First Graph / Hard Knee House Curve / Target Level / Target Line Questions

[myn]

Here are my intial measurements from my IXL18.4 build recently posted here

How's it look?

.


This second graph is with the Wayne's sample hard knee house curve applied.
30 6.0
35 4.4
40 3.1
45 2.0
50 1.1
60 -0.1
70 -0.6
80 -0.5
90 0.0


I've read through the really informative long article on minimal EQ, target levels and house hard Knee curves. I've got a couple of questions.

1) Upon applying a positive house curve the target line especially in the frequencies that need a bump to achieve a house curve are raised. In my case the target level when I took the measurements was 75.2db. The house curve represents a postive gain above the 75.2 baseline. I've heard much mention of not boosting frequncies in the eq. As the target line represented by the house curve is raised how are you supposed to eq without boosting those frequencies?


Out of curiosity sake I applied some EQ with lots of boost to match Wayne's sample hard knee curve above:




And the result of this EQ. Note that when I calibrated the target level after this filter was applied it was raised up to 77.6dbs (this target line now reflected in the image below). Should I have maintained the original 75.2db target level as was originally measured or should you (like I did) set the target level after each measurement with the filters applied.:



Any help/guidance/direction would be greatly appreciated.

[myn]

I also see on that long "hard knee house curve" thread you should use 1/3 smoothing while creating your eq filters to promote less filters. Yet I see many people are still using all 12 of their BFD filters to smooth. What is the final word here on usage of smoothing for eqing a subwoofer?

[Wayne A. Pflughaupt]

I've heard much mention of not boosting frequncies in the eq. As the target line represented by the house curve is raised how are you supposed to eq without boosting those frequencies?

The house curve target, whether it is derived from "positive" or "negative" values, can be manually adjusted under the "Target Settings" button in REW (left side of the screen) to a location appropriate for your response reading.

And the result of this EQ. Note that when I calibrated the target level after this filter was applied it was raised up to 77.6dbs (this target line now reflected in the image below). Should I have maintained the original 75.2db target level as was originally measured or should you (like I did) set the target level after each measurement with the filters applied.

In your case I think I would have moved the Target (with house curve file in place) to about 78 dB. That would have required less boost at 20 and 30 Hz, would have left you with flatter response all the way out to 10 Hz, instead of sagging below 20 Hz as it does now. Of course you can try it both ways and see which results in the best <20Hz sound quality...

I also see on that long "hard knee house curve" thread you should use 1/3 smoothing while creating your eq filters to promote less filters. Yet I see many people are still using all 12 of their BFD filters to smooth. What is the final word here on usage of smoothing for eqing a subwoofer?

Oh, I doubt there is anything such as a "final word" on this topic.

The 1/3-octave smoothing recommendation was only for when you let the REW program craft your filters. That was based on my observation that it obtained acceptable results with without the notch filters REW often generates. For manually creating filters, as it appears you're doing, I used 1/6-octave smoothing and a 20-130 dB window. That was also recommended to help reduce the inclination to over-equalize. It's not necessary to smooth your graphs unless you find you are inclined to over-equalize.

It's been a while since I've seen anyone trying to use all 12 filters, but you'll usually find that most of the people doing that are either novices, or they're trying to equalize based on waterfall graphs, which is a different approach than the response-smoothing method.

Regards,
Wayne

[myn]

I wanted to clarify something. I’ve seen many posts that state “minimize positive gain, only negate” and “Boosting is bad”.. etc.

Assuming you lowered the target line and negated to match a verbatim curve achieved by boosting (lets say a house curve) And then raised the attenuator on the subwoofer amplifier since the signal is now lower from all of the negative gain, how is this any different from the boosting positive gain scenario? Doesn’t the curve signature stay the same? If you are compensating by increasing the gain on your sub amplifier I don’t understand why boosting would be such a bad thing?

[brucek]

how is this any different from the boosting positive gain scenario?

