PB12+ Room Response

[daxie]

This is my room response with a PB12+... Measurements were made with sub+mains.

First place I put it had a 35 wide canyon dip between 20-70Hz, too bad I didn't save that one...

It's going pretty straight but dropping pretty steep until 40Hz, but then the misery starts :-).

What would you guys suggest I'd do? Using the different tunings of my PB12+ I can shift the peak @16 Hz to 20Hz, but it remains the same for the rest...

I have also included the mdat file of my measurements, in case somebody wants to play around :-) (click here

(DSP1124P, SR7500, PB12+)

Thank you

daxie

[daxie]

Playing around with the filters in REW I get the following...

filters used:

41,50 - +1dB BW/60: 4
51,30 - +5dB 2
55,85 - -5dB 4
59,75 - -3dB 3
67,25 - +8dB 5

What are your thoughts on this, especially the fact I have to boost a bit? (especially the +5dB and +8dB boost)

(I can adjust the gain of the sub a bit to get it all 2dB higher)

Thanks

[brucek]

Measurements were made with sub+mains.

This is a problem.

It's impossible to really see what a sub is doing when the mains are attached. Subs are equalized first by themselves and then the mains are added to check for level and interaction around the crossover.

Can you redo the graph with the sub only?

brucek

[daxie]

Yes, I can . Will disconnect the mains.... However, since you are listening to both main + subs, istn't it more logic to measure them both? Or am I wrong?

(but not immediately though, don't have the measuring equipment this weekend)

Btw, crossover of the receiver is set at 80 now.

[lovingdvd]

daxie wrote:

Yes, I can . Will disconnect the mains.... However, since you are listening to both main + subs, istn't it more logic to measure them both? Or am I wrong?

(but not immediately though, don't have the measuring equipment this weekend)

Btw, crossover of the receiver is set at 80 now.

Yes it does seem logical what you are saying and likewise I was confused by this myself. My thought was why bother just doing the sub when ultimately you need to do the mains+sub.

But after playing with it a bit it seems that you have better control to first get your sub tamed. Then add in the mains and see what additional tweaks are needed.

The one thing I am still confused about is that I keep reading remarks about how the main and sub interact at the crossover point. However in my case they interact far lower than just a the xo point.

For instance my mains are rated to 32dB and even though I have the xo set to 80hz and the mains set to Small, I still get say 30-40dB of bass below even say 60hz when using 75dB as a reference. I don't quite understand this but apparently the slope is a lot slower than I would have thought. For example I would think (for no particular reason really) that with a xo of 80hz there'd be nothing much coming out below 60hz of my mains, but this is far from the case.

[cyberbri]

If the xo slope is 12dB per octave and crossed at 80Hz, if you send an 85dB sweep going down, the signal would start to lower in volume a bit higher than 80Hz, and then by 40Hz it would be down 12dB, or 73dB. If the slope is steeper, like 24dB per octave, it would be down to 61dB at 40Hz. This is assuming no room gain/interaction, so what the speaker is outputting before it gets shaped by the room. But the slopes are to make the speakers blend together, just like crossovers in regular speakers that blend the tweeters and drivers. You wouldn't want a brick wall cut-off between sub and speakers.

Personally I've found that even though the mains still play below the xo point, the main place where the bass interacts between the sub and the mains is above the xo point, the 80-160Hz range. That's where the biggest differences are noticed when checking phase, etc. Of course if the mains are more capable and play strong lower, and/or if they are closer to the front wall (reinforcement/gain), then there could be more interaction there than I notice in my situation. I have Ascend 340SEs, pretty far from the walls, with a VTF-3 MK2 sub.

[DrWho]

How far apart are your mains and sub? Are the mains set to small?

If you look at the Energy Time Curves you can see that you've got an early reflection from a boundary at a distance of about 4 feet from the sub. I'd be willing to bet that your mic was about 4 feet in the air when you took the measurement. If you move the mic around, you should see the 65Hz dip shift around in frequency (and its subsequent harmonics). In other words, don't try to boost the null.

As far as your low frequency response...I get much more normal results when adjusting the windowing of the impulse response. With a longer window you are introducing more noise into the calculation, making it seem like there is more bass than there really is. It also looks like your measurement was taken at a very low volume? You've only got a 40dB noise floor and your input was at -60dB.

Anyways, this is the frequency response I see with the file you provided:
untitled.jpg

Your waterfalls look really good too - like I wouldn't recommend changing anything with your setup - except maybe improving upon the crossover transition (but even that looks better than normal).

[brucek]

As far as your low frequency response...I get much more normal results when adjusting the windowing of the impulse response.

