Crossover Slopes?

Trick

Hi, a question on the crossover slope settings.

It was my original understanding the the slope begins at the frequency selected, and then falls off accordingly to whatever db/octave rating specified. So if there was a 18 db/octave lo pass at 100 hz set to go to the subs, I thought that it would let 100 hz pass with no db reduction and then fall off. After listening and visually looking at it, I hear and see that 100 hz is effected and begins to roll off way before it. So with that 18 db/octave I was assuming that 200 hz would be 18 db down. Basically, how do i figure out what frequency the crossover slope begins to fall off or begin with each filter type with no db reduction? Another question, if I set crossover for a 2 way setup at 100 hz. It automatically applies -3 db reduction at the xover point, correct?


thanks for your assistance,
Sean

Dra

Imagine a horizontal line (representing the unaffected frequencies). At one end (or the other depending on which section of frequencies you are going to affect) there is an intersecting / joining line that is angled. The angle severity is the slope, with smaller numbers being very slight and higher numbers very steep. These numbers are fixed to a degree of steepness, regardless of filter type. The difference comes with the filter type chosen (BW, LR, BS, etc) as it applies to the "knee" of the joint. The "knee" is the transition of the unaffected frequencies to the affected frequencies. Some "Knees" are smaller/tighter and some are larger.

This transition is in the form of a radius. The radius size is the reason your 100hz x-over actually has a loss of energy (filtering) that begins before the 100hz chosen. This loss (typically -3db) is regained when the adjacent spectum's speaker it give the same filter and slope type. High's speaker is -3db @ 100hz, low's speaker in -3db @ 100hz. The 2 speakers are reproducing the same 100hz content, and while each is producing it at a reduced level, they both are reproducing it, so it sums or combines to bring the output up to what would be the normal unfiltered level. If memory serves... the specified freq is at the mid-way point of the radius.

Generally, we recommend LR24 for all filters that are going to sum with other filters. BW18 is typical for the subs' bottom end, and usually set at the -3db point of the spec.

Hope that helped.

DRA

Trick

thanks for the help...I understand your explanation on the -3db xover point and am still not totally clear on which freq's are unaffected, so here is another example:

I still have the lo pass at 100 hz, and have now set the filter type to BW 6. This is clearly taking out more than -3 db at 100 hz. Visually it looks like the knee begins at 60 hz or so. It looks as if 100 hz is where the attenuation is at it's maximum.

I would not use the BW 6 but am giving an example to help me understand where the attenuation begins given the freq chosen.

thanks for the help

Dra

You keep saying "see" and "visually". What are you talking about? Are you using SMAART or some other detailed measuring system?

DRA

Trick

I am saying 'see' and 'visually' because of the front panel display on the DR. In the top row, the freq's are printed above the display. I hear the drop out also. If I have it set at 100 hz, shouldn't the different filter types let 100 pass with either no attenuation or -3db, and then the slope of the knee would change affecting the attenuation of freq's after 100.

Dra

Are you saying that the drop out is audible with the sub crossed at the same frequency & slope?

DRA

Gadget

Ah but then we add a monkey to the works...the display might be only a representation, but is it the acoustical rolloff point or the electrical rolloff point... they are never the same...

Trick

I made a short video to better explain my question. Unfortunately nothing is in focus but I know you'll understand explanation and question.

thanks

http://www.youtube.com/watch?v=lkvzaVygbD4

Dra

That is a good tool to explain what you are seeing, but we covered the reason why already.
The frequency is IN the knee not at the start of the knee. The loss will be made up when the other range speaker is added.

Can you do this?
Send the 100hz tone to the driverack and send the high and low outputs (each 100hz LR24) into 2 inputs on the Presonus. Then raise or lower each filter to 80, 90, 100, 118, 125, 130, 200, etc. You should see the level stay the same at all changes. If there is still a loss, try placing a one notch gap, or a one notch overlap. Try mixing filter types. This may or may not be futile, since thre is no "acustical audio" to measure here, ut is looks like fun, none the less.

DRA

Trick

The frequency is IN the knee not at the start of the knee.

I always thought it wasn't supposed to be in the knee, but start before it. thanks for clarifying. The part I'm concerned with then is setting the crossover properly. Here's my concern: If I'm setting 100 hz for the lo and hi (2 way system), is 80 hz (fundamental kick drum) getting attenuated in the lo and hi and getting the full output of 80 by summing together. I would like to find the appropriate crossover freq that 80 will not have any attenuation. What you mentioned to test sounds like a good way for me to achieve this. I feel my crossover point is going to end up too high though. It would be so much simpler if the crossover freq was at the start of the knee, wouldn't it?

The BW 24 sounds nice to me on the LO because it's in fact giving me more output at 100 sending a kick drum through. I will read elsewhere about it, but is there a simple understanding why the LR is preferred. With the system I have listed below I feel like too much thump is getting through to the tops at 100.

the system I'm tuning is: QSC KW 181 & Yamaha DSR 112

any suggestion on setting would be helpful.

Trick

Ok, I did test running both lo and hi into inputs of the console and moved crossover around. With that version of smaart on there I cannot see Left and Right linked only individual, so that sort of sucks.

So using a LR 24 filter set at 100 hz for the lo and hi, I input an 80 hz sine. There was 6 db less of 80 going to the hi's. That isn't much attenuation is it? I mean the kick drum is still going to come ripping through the tops. And it does cause I've listened to it and seen the woofer. Is this acceptable with the system I have mentioned?

Dra

Have you tried PSW? There are a lot more techies that hang out there.

DRA

Gadget

Look, I haven't got the time right now to get into this much.. I gave you a hint here and that is that the electrical and acoustical properties WILL be different. Where you set the Electrical (typical crossover point you see on the driverack or other DSP...) crossover point is NOT where the acoustical (actual) crossover point ends up...

I would NOT be putting all my eggs in the SMAART basket that comes with the Presonus either...it is FAR from a full blown SMAART package and is little more than a glorified RTA with spectrum display...I think they rushed it and I haven't even updated to 1.6 yet because of that reason.

Now, hook up to a full blown version of SMAART, and offer up the corresponding knowledge of the use of SMAART and then we can talk, but SMAART is a VERY deep program and there are very few people that REALLY understand what they are looking at with the SMAART program itself. Dennis is one such person and Mikey is as close to an expert as your going to find on these forums.

There is a lot more to crossovers than you can imagine, and if you want to get into it you will ....be sorry( I was...)... I have a crossover cookbook, and I have tried and tried to understand it but the math goes so deep my brain starts to smoke and degenerate into gelatinous goo every time I try and understand how the numbers go, you may fair better...there are simpler explanations, some we have included here, but they really don't do crossovers justice...

If you really want to know, get SMAART, and take the classes.. you'll probably need MANY to simply be able to interpret the information provided by the software...otherwise your shooting a shotgun in the dark trying to hit a bat...

Dennis

I have been following this thread and I am having difficulty understanding the issue(s) so I don't really have any answers. But I do have some info to share that may shed some light on the topic.

First is a diagram that compares LR filters to BW filters


It shows differences between the two.

Second is the anticipated summation when two filters are used together (high-pass and low-pass). If phase is aligned between the two components at the crossover point, all goes as planned but if there is no phase alignment you will have a "notch" at the xo point that filter summation will not cure. You will need to align by adding delay to one of the components. A measurement program such as SMAART is useful in doing this. As long as you know for a fact that you have correctly aligned the system and have a flat response through the xo point, you don't have to worry about stepping on 80Hz. (I use 80Hz as a xo point in several situations).

Trick

thanks for the graph...that helps explain some of the results I was getting.