Gap in my response

[subby]

Hi

I just moved into a new house and my HT room is way bigger now then my previous. I have the SVS PB10. After weeks of work in the new hous I finally got some time to calibrate my sub yesterday. Unfortionatly I got a very weird measure result.





As you can see the graph peaks at 20Hz and at 25Hz it just drops down and comes back up only about 50Hz. I tried moving it around a bit in the front from left to right but it doesn't change much. for now the sub was only positioned in the front corner of the room. Maybe I have to try to position it behind or a couple of meters aside the listening position. The problem might be that my room is not only much bigger (8m x 5m), but as you the ceiling also goes downwards at the sides like a roof of house (see pic 3, it's a view from the side) and there are 4 windows so there is kinde of a niche on each window (see pic 3). Maybe the frequenties above 25Hz get lost in the window niche right next to the sub?

Do you have any tips where I can try to position my sub?
And is it a good idea to boost the gap of frequenties or will it stress the PB10 too much?

sorry for my poor english but it's not easy to explain.

[brucek]

Do you have any tips

You're using a rather exaggerated vertical scale in your graph that make the situation look worse than it is.

Post your graphs in the standard we normally use here at HTS.

Vertical scale = 45dB-105dB.
Horizontal scale = 15Hz-200Hz.
Be sure to show the target at the crossover frequency you are using.

brucek

[subby]

ok, I'll try to do another measure next weekend because I'm only home in the weekends.

[JBravo]

subby wrote:

ok, I'll try to do another measure next weekend because I'm only home in the weekends.

you shouldn't need to remeasure unless you didn't save your measurements. You can just reload them and change the graph settings, then re-save your graph and repost here.

[subby]

I left my graph at home. I only took the JPG with me. I'll change the settings over the weekend.

cu then

[Wayne A. Pflughaupt]

In the meantime:

And is it a good idea to boost the gap of frequenties or will it stress the PB10 too much?

You really can't stress that subwoofer because it has a built-in limiter. Boosting will reduce the amount of headroom you have, which means you won't be able to get as much SPL as you would otherwise. But let's wait and see what your new graphs looks like...

Regards,
Wayne

[subby]

Ok, did another measure today.




I found a picture of my room before we moved into the house. In the right corner where the sofa is located, behind the soccergame table, that's where my sub is now located. Just to give you an idea how my room is shaped. So I did a new measure of the same graph from last week and did the settings as you asked, but not much changed.



then I moved the sub to the window corner right next to the hall (see top shot scetch of my room at the top of this page). Out of this measurement came this graph. The gap is much less But I lose out a few frequenties at the bottom end.

So look like I have to choose. keep the sub's current location and have the 20Hz but with a gap after 25z, or have balanced line but lose the bottom end frequenties. Maybe I've come to the limits of the pb10 and my room is too big for this decent but entrie model. Either way, I was hoping to upgrade to the pc13 if my poor student budget allows it somewhere in April or so.

[brucek]

I'd keep the current location with the extended bottom end....

The dip (gap?) will not be as noticeable as a loss of bottom end...

brucek

[JBravo]

that original with your dip doesn't look that bad. If you have an eq, you could cap some of those peaks and smooth things out a little bit. But that sub looks like it is handling the room fine.

I would read up on adding a house curve next.

[subby]

Buknakyd wrote:

that original with your dip doesn't look that bad. If you have an eq, you could cap some of those peaks and smooth things out a little bit. But that sub looks like it is handling the room fine.

I would read up on adding a house curve next.

Where would you change the peaks?

I've heard of a house curve before, but don't have a clue of what it is or how to do it. :-p

[JBravo]

here are the links to Wayne's articles on house curves:

House Curve: What it is, why you need it, how to do it.

Minimal EQ, Target Levels, Hard-Knee House Curves.

[brucek]

Where would you change the peaks?

Simply lower them with a BFD. REW will recommend the filters to use..

brucek

[subby]

brucek wrote:

Simply lower them with a BFD. REW will recommend the filters to use..

brucek

I already tried, but it always says: "there are no peaks to be equalised" or something like that. I must admit that I haven't worked with REW for over a year so I probebly overlooked some steps. unfortionatly I also have very few time left these days as school projects are requiring more and more time as the hollidays near.

[Wayne A. Pflughaupt]

I already tried, but it always says: "there are no peaks to be equalised" or something like that.

That's because REW looks for peaks in relation to the Target Curve. Your first graph had the signal so low that it was all below the Target:





Your new graph has the levels up where they should be, and as you can see the peaks are now above the Target Curve. REW will recommend filters this time.