Yes, you will get a nicer looking graph, but it's not representative of the rooms response for the sub. Generally you want to adjust the window widths to include all of the response up to the point where it disappears into the noise of the impulse. This is why for a sub John chose 125ms as the pre ref window and 500ms as the post ref window for defaults. He also indicates that Tukey 0.25 is the best default for subs.

brucek

[daxie]

DrWho wrote:

How far apart are your mains and sub? Are the mains set to small?

If you look at the Energy Time Curves you can see that you've got an early reflection from a boundary at a distance of about 4 feet from the sub. I'd be willing to bet that your mic was about 4 feet in the air when you took the measurement. If you move the mic around, you should see the 65Hz dip shift around in frequency (and its subsequent harmonics). In other words, don't try to boost the null.

As far as your low frequency response...I get much more normal results when adjusting the windowing of the impulse response. With a longer window you are introducing more noise into the calculation, making it seem like there is more bass than there really is. It also looks like your measurement was taken at a very low volume? You've only got a 40dB noise floor and your input was at -60dB.

Anyways, this is the frequency response I see with the file you provided:
Attachment 1886

Your waterfalls look really good too - like I wouldn't recommend changing anything with your setup - except maybe improving upon the crossover transition (but even that looks better than normal).

Mains are set to small...
Sub is about 5 feet away from right speaker, it's a bit to the right and front of the right speaker...

4 feet high, euh, nice guess, indeed it was about 3-4 feet high :-).

So your advice would be not to change anything, just leave it as it is?

Thanks a lot for the help.

I will check into the Energy Time Curve and try to understand how you got to your conclusions. Is there anything I can read that will help me to understand them better?

(Btw, I have included some pics of my room setup for an impression (and to boast ))




(The sub is placed about 1 feet further to the door now, and the tag was removed off course )


(I can move the pb12+ along the right wall from just next to the door up until the back corner - 1 feet (for the back speaker)

[Sonnie]

Daxie... you should post those pics in the gallery... nice...

[Malice]

With the PB12+ so close to your bed, how often do you feel the earth move?

[Wayne A. Pflughaupt]

lovingdvd wrote:

The one thing I am still confused about is that I keep reading remarks about how the main and sub interact at the crossover point. However in my case they interact far lower than just a the xo point.

What that refers to is response changing around the crossover point once the mains are added. Typically it’s a phase issue introduced by the crossover. Response deviations that appear with the addition of the mains can show up as far as an octave above or below the crossover frequency.

For instance my mains are rated to 32dB and even though I have the xo set to 80hz and the mains set to Small, I still get say 30-40dB of bass below even say 60hz when using 75dB as a reference. I don't quite understand this but apparently the slope is a lot slower than I would have thought. For example I would think (for no particular reason really) that with a xo of 80hz there'd be nothing much coming out below 60hz of my mains, but this is far from the case.

You have to keep in mind that the crossover’s slope is subject to the acoustical output of the speakers, as well as the frequency response of the source signal.

For instance, let’s say your speakers have substantial bass output and do a 6 dB/octave rise between 100 and 30 Hz. Your receiver’s high pass is rated at 24 dB/octave. So, what you end up with is a functioning roll-out of about 18 dB/octave. The same thing will happen with room modes that cause a peak in response at say, 40 Hz – instead of being more than 24 dB down (assuming a 100 Hz crossover), it might not be more than 12 dB down, or even less, if it’s a really bad peak.

The other issue is the exaggerated bass response that action movies have. An explosion that gives a 20-dB boost below say, 30 Hz is going to essentially reduce the crossover to near-flat response.

A good way to see if your crossover is working to spec is to play sine wave tones at the crossover frequency, and an octave lower, and measure both with a SPL meter. The lower-frequency sine should be in the neighborhood of 24 dB lower (or whatever your slope is).

Regards,
Wayne

[lovingdvd]

Wayne A. Pflughaupt wrote:

A good way to see if your crossover is working to spec is to play sine wave tones at the crossover frequency, and an octave lower, and measure both with a SPL meter. The lower-frequency sine should be in the neighborhood of 24 dB lower (or whatever your slope is).

Regards,
Wayne

Thanks. Sorry to ask such a newb question, but how do you determine what is an octave higher and lower than something? I frequently here people say things like "set your xo to be one octave higher than your rated lowest frequency of the speakers" but not sure how to calculate that.

[cyberbri]

An octave is halving/doubling of frequency. So an octave above 80Hz would be 160Hz, and an octave below 80Hz would be 40Hz.

But 80Hz is a good starting point for crossovers, unless the speakers can't do 80Hz. But just because speakers only go to 60Hz, you don't have to set the crossover to 120Hz. 80Hz would likely work fine in that case as well.

[lovingdvd]

cyberbri wrote:

An octave is halving/doubling of frequency. So an octave above 80Hz would be 160Hz, and an octave below 80Hz would be 40Hz.

