at point A .

OK assuming you get that. Lets picture the 2 slopes of a HP and LP. If we stick with 6 db then the at A point the signal will be full strength for the low pass but the HP signal will be down 6 db and vice versa at point B. In the middle, both slopes will be 3 DB down.


Now we know if we add a second speaker like a 2nd sub then the 2 subs couple and basically double the output, all else being equal compared to one sub. The same is tru for a midbass and sub working at same freqs. so in the middle between AA and B the 2 combine and so we instead of the signal dropping at the middle they combine and w have the to - 3 db points adding and we end up with the signal not dropped at all.. now if we move to half way between the mid point and point A one signal will be down 1.5 db and the other 4.5. these add and we are back at a flat signal.

So by having a gap we actually end up with a flat response between the 2 crossover points. Having a gap is called underlap btw. Now if the xovers were total filtrs and cut off everything then we could have the same xover point for LP and HP.

I really wish I could just use diagrams but thats the best I can do with words. Hope it helps
peace
Cyberpunky

[bodapa]

Just want to add...with overlapping and underlapping the crossover points and varying the slopes you basically using them to counter-act the cabin gain of the vehicle. I think someone mentioned this in the previous posts. As to whether you need to overlap or underlap (or even sharing the same points) depends on the interior of the vehicle. This is one of those situations where an RTA is useful, because you can visualise the frequency response of your vehicle and make adjustments where necessary, though setting them by ear are OK too.

Cheers,

Bon

[Bassaholic]

One point to consider, is the rolloff doesn't magically go from 0db/octave to 12db/octave (or whatever slope), the rolloff is a bit more gradual, depending on the Q of the filter.

In the case of a Butterworth filter (Q=0.707), the crossover point will be the 3dB down point.

In the case of a Linkwitz-Riley filter (Q=0.49), the crossover point will be the 6dB down point.

(6dB/octave crossovers have a fixed Q of about 0.5)

Secondly, if you were to combine the output of two speakers, which happen to be playing at the same volume level, at low frequencies you would gain 6dB. (when the signal from each speaker is practically in phase) This means that (assuming the speakers are mounted close together enough relative to the wavelength), if you were to match the crossover points to the same frequency, you would want them both to be -6dB down at that crossover point - you would choose Linkwitz-Riley filters.
Secondly, you have to consider the relative phase shift of the crossover - in the case of 12db/octave crossover, the relative phase shift is 180 degrees (easily compensated by reversing the polarity of the wiring to one of the drivers). In the case of a 24db/octave crossover, the relative phase shift is 360 degrees (0 deg) (so the net result is that the drivers will be in phase. However, the phase shift of 6dB/octave crossovers is 90 degrees and the shift of 18dB/octave crossovers is 270 degrees (-90 deg) which is a bit more difficult to compensate for, so the combined frequency response will be a bit different. However, this is assuming that the drivers are mounted virtually next to each other.

In the case of car audio, the front stage midbass drivers are not that close to the subwoofers. So the relative phase at the crossover point will also depend on the distance between the midbass driver and the subwoofer, vs the wavelength of the frequency at the crossover point. (to work out approx wavelength, divide the velocity of sound by the frequency, ie with an approximate velocity of sound through air of 344 m/s, the wavelength of 80hz would be 344/80 = approximately 4.3m.)

Thirdly, you have the transfer function (cabin gain) of the vehicle which will have its own effect on the frequency response. As Bodapa mentioned, this is one of the few situations where an RTA is actually useful - if you can get a smooth transition then everything is fine.

Jas - the filters may be digital, but that doesn't necessarily mean that they are FIR filters...

Oh yeah, here some pics showing what I mean in regards to the Q of the crossover. (borrowed from http://www.bcae1.com/xoorder.htm )

Second order (12db/octave) Linkwitz Riley:

http://www.bcae1.com...fs/xo2ndlra.gif

Second order Butterworth:

http://www.bcae1.com...fs/xo2ndbwa.gif

Notice how, since the response of each driver is only -3dB at the crossover point, the summed response is +3dB at the crossover point.

[jas]

the digital filter can be manipulated in any way you want. i would be surprised if the designers copied any known analogue filter type due to the fact that they would also mimic their phase or frequency short commings. The digital filter will suffer from analogue filter short commings due to the fact that you require a filter to "translate" the digital stream (PCM) to analogue. 12db/oct low pass op-amp filters and current to voltage filters must be used to reconstruct a digital signal using a multi-bit dac. These extra phase and frequency anomolies are not as bad as having this filter inside the "pass band" (20-20khz).

the digital filter doesnt have the phase or frequency problems that analogue filters have. Their phase delay is linear and is totally dependant on half the length of the FIR impulse chosen. In the end the digial filter is far from perfect but has great advantages over its old analogue cousin.

BTW id be very surprised if they use IIR filters

BAssaholic : good point about the x-over -6db(linkwitz) and a few more variations with different Q filters. LAst time i did these calculations i used the good old butterworth (-3db). Good pickup and very well explained