New sub, strange response...

**Dr. Spankenstein** · **08-22-09, 04:05 AM**

Just took ownership of an eD A7s-450 after coming from a Mirage Omni S8. Recently acquainted myself with REW and created a nice graph of the Mirage in my 21'x13'x9' room. 4 wide Q filters, positioned against the side wall adjacent to L main-firing across the room:

Unpacked the new sub and took a measurement after trying a couple locations:

Ran Audyssey, dropped 5 filters on it and came up with this:

Tried boosting those troughs...could this sub (in the same location as the Mirage was) be helped by placement?

Thanks,

Bryan

**Wayne A. Pflughaupt** · **08-22-09, 09:24 PM**

Hmm, that first graph looks like it has some serious smoothing applied. Typically you only apply smoothing to full-range measurements.

The other graphs, you need to re-do them with the Frequency axis switched to logarithmic, and the vertical graph parameter to 45-105 dB. The bad news is, they'll probably look even worse than they do now...

But to answer your question, yes repositioning can often improve response considerably. Or at least get you something that's easily equalized.

Regards,
Wayne

**Dr. Spankenstein** · **08-23-09, 05:56 PM**

I thought mine were looking strange. I forgot to view in logarithmic form...
Here is my graph of the new sub but in compliant form:

Now with a slight house curve and 5 filters:

Can anyone see where my house curve is wrong? All the others I see drop below 48dB before 120Hz. Is mine not steep enough? Oddly, 30Hz sine wave was the same SPL as 80Hz so I was unsure about how many dB my house curve should be. Also, is it true that I don't need to be concerned with that wide peak above my 80Hz crossover frequency. Oddly, my receiver doesn't seem to be applying the crossover while using the subwoofer multichannel input.

I was thinking I should be doing better on extention with this sealed 18 incher...
The last graph was with a 10dB boost @ 20Hz.

**Wayne A. Pflughaupt** · **08-23-09, 09:25 PM**

Wow, I had it backwards - the graphs look much, much better with the correct scaling.

As far as your house curve target on the graph, first make sure you have REW set for the correct crossover frequency and slope that your receiver has. Yours looks like the frequency is high and/or the slope is only 12 dB/octave. Of course, if that’s correct for your receiver, then it is what it is...

Quote:

Oddly, 30Hz sine wave was the same SPL as 80Hz so I was unsure about how many dB my house curve should be.

If those two sine waves are reading the same SPL, then you don’t have a house curve, you have flat response. As it is now, I’ll bet that 80 Hz seems louder than 30 Hz? If that’s the case, increase the volume of the 30 Hz tone until it sounds like it’s as loud as 80 Hz. Naturally then, 30 Hz will be a higher SPL reading. The difference between the 30 and 80 Hz SPL readings is what your house curve slope should be (or at least it’s a good place to start). Create a house curve text file, and if you want the hard-knee house curve, check the “logarithmic interpolation” box under the House Curve tab (“Settings”).

As far as your crossover, if you just use the regular inputs (“Aux,” “CD,” etc.) you should be fine, as long as you have your speakers set to “small” and subwoofer “on” in the menu. Just unplug the speakers for a subwoofer reading.

Regards,
Wayne

**Dr. Spankenstein** · **08-23-09, 10:19 PM**

Well, I'm just using the default setting of 80Hz crossover & 24dB/octave, can't determine why my target curve is up so high.

I'll take another listen to those sine waves and come up with a proper house curve dB differential.

Thanks for the input and replies!

**Dr. Spankenstein** · **08-25-09, 10:24 AM**

Found a better position for the sub. Rear left corner firing @ 45 degree angle into room.

I may have gone with too strong of a house curve...
Listening to the 30Hz and 80Hz sine waves seemed like it required a 10dB boost to sound the same level.

The following graph was what was obtained using 3 filters. 30Hz null and 49Hz peak just won't go away.

Listening to a couple of spots in Kung Fu Panda made the overdone low-end obvious! Even though I calibrated to 80dB, there was so much low-end energy that even at -42dB in the receiver, the walls were shaking when the frequency dropped (skidoosh). It was frightening..

So back to (correctly) determining my house curve...maybe I should just run it flat? Could that potentially net me more extension?