But 80Hz is a good starting point for crossovers, unless the speakers can't do 80Hz. But just because speakers only go to 60Hz, you don't have to set the crossover to 120Hz. 80Hz would likely work fine in that case as well.

Boy that was even a simpler explanation than I expected. Thanks. I will try some tests with a 60hz xo and see how it goes. Although with the house curve I'm planning that doesn't leave a lot of room for the sub to work (mainly would be pushing out 20-30hz strong and then sloping to 60hz quickly as the xo comes...

[DrWho]

brucek wrote:

Yes, you will get a nicer looking graph, but it's not representative of the rooms response for the sub.

Well, it's more representative than the default windowing.

Generally you want to adjust the window widths to include all of the response up to the point where it disappears into the noise of the impulse.

And if you look at the windowing of this particular impulse response you'll notice that I did just that.

This is why for a sub John chose 125ms as the pre ref window and 500ms as the post ref window for defaults. He also indicates that Tukey 0.25 is the best default for subs.

There is no way anyone can choose a proper window without first interpreting the data. Even with the same speakers no two rooms are going to have the same impulse response, nor the same noise floor.

The Tukey 0.25 is going to show you more of what the sum of the speaker + room is doing, where something like Blackman-Harris is going to show less of what the room is doing. Depending on what problems you're looking to deal with and what the Impulse Response looks like will determine what kind of windowing provides the most useful data. There are no absolutes in this regard, so I'm sure there are plenty of opinions with different ways to approach the situation. In other words, conclusions about the measurements must take into account the windowing for the conclusions to have meaning.

[DrWho]

Is there anything I can read that will help me to understand them better?

Yes of course - I'll see if I can't hunt down any of the literature.

Btw, sound travels roughly 1 foot every millisecond - so if you see an 8ms delay in the ETC, then the reflecting surface was probably around 4 feet away (resulting in a total of 8 feet extra distance travelled, or 8ms delay). Zoom in on the ETC and you should notice a spike around 8ms. In fact, just about every spike you see should correlate to some dimension in your room. And then when you calculate the wavelengths that result in 180 degree phase shifts from the difference in distance, you should find a resulting dip in the frequency response.

http://www.peavey.com/support/techno...ncellation.cfm

[lovingdvd]

All this talk about being able to get a graphical view/ representation of how your room surfaces are affecting sound and bouncing sound around is very interesting. However I'm unfamiliar with what this is all about and what features in REW can be used to see this. Is there a tutorial somewhere on this?

[DrWho]

http://www.prosoundweb.com/install/s...2/interact.php

[daxie]

thanks!

I will read into this later, all very interesting!


daxie

[DrWho]

I found another short article that gives a broad overview of what windowing does and the different types with pictures of the time and frequency response:
http://www.siasoft.com/pdf/tecnote3.pdf

And here's another that is talking about interpreting the impulse response:
http://support.siasoft.com/Downloads...udy2/case2.pdf
I freaked out when reading this article until I read the footnote at the very bottom.

Sound System Engineering is a great book if you want to get more into the acoustics side of things:
http://www.soundsystemengineering.com/

[JohnM]

Window shape selection for impulse responses is somewhat different than for the analysis of a continuous signal, as one typically encounters in discussions of windowing. With continuous signals of which only a portion is sampled for analysis it is essential to remove the discontinuities that occur at the edges of the sampled section. The shape of the window chosen for that will affect the trade-off between frequency resolution, sideband suppression and amplitude accuracy.

Impulse responses are different, as they have a finite duration. Our aim when analysing the full impulse response is to include everything to the point the impulse disappears into the noise floor, then window out the remaining noisy parts of the response as they will only degrade the result. Including all that is relevant and getting the best resolution from it is best achieved by a window shape which is flat at 1.0 through the region we are intersted in and fades out after that, which is why the Tukey windows are the REW defaults.

The default durations for the left and right windows are reasonable starting points for room response analysis, but to get the best results it is essential to look at the impulse response and choose the window durations according to the points the response reaches the noise floor. Use the dB FS Y scale when doing this, as on a linear scale the response will have dropped below the point where it is visible long before it actually reaches the noise floor. Windows that are longer than they should be include more noise than they should, so the frequency responses can take on a ragged appearance. Windows which are too short typically cause some ripples at the lowest frequencies and have reduced frequency resolution.

The situation changes when we only want to analyse part of the impulse response. In REW this happens when generating the waterfall and LF decay plots, which analyse successive portions of the response, moving along it to make each new slice of the response. Short windows are also used to exclude the effects of the room by analysing only the part of the response that occurs before the first reflections are seen. In these cases the tradeoffs are like those made when analysing continuous signals, and the choice of window needs to be made bearing in mind the characteristics of the various shapes.