^These “why” answers^ could be more important than just following the “obey” rules?
Wait, what?
I can't just give the answers, I have to show my work?
It's like 7th grade math all over again, LOL.
Maximum clean input signal.
I agree that -10db is probably better than -6db. Back in the day, the standard for recording was -12db, but I honestly don't know if that's been bumped up in recent years; it seems to be less of a standard to my ear, now. As I've said in other posts, I tend towards the conservative in these matters. I'm happy to hear differently from anyone here with knowledge.
So, for the OP: if you're setting up your system with a -3db reference tone, you're telling it to amplify based on that "expectation." But if your music is coming in at -12db, it's going to be a lot quieter than it "expects," leaving a lot of headroom on the table. Having
some headroom isn't a bad thing, for variations in recording standards, but you also want to have as strong an input signal as possible.
Or you can max it out and rely on your ear to hear where on your volume knob things start to distort. If you follow Andy's advice, you
never trust your ears.
My advice, as a recovering newbie, is to start safe and get comfortable with how your changes affect the sound before you decide where you're willing to take risks.
Your call...
Crossovers.
For clarity, I'd consider anything other than 4th order (24db slope)
exotic. There may be good reason in home audio speakers to go with shallower slopes, given that the disparate drivers are mere inches apart and the listening environment is more controlled; that's out of my experience, however.
But the reason to go with 24db slopes in general comes down to how phase is altered by the filter: Each 6db of slope added rotates the phase by 90 degrees, so a 12db slope will give you a 180 degree offset in the phase at the crossover frequency, giving you a cancellation (dip). 24db gives you phase coherence, and a smooth transition from one driver to another. Very important when the drivers can be physically separated by feet, not inches.
Now, you
can correct that 12db phase cancellation by rotating the phase of one of the drivers 180 degrees in some DSPs, but that has implications throughout the spectrum, and needlessly complicates the tune, introducing phase conflicts elsewhere.
There may be times where this is warranted, but those would likely be the exception, and best left to when a robust understanding of the basics is well ingrained.
As for a 48db roll-off, while it does keep everything in phase agreement, it's a hella steep drop. You're likely to have more discrete sounding drivers, with a sharper handoff from one to another. In this case, there's less wiggle room in choosing the right frequency, based on the acoustics you your environment.
There's a pretty good discussion of it in Hanatsu's thread on tuning with REW, which is a good read in it's entirety (
here). While I prefer Andy's method of tuning, Hanatsu's is really great for identifying where, acoustically, you should set your levels and x-overs.
Bottom line: a 24db slope will cover you for 99% of your scenarios, and there are plenty of other ways/settings to screw up your tune; eliminating this one is a no-brainer.
So, cuts not boosts.
Given an evenly leveled full-spectrum test tone (pink or whatever), peaks and dips in your audible output (what's coming out of your speakers) will be coming from:
- your equipment (EQ curve built into factory head, deficiencies in the amp design, distortion introduced by gains, etc)
- your listening environment (cabin gain, nulls, phase boost/cancellation, reflection)
There will undoubtedly be elements of both, but I'd say that the overwhelming majority of problems come into the second group: the interior of a car is the most absurd place imaginable to attempt to accurately reproduce a recording.
Peaks are the least problematic to correct, with no danger to your equipment if you go overboard with cuts; the worst you can do is make it sound lackluster (or cut to your maximum and still have a problematic peak. I have one around 300hz that refuses to be tamed).
But dips, particularly deep ones, are
most often the result of phase cancellation, where there's a reflection (or discrete waveform from another driver) in that frequency that is anti-phase to it (or to some degree). These types of nulls cannot be filled with boosting; any attempts to boost them will also boost the reflection, cancelling out any benefits at the expense of pouring more power into that freq. And lest we forget, a +3db boost requires double the power.
If you're starting with a maximum clean power input, boosting from there is going to "bring the noise," as the kids and Anthrax/PublicEnemy/Leo say. You can be clipping that frequency (distorting it), at the same time that it is still below your reference curve; read: can barely hear it, but it still sounds like ass. Better to just barely hear it, without the ass flavoring added.
I think that covers it?
As an amateur tuner, I'm happy to hear what I've gotten wrong. There are (symbolic) greybeards here that have forgotten more than I'll ever know about this, and I appreciate the opportunity to replace ignorance/misunderstanding with solid knowledge...