**Wayne A. Pflughaupt** · **08-25-09, 11:23 AM**

Looks good. If the 49 Hz peak won't go away, it's probably your main channels. If you're running them full range, change the setting in the receiver to "small."

Do some music listening before you decide of your house curve is too steep. If everything seems overly bottom-heavy, lacking in upper bass, etc. then dial the curve back and re-EQ.

Even as it is, the 10 dB house curve has pushed up your Target Curve pretty high. You might want try lowering it down to something between 75-80 dB, especially if you're having to use boosted filters to push everything up to it.

Regards,
Wayne

**Dr. Spankenstein** · **08-25-09, 12:03 PM**

Quote:

Wayne A. Pflughaupt wrote:

Looks good. If the 49 Hz peak won't go away, it's probably your main channels. If you're running them full range, change the setting in the receiver to "small."

Hmmm, this is the sub alone.

Quote:

Do some music listening before you decide of your house curve is too steep. If everything seems overly bottom-heavy, lacking in upper bass, etc. then dial the curve back and re-EQ.

I will take a listen. There didn't appear to be any weakness in the mid-bass area, I was just concerned at the amount of ULF coming through even when the receiver volume was drastically reduced (-41dB).

Quote:

Even as it is, the 10 dB house curve has pushed up your Target Curve pretty high. You might want try lowering it down to something between 75-80 dB, especially if you're having to use boosted filters to push everything up to it.

No real boosts applied. All cuts (as suggested by numerous folks!) In fact the 51Hz region got a -25dB cut with a 1/6 octave filter! Just adjusted the sub's volume to bring the level back up to 80dB with pink noise.
*light bulb* do I need to utilize another test tone to calibrate? Full-band pink noise?
I just can't wrap my head around why it sounds well balanced during moderate low thumps and then when the frequency really drops-I find myself in a sonic blender!

**brucek** · **08-25-09, 12:19 PM**

Quote:

Hmmm, this is the sub alone.

We can't tell that since you aren't posting your graphs with the axis we ask of everyone for subs.
Vertical = 45dB-105dB
Horizontal = 15Hz-200Hz.

Quote:

I was just concerned at the amount of ULF coming through

ULF?

Quote:

In fact the 51Hz region got a -25dB cut with a 1/6 octave filter!

I would remove that filter - far too much cut.

Quote:

when the frequency really drops-I find myself in a sonic blender!

Can you explain what this means?

brucek

**Dr. Spankenstein** · **08-25-09, 12:33 PM**

My apologies. I will have to go back and get new JPEGs with the correct scaling. I thought you were just concerned with the vertical scaling. Sorry.

ULF=Ultra Low Frequency. Forgive my cryptic acronyms.

I took the sub on a tour of the room last night. The location mentioned had the smallest peak at that frequency. Every position and orientation required a MAJOR cut there. (Must be the wood panelling and the glazed brick wall that is causing all the gain.

The blender comment was an attempt to explain the curious and sudden pressurization that happens when the signal's frequency falls into the upper teens. The sound seems balanced during normal passages.

**brucek** · **08-25-09, 01:13 PM**

Quote:

explain the curious and sudden pressurization that happens when the signal's frequency falls into the upper teens.

Your response graph doesn't reveal that problem, so the trouble is likely in the time domain.

Lets see a waterfall plot (using the same axis as the response graphs - and be sure it's log).

It will probably show lots of ringing over the area you're concerned with.

brucek

**Dr. Spankenstein** · **08-25-09, 01:30 PM**

That's probably true. I forgot to re-run Audyssey after I moved the sub.

I get to have tomorrow evening all to myself. I'll put together the required graphs and report back.

Thank you so much for the troubleshooting and suggestions!

**Wayne A. Pflughaupt** · **08-25-09, 05:38 PM**

Quote:

Dr. Spankenstein wrote:

Quote:

Wayne A. Pflughaupt wrote:

Looks good. If the 49 Hz peak won't go away, it's probably your main channels. If you're running them full range, change the setting in the receiver to "small."

Hmmm, this is the sub alone.

Interesting - the only time I’ve ever seen people having problems with peaks in the sub range that won’t EQ away, it’s always because they’re running their mains full range and since they have no EQ, there’s no way to get rid of the peaks they may be causing.

