Please help me eq/place my sub - plots inside!

Kincade · 06-11-2004, 03:15 PM

So i've taken readings of the BP1503 based sub I built here:

http://www.hometheaterforum.com/htfo...hreadid=200777

Below I have the plots. I took readings at the 3 main listening positions (the middle however is the MAIN position) with the sub placed both at the right front and left front corners. All the measurements are corrected using the excel spreadsheet on the BFD guide (snapbug.ws).

At Wayne F's suggestion in the other snapbug thread, I took measurements WITH my mains on, as I like to listen to music as well.

These 3 graphs represent the sub placed at the right front corner, and illustrate listening from the couches and recliner, oriented exactly as I listed in the diagram below:

These 3 graphs represent the sub placed at the left front corner, and illustrate listening from the couches and recliner, oriented exactly as I listed in the diagram below:

This is the room orientation:

So, which location is a better starting point for eq'ing? I haven't touched the BFD yet, because i'd like to get placement correct first.

Thanks for any advice!

Wayne A. Pflughaupt · 06-11-2004, 07:37 PM

I’m surprised to see such a difference between the right and left locations in that room – I guess that entryway makes more of an impact than you might think.

The question (which is the best corner) is really academic, because the BFD can equalize any of these curves. However, from the standpoint of which would be easiest (i.e., require the fewest filters) the left corner looks better.

On the other hand, since the left couch extends so far to the front of the room, keeping the sub on the right will keep it further away from viewers – might be less prone to overwhelm.

One think to keep in mind, Kincade: It will be impossible to get perfect response for all seats in this room. This is common in rooms that are essentially rectangular. I’d EQ for the center sweet spot, since that’s where you’d be sitting for music. “Perfect” response is more important for music than movies.

However, if this is primarily a movie room, you might want to consider an averaging curve that doesn’t give perfect response to any particular seat, but improves from what you would have with no equalization.

Regards,
Wayne A. Pflughaupt

Kincade · 06-11-2004, 07:45 PM

Wayne,

I think that the little entryway is exactly what is changing it so much left to right. That's the primary reason I decided to try both corners...

I'd like to eq for the center spot, as that's where I do my listening on multichannel music.

So you think the left side looks easier to EQ? I was thinking the left side (bottom graph) looked much more peaky on the left and right positions, and only slightly flatter in the center position.

FWIW, i'm taking reading with my mains on as well - per your suggestion in an earlier BFD thread.

Kincade · 06-12-2004, 03:58 PM

Well, I did another plot today, both from the main listening position; left and right. This time instead of using the autosound 2000 CD's, I downloaded the sine waves from snapbug.

The result is much different, and how that's possible, I don't know.

Blue is sub on left, pink is sub on right.

So who knows... I think I'm just going to leave it on the right, and try to EQ that.

Also, in the BFD guide it says this:

"Something that you should remember to do when you are setting up your BFD is add a foot to the value you enter for sub distance in your pre/pro or receiver set up. The 1 msec DSP processing delay in the BFD would account for approximately a foot in distance. If you add a foot to the distance you tell your pre/pro or receiver that your sub is from your ears, then it will advance the sound the 1 msec required. This will compensate for the BFD's delay."

However, I do not have a delay setting for my sub. Only the center, side, and rear surrounds. How do I compensate for the delay?

If you have any insight as to what is going on here, please let me know!

Wayne A. Pflughaupt · 06-12-2004, 06:45 PM

The right one certainly looks good. Amazingly good. The fact that the low spot between 63-71Hz is so wide, and the fact that it wasn’t there when the sub was on the left, should indicate that it’s not a null. You’ll probably be able to equalize it away.

Quote:

However, I do not have a delay setting for my sub. Only the center, side, and rear surrounds. How do I compensate for the delay?

A continuously variable phase control would do the same thing. If you can’t do it electronically, the only alternative is physically moving it.

However, in your case, since your sub is so close to the mains, it’s really nothing to worry about. I have no delay on mine and I’ve never noticed any ill effects.

Regards,
Wayne A. Pflughaupt

**HTF Ads**

Ethan Winer · 06-13-2004, 11:36 AM

Kincade,

> I took readings at the 3 main listening positions <

Those 1/6 octave measurements hide most of the detail. I guarantee that the peaks are higher, and the nulls are much deeper, than what's shown in your graphs. You need to measure to at least 1 Hz resolution to really see the true response. I do not say this to discourage you! But you risk making things much worse when you start fiddling with the EQ for two reasons:

1) The exact center frequency of the peaks and nulls is surely different from what the graphs show. If you have a deviation that falls near, but not on, one of 1/6 octave frequencies, you'll be tweaking the wrong frequency. This can make the response worse, yet you'll never know it and the graph may even look perfectly flat.

2) Since the magnitude of the peaks and nulls is not what you think either, the amount of EQ compensation will be incorrect.