Since any gain used in the BFD requires a lower input voltage, it reduces the number of bits you'll use in the BFD.

brucek

[myn]

Another thing,

In this thread thewire has applied the subwoofer calibration file to REW and he has his Radio Shack digital SPL meter's Weighting set to C. He states that he DID NOT tick the "C Weighted SPL Meter" box in the mic/meter settings within REW

The help files indicate:

"If using an SPL meter set it to C weighting and tick the C Weighted SPL Meter box in the Mic/Meter Settings. Set the meter range to suit the measurement level used in the check levels process (the 80dB range is recommended for the Radio Shack meter). If using a mic and preamp leave the C Weighted SPL Meter box unticked "

As I am measuring only the sub and have the SPL meter calibration file loaded should I have "C Weighted SPL Meter" box in the mic/meter settings within REW ticked or unticked

[brucek]

He states that he DID NOT tick the "C Weighted SPL Meter" box in the mic/meter settings within REW

Matters not, either way. REW uses the meter calibration file to its extremes.

brucek

[Wayne A. Pflughaupt]

myn wrote:

I wanted to clarify something. I’ve seen many posts that state “minimize positive gain, only negate” and “Boosting is bad”.. etc.

Assuming you lowered the target line and negated to match a verbatim curve achieved by boosting (lets say a house curve) And then raised the attenuator on the subwoofer amplifier since the signal is now lower from all of the negative gain, how is this any different from the boosting positive gain scenario? Doesn’t the curve signature stay the same?

That is essentially correct.

If you are compensating by increasing the gain on your sub amplifier I don’t understand why boosting would be such a bad thing?

Generally speaking, any gain boosting with an equalizer increases noise. Fortunately, noise is virtually a non-issue at low frequencies.

Regards,
Wayne

[myn]

Wayne A. Pflughaupt wrote:

In your case I think I would have moved the Target (with house curve file in place) to about 78 dB. That would have required less boost at 20 and 30 Hz, would have left you with flatter response all the way out to 10 Hz, instead of sagging below 20 Hz as it does now. Of course you can try it both ways and see which results in the best <20Hz sound quality...

I used 1/6-octave smoothing and a 20-130 dB window. That was also recommended to help reduce the inclination to over-equalize. It's not necessary to smooth your graphs unless you find you are inclined to over-equalize.

Regards,
Wayne

So I took your advice and created another filter set lowering the target line down around 72db's which resulted in less boosting in the lower frequencies.

In this next iteration I have 4 total filter EQ's, 1 more than I originally had:



And here are the new results:



Thoughts? Any suggestions or improvements?

[brucek]

Any suggestions

Let's see the waterfall..

Vertical = 45-105dB
Horizontal = 10Hz-200Hz
Logarithmic mode (important)
X=1, y=100, z=150.

brucek

[myn]

As requested, here is the Waterfall.

I'm brand new to waterfall charts. Hows it look?

[JohnM]

That filter at 43.5Hz is pretty broad, and the 31Hz filter is a bit narrow. You are generally better off applying narrow cuts and, if they are needed, broader boosts. A narrow boost causes a slow decay at the boosted frequency, it is in effect an artificial resonance. This may help show what is going on, the red/green/blue/yellow traces are the individual filters and the cyan area is the net effect relative to the target.

[myn]

Very interesting graph. I appreciate you mocking that up for me.

As a guideline, what should the boundaries be for narrow cuts and broader boosts? Whats a good number?

Any other suggestions would certainly be welcomed. Thanks again

[Wayne A. Pflughaupt]

As a guideline, what should the boundaries be for narrow cuts and broader boosts? Whats a good number?

Hopefully John will chime in with his recommendation, but personally I try to keep them at ~1/6-octave are broader (10/60 in BFD-speak).

Looking at your waterfall, though, I don't see any evidence that decay time has increased in the 31 Hz area (relative to areas above and below that point)...

Regards,
Wayne

[JohnM]

1/6th octave is reasonable as an absolute minimum for a boost, 1/3 octave is safer. Cuts can be as narrow as needed to match the peaks they are countering.