Quote:

No real boosts applied. All cuts (as suggested by numerous folks!) In fact the 51Hz region got a -25dB cut with a 1/6 octave filter!

Boosts are fine, if they’re needed, as long as the depressed area is not a null that does not respond to EQ (like the 30 Hz problem you mentioned).

I can’t imagine a peak being as severe as 25 dB or more. Either you need to reposition the sub, or you have something in the house (besides the sub) making noise at that frequency when you’re measuring. Perhaps a before-EQ graph is on order...

Regards,
Wayne

**Dr. Spankenstein** · **08-27-09, 12:32 PM**

Took the sub on another tour of the room (sans Audyssey, no EQ), no luck in making the 21db peak go away even when measured from a few different listening positions. It's just there. The room surfaces are so reflective that changing the sub's orientation at each position made little to no difference. the biggest differences from graph to graph is when I change the listening/measuring position. And I'm not talking "move a few feet over", I'm talking 8" one way or the other drastically changes the peaks and nulls (even after Audyssey.) Basically I'm listening in a echo chamber!

Attempted to add ~4dB of boost around 30Hz and that allowed me to reduce my amount of cut (at the 51Hz peak) to -11dB. Graph is looking pretty good (when measured from my 8"x8" sweet spot) now that I reduced it to a +6dB hard-knee house curve.

Waterfall graphs tell the story, sorry I don't have them on hand. But there is upwards of 800ms of decay from 60Hz on down. Guess the next route is room treatments.

Is anyone partial to SuperChunk traps vs DIY Tube traps?

**brucek** · **08-27-09, 01:55 PM**

Quote:

Is anyone partial to SuperChunk traps vs DIY Tube traps?

How much effect do you expect them to have in the under 60Hz bandwidth?

brucek

**Dr. Spankenstein** · **08-27-09, 02:04 PM**

Not much, just a precusor to some sonotube hemholtz resonators as I doubt it would sound very satisfying even if I were to just tame the 52Hz peak.

**Wayne A. Pflughaupt** · **08-27-09, 02:17 PM**

I was a bit off in my earlier post. You indicated that the peak was really severe, not that you couldn't get it to go away. In the scenario I mentioned with the main speakers, EQing the sub to eliminate a peak wouldn’t register at all, because the mains were still generating the peak. Sorry if I mislead you.

Quote:

Dr. Spankenstein wrote:

Took the sub on another tour of the room (sans Audyssey, no EQ), no luck in making the 21db peak go away even when measured from a few different listening positions. It's just there. The room surfaces are so reflective that changing the sub's orientation at each position made little to no difference. the biggest differences from graph to graph is when I change the listening/measuring position. And I'm not talking "move a few feet over", I'm talking 8" one way or the other drastically changes the peaks and nulls (even after Audyssey.) Basically I'm listening in a echo chamber!

I see a lot there about moving the measurement or listening locations, but nothing about trying the sub in different locations. Often a horrendous peak is the result of a really poor location.

Quote:

Waterfall graphs tell the story, sorry I don't have them on hand. But there is upwards of 800ms of decay from 60Hz on down. Guess the next route is room treatments.

The “severity” of decay times a waterfall displays is as dependant on the SPL level you’re measuring at as it is the actual ringing in the room. For instance, a measurement taken at 85 dB will look far worse than an identical one taken at 75 dB, even though the rate of decay will obviously be the same. 800 ms would be pretty unusual for everything below 60 Hz, especially at a reasonable SPL level. Below 30-40 Hz, maybe, but not 60 Hz.

Regards,
Wayne

**brucek** · **08-27-09, 02:29 PM**

Quote:

For instance, a measurement taken at 85 dB will look far worse than an identical one taken at 75 dB, even though the rate of decay will obviously be the same.

Yeah agreed, it's important to ensure a 75dB target was used when evaluating a waterfall.

brucek

**Dr. Spankenstein** · **08-27-09, 02:42 PM**

Quote:

Dr. Spankenstein wrote:

Took the sub on another tour of the room...

Quote:

...changing the sub's orientation at each position made little to no difference.

These are both references to my attempts to remove the peak via sub position and firing orientation. I must need work on my communication skills...