I regularly measure rooms (using the ETF program) where a peak and deep null both fall within the same 1/6 octave band. Tests that use 1/3 or 1/6 or even 1/12 octave bands average all the values within the band. So these tests can display a perfectly flat response even when there's a big peak and also a deep null.

--Ethan

Kincade · 06-13-2004, 12:49 PM

Ethan,

Thanks for the response. So is everyone that uses the BFD w/ the Snapbug worksheet acutally making things worse? It comes very highly recommended.

What do you recommend to measure with? That's the only worksheet i've been able to find that has correction values built into it. I do have the autosound 2000 CD w/ 1 Hz intervals from 10-98 Hz, but no way to measure them...

Wayne A. Pflughaupt · 06-14-2004, 01:12 AM

Kincade,

I have some disagreements with Ethan’s recommendations that you might want to consider.

Quote:

The exact center frequency of the peaks and nulls is surely different from what the graphs show. If you have a deviation that falls near, but not on, one of 1/6 octave frequencies, you'll be tweaking the wrong frequency. This can make the response worse, yet you'll never know it and the graph may even look perfectly flat.

I won’t dispute that you will not get perfect accuracy with 1/6-octave readings. However, more people than you can count on this and other Forums have used 1/6-octave readings to equalize their sub and have been very pleased with the results.

Granted, as Ethan indicates, it has to be true that most of them have not actually achieved the perfectly optimal response that they imagine. Yet, if you do a search on this Forum and the Speakers and Subwoofers Forum you’ll find their testimonials about how smooth response is, with increased low frequency detail and no more “one note” peakiness, etc. That doesn’t sound like they “made their response worse” because they didn’t nail the exact frequency where the null or peak was, does it?

Ethan fails to keep in mind the difference between musical octaves and individual frequencies. For instance, there are only 12 musical notes between any two octaves. But if you’re talking individual frequencies, the number (given as whole numbers, at least) doubles or halves with each higher or lower octave.

For instance, taking per-frequency readings, there would be twenty readings between 20 and 40Hz, and forty between 40 and 80Hz. If you carry this to the ultimate extreme, you would be taking 10,000 readings between 10-20kHz – even though there is only 12 notes up there, too! Certainly, being concerned with 1Hz, per-frequency readings is the sonic equivalent of shredded cheese.

If each octave has the same number of musical notes, doesn’t it make sense that you would want the same number of samples for each octave? I mean, if you really need forty samples between 40-80Hz, shouldn’t you also have forty samples between 20-40Hz, not just twenty?

As you can see, it really is not logical to deal with per-frequency readings when the number of samples changes with each octave. This is why properly-executed equalization seeks to achieve smooth octave-to-octave response, not frequency-to-frequency.

Quote:

I regularly measure rooms (using the ETF program) where a peak and deep null both fall within the same 1/6 octave band.

Again, I won’t contest that you can have a peak and null within the same 1/6-octave band. But what Ethan fails to consider is that nulls and peaks that narrow and/or that close together are virtually inaudible in the bass frequencies. This is because ultra-narrow problems like that are easily masked by the program material and the adjacent frequencies.

Just to give you an example of how difficult it is to hear narrow low frequency problems, I have a 1/6-octave peak in my response at 45Hz (i.e., it shows up there with sine-wave tones, but not at 40 or 50Hz) that I can’t address because I’m using 1/3-octave equalizers. I forget how bad it is – probably several dB. The test track I use to verify good bass response and proper house curve (after all the reading, charting and tweaking) has a bass line that runs up and down the neck of the instrument, from the lowest notes to high notes. All notes are at the same volume; all are heard equally well. Furthmore, I’ve never noticed a “hot spot” with any other musical recordings, either. In other words, while my response does not measure perfectly, it nevertheless sounds perfect. There is no audible evidence of that 1/6-octave peak.

There are other, more practical reasons why to not be concerned with a hundred per-frequency readings.

As Ethan alluded, you will surely find that the individual readings will be all over the map. It will be common to find deviations of several dB between each frequency, or deviations only a few frequencies wide. I haven’t dealt with per-frequency readings before, but people have sent me every-other-frequency readings to analyze (i.e., even-numbered readings), and this is what I’ve seen. You’ll drive yourself crazy trying to deal with all those tiny aberrations and in doing so the tendency will be to over-equalize. You’ll probably “burn” all 24 of the BFD’s available filters, most of them set to ridiculously narrow bandwidths like 1/20 or 1/40-octave trying to “fix” inaudible “problems.” This is not an efficient way to use an equalizer.

Bottom line, Kincade, as countless dozens have done before you, you can successfully equalize your sub using only 1/6-octave readings, and you’ll be happy with the results. If you really want to go to the nth degree, you can use 1/12-octave spacing. But there is no need to be concerned with 1Hz, per-frequency readings.

Regards,
Wayne A. Pflughaupt

Ethan Winer · 06-14-2004, 03:20 PM