On Minimal EQ, Target Levels, and a Hard-Knee House Curve (long)

[daryn]

I somehow missed this thread until the past weekend...

I despised a flat response, and was never really happy with a house curve but "tolerated" it because it solved my nasty resonance problems. After using Wayne's hard knee curve, I'm in heaven! I secretly re-eqed while my wife was off to an auction. I was listening to music when she returned home, and she said: "What did you do?! I can feel the bass so much better, but I don't have that dull painful throbbing in my forehead!". That's a ringing endorsement since she normally can't tell a difference even when she knows I've done something.

Thanks Wayne!

[Chrisbee]

Thankyou, Wayne.

Lots to think about and try on a wet Sunday afternoon. I have been adjusting REW target levels from the start to find a best fit to my response curves in REW. It seemed the obvious thing to do.

I am increasingly of the (humble) opinion that a housecurve should follow the equal loudness curves as closely as possible. The problem is that the curves change constantly with increasing and decreasing average SPLs. I really need a perfect bass transducer and an active digital equaliser which can match the equal loudness curves to average SPLs and my own hearing.

Here's my (near) perfect bass transducer using four fairly broad BFD filters. I think you'll agree that it matches the equal loudness curves nicely at average listening levels on music.



4 x Acoustic Elegance AEIB15s in an IB manifold. Galaxy 140 SPL meter as test microphone.
I'm hoping that adding four more drivers might give me a bit more useful extension below 12Hz. I'd hate to be missing anything.

[brucek]

After using Wayne's hard knee curve, I'm in heaven!

With the new version of REW (presently in beta testing), the need to manually create this type of house curve is obviated by a new logarithmic interpolation feature.

interpol.jpg

brucek

[Avus_M3]

More things to do when I finish my IB

Great info!!!

[publius]

While not identical, JBL uses a similar house curve.





source: http://www.jblsynthesis.com/technolo...x?Language=ENG

[tdamocles]

Wayne,

In the example above you are boosting some frequencies a large amount...I thought this was a no-no?

[publius]

tdamocles wrote:

Wayne,
In the example above you are boosting some frequencies a large amount...I thought this was a no-no?

he does not seem to be using large positive EQ in order to try to fill-in any large holes (modal suckouts in the frequency response). you are correct in that trying to fill-in such holes would be a losing strategy.

what he _is_ doing is raising up the target level to the midpoint between the peaks and valley's BEFORE applying EQ. the result of this is that some frequencies require significant boost. this is okay and the whole strategy has reduced the overall number of filters. too many filters was identified as being the culprit of degraded sound quality.

the confusion stems from the complicating factor that he has also changed the target response from flat with a gentle roll-off above 60hz (aka 6db house curve) to flat to 30hz, then decreasing to 100hz, then further decreasing after that (aka 6db hard-knee house curve).

the two methods for applying filters can be contrasted in this way:
OLD WAY:
1. apply a large number of filters in order to try to bring frequency response to the old target ideal (6db house curve).

PROBLEM WITH OLD WAY:
1. because achieving minimal variation from the target response often requires a large number of filters AND because a large number of filters degrades sound quality, the old way results in the enigma of systems that appear to measure well, while not sounding right.

NEW WAY:
1. establish the new "target response" as the midway point between the peaks and the valleys that you measure in your room. (use the 6db hard knee house curve as the target not the typical 6db house curve)
2. apply a small number of filters in order to achieve this target. do not worry about trying to get the response to match the target response perfectly. this is not required for good sound.

BENEFIT OF NEW WAY:
1. 6db hard knee house curve sounds better
2. using fewer parametric eq's sounds better

[Wayne A. Pflughaupt]

Wow publius, that was a geat synopsis - simple, eloquent, and to the point!

tdamocles,

To answer your question a bit more technically, it looks like you’ve dug up the typical knee-jerk reaction to filter boosting: Someone ponders the situation, does a little math and comes up with, "Oh wow, if you boost 10 dB at 430 Hz, your 200 watt amplifier will need 2000 watts at 430 Hz to pull that off!" (Or whatever the correct figure is).

Where did our mathematician-cum-audio-expert get the idea that an amplifier is running at maximum capacity anytime you have it turned on? Obviously, what’s not being considered is the real-world use of the HT system, which is probably never maxed out. Maxed-out systems sound bad - everyone knows it. They distort and their subs make rude noises. So people typically keep a stressed system turned down, or they upgrade to something with adequate power reserves.

Assuming the system has adequate headroom, the next thing our mathematician overlooks is "relative gain" (new term I just coined). He also has a fundamental misunderstanding about how filters work, and the way they alter electronic response.

Here’s what happens in the typical "cut only" approach to equalization, three filters in this case. The dotted line is predicted response after equalization; the lower dark green line with smooth curves is the equalizer's electronic response with those filters applied.

graph with three cut filters enhanced.JPG

As you can see in the predicted response line, overall subwoofer level after equalization is going to be greatly reduced, probably 8-10 dB in this case. Naturally, you're going to have to increase the sub's level to restore the lost gain:

re-adjusted graph with three cut filters enhanced.JPG

So, did we avoided demanding undue power from the amp by not boosting? Well, I dunno. Take a closer look at the equalizer's electronic response. Did we make three cuts...

three cutting filters - electronic response w -75 dB enhancement.JPG

...or did we make three boosts, one being a power-robbing shelving filter?

re-adjusted three cutting filters - electronic response w -75 dB enhancement.JPG

It should be plain to see that once levels are re-adjusted, downstream amplifiers and drivers will see gain-reduced filters no differently than boosted filters, because subtracted filters leave peaks between the cuts. Any frequency where response is peaking, more power will be demanded.

In my own installation, using the filters you inquired about, we have a somewhat reversed, yet similar situation. I know that makes no sense, but here's the electronic response of the eight filters I was initially using:

eight filters hard knee - electronic response reduced.jpg

As you can see, even using cutting filters almost exclusively, I ended up with a honkin' boost at ~53 Hz, and another one at 20 Hz after I re-adjusted the level.

Now let's take a look at the response of the four "hard knee" filters that have mostly-boosted values:

four filters hard knee - elctronic response reduced.jpg

Looks pretty scary, huh? I mean, boosted out the wazoo. But since this set of filters increased overall gain so much, I had to turn my sub's level down to get it back to where it was before.

Are you seeing what I mean about "relative gain?"

Also consider, look what the severely boosted 25 Hz filter in the second graph did to response. You can plainly see that below the peak, response is forced to roll off fairly rapidly. This is ultimately going to conserve more power as opposed to the cutting approach, which left me with a de-facto boosted shelving effect that flattened response below 20 Hz.

As I've been telling folks for years, there's no free lunch when it comes to equalizing: boosting or cutting, you're usually going to end up placing increased demands on your subwoofer amps and drivers, as either method leaves peaks in electronic response. You just have to make sure you have enough headroom going in - it's as simple as that.

There are other issues with boosted filters that concern the equalizer itself, such as the potential for added noise and distortion, but we'll save those topics for another day. Fortunately, their impact on subwoofers is minimal at best.

Regarding my installation specifically, the default level setting on the Chase RLC-1 I use to remote-control my subs reduces the signal level considerably before passing it on to the sub amp. Plus, my equalizers have a 25 Hz high pass filter engaged to roll out the lowest frequencies (my throw-down subs aren't doing much below that point anyway).

Here's some more reading on the subject, from a recent thread.

Regards,
Wayne