Quote:

Master Wayne wrote:

The “severity” of decay times a waterfall displays is as dependant on the SPL level you’re measuring at as it is the actual ringing in the room. For instance, a measurement taken at 85 dB will look far worse than an identical one taken at 75 dB, even though the rate of decay will obviously be the same. 800 ms would be pretty unusual for everything below 60 Hz, especially at a reasonable SPL level. Below 30-40 Hz, maybe, but not 60 Hz.

Maybe you can help me in this area. This also harkens back to my question about determining my house curve. When using REW, I set the levels by the VU meters in the "Settings" tab, calibrate the SPL meter to 75dB and my readings still come out around 88dB at 25-30Hz. Sweeps don't clip and I have around 9-4dB of headroom. Any ideas as to why?

**brucek** · **08-27-09, 02:52 PM**

Quote:

calibrate the SPL meter to 75dB and my readings still come out around 88dB at 25-30Hz.

Do you run the CALIBRATE routine after the Check Levels routine?

brucek

**Dr. Spankenstein** · **08-27-09, 03:33 PM**

Yes sir.

Check levels (via the "Use subwoofer tone to set levels" in the "Settings" tab) to set the SPL meter to 75dB (by adjusting the master volume on receiver), then adjust the level on the "recording device"/line-in so that the VU meter reads -12dB. Click "FINISH"
Click the "Calibrate" for SPL meter. Verfiy that the meter is reading ~75db (averages 76dB even on SLOW setting) and set the calibration number to 76dB. "OK".
Run a 4 loop sweep with default settings (0Hz to 200Hz, -12dBfs, 256k) and the result is still a graph that shows @ 88dB. This is verified after the sweep by running the "set target level" in the "Target Settings".

I don't get it? AVR setting for sub is -6dB and sub gain is at 12 o'clock....

Even when I use the receiver Test tones, they are still at the same level (although the mains & surrounds @ 75dB adn subwoofer to ~80dB.) I was told this was kosher as the Radio Shack meter reads low.

**brucek** · **08-27-09, 03:52 PM**

Quote:

calibrate the SPL meter to 75dB and my readings still come out around 88dB at 25-30Hz.

OK, you do have to realize that the Check Levels routine using the subwoofer pink noise as the test signal outputs a band limited pink noise with the low cut at 30Hz and a high cut at 80Hz. It's a compromise in an attempt to cover most situations.

If your system happened to output a lot of energy below 30Hz, then this wouldn't be considered in the levels setup since the low cut of the test signal is at 30Hz.

You can see the problem.....

brucek

**Dr. Spankenstein** · **08-27-09, 04:01 PM**

I honestly was not aware of the cutoffs for the band-limited pink noise. I am very new to ALL of this.

I can't express how much I appreciate your and Wayne's help and suggestions, but now that I have been made aware of this shortcoming, what would be a suggested solution?

**brucek** · **08-27-09, 04:20 PM**

Well, it's not really a problem. The Check Levels routine attempts to set a decent level for output and input. If you get a result in the green then that's fine.

When you examine the waterfall, it's very easy to look at slice #1 (of 30) and use the Left Panel 'Trace Offset' and 'Add Offset to Data' feature to position your response trace around the 75dB target and then generate an appropriate waterfall.

brucek

**Dr. Spankenstein** · **08-28-09, 11:22 AM**

Don't know if this will shed any light on my situation, but here are the graphs from the other night.
The positions and orientations might not mean much and not all are posted here (to respect your bandwidth), but they are in order:
1.) Rear left corner-firing along long wall
2.) Rear left corner-firing into corner @ 45 degree
3.) Rear left corner-firing into rear wall from 24" away
4.) Rear left corner-firing 45 degrees into room
5.) Front right corner-firing 45 degrees into room
6.) Front right corner-firing across width of room
7.) Front right corner-firing into front wall
8.) Front right corner-firing into corner @ 45 degrees

Two places I haven't tried are the front or rear center of room width and midway along a side wall (both because of current obstructions-couch and table.

Here are the graphs I get from the main LP (utlilizing the right front corner firing 45 degress into room). Each one varies from the center position by a mere 8".
In order they are Left, Center, Right:

Again, it's strange that the target level is reading so high as I've taken great pains to calibrate to 75dB.

Any comments or suggestions?

Advertisement