Regards,
Wayne

[subby]

I just tried the "assign filter" button and it still says "there are no peaks to be corrected".

[brucek]

Hopefully, you pressed FIND PEAKS first?

brucek

[subby]

Ok thanks, I set the filters. I only should do another measure to see if the result of the filters are really as good as REW predicts.

I don't want to seem lazy. But can you tell me in short what a house curve does and how I can do it? I really have very little time on my hands these days and I don't want to get it all to technical.

[tenzip]

The short, very simplified version of "Why do I need a house curve?" is:

The human ear is less sensitive to lower frequencies, so you need to reproduce them louder to make them seem as loud as the higher frequencies.

The even shorter version is:

It will sound better.

As for how, and a more detailed explanation, read the links that Buk gave in post #11 above.

[DrWho]

brucek wrote:

You're using a rather exaggerated vertical scale in your graph that make the situation look worse than it is.

Can you please explain how changing the vertical scale changes the data in the measurement?

Subby, what mic are you using for your measurement? Also, can you post some waterfall plots? Behavior like you're seeing is very typical of room modes - the most reliable band-aid will be moving your listening position and subwoofer.

[DrWho]

tenzip wrote:

The short, very simplified version of "Why do I need a house curve?" is:

The human ear is less sensitive to lower frequencies, so you need to reproduce them louder to make them seem as loud as the higher frequencies.

The even shorter version is:

It will sound better.

As for how, and a more detailed explanation, read the links that Buk gave in post #11 above.

I would argue that the non-linearity of human hearing is compensated for during the recording process...even in the most simplistic of recording techniques, you're still choosing the mic location with your ears.

I'm not saying that flat response after room gain is the ideal, but I do find that a "house curve" following the Fletcher Munson curves is going to be way way overexagerated. I think the ideal lies somewhere with the effects of the group delay from room gain...

[brucek]

brucek wrote:

You're using a rather exaggerated vertical scale in your graph that make the situation look worse than it is.

DrWho wrote:

Can you please explain how changing the vertical scale changes the data in the measurement?

It doesn't, those are your words. Re-read the sentence.

Graph scaling is a major cause of data misinterpretation, and is often used in many fields (and particularly advertising) as a tool to exaggerate or support an argument that the data alone doesn't.

A great many new members have little or no experience with graphs and require instruction on interpretation in a language they understand.
To that end we have decided to use a standard scale set for all REW graphs (unless someone wants to specifically point out a section of data).

The scale we chose is quite appropriate for the audio data it represents and has been chosen as the default in REW software. This allows everyone to represent their data in a way for easy comparison without confusion. You need to be aware that not everyone is as experienced as you obviously are in interpreting graphs.

Even with this policy, I have many examples where new members have posted inappropriate scaling for their graphs, and as a result have received incorrect advice from rather seasoned members who misinterpreted the graphs.

As an example, here are three graphs using the exact same REW mdat file, with difference scaling chosen. I have seen all these scales used here at HTS. I suspect that someone who is new to interpreting graph data might think the data is derived from different measurements.







Subby, I apologize for hi-jacking your thread. I thought the info might be useful.

brucek

[tenzip]

DrWho wrote:

I would argue that the non-linearity of human hearing is compensated for during the recording process...even in the most simplistic of recording techniques, you're still choosing the mic location with your ears.

I'm not saying that flat response after room gain is the ideal, but I do find that a "house curve" following the Fletcher Munson curves is going to be way way overexagerated. I think the ideal lies somewhere with the effects of the group delay from room gain...

Mike, I'm not going to argue any of your response, but I will say that lots of people seem to be very happy with house curves. I've never seen one that looked like an equal loudness curve or F-M curve.

If you can show us a better way, lead on, man, lead on!

[DrWho]

brucek wrote:

It doesn't, those are your words. Re-read the sentence.

I'm not saying that a standard for comparison is a bad thing - in fact, it's great for quick comparisons.

I brought it up because there is an underlying mindset that a "good graph" somehow validates the performance of a system. A perfectly flat frequency response can actually sound quite horrendous...it just means the frequency response is the wrong tool for identifying the problems. So what's the point in comparing against other plots?

Isn't the whole point of measuring to identify and then fix problems? If you're not going to fix anything, then what's the point? Extending from that, proper scaling is pretty much just zooming in on the problem so that it can be identified when it is fixed.

I guess I just don't see the sense in asking for less resolution when the original plot was scaled quite appropriately? What extra information is to be gained? Are you archiving plots or something?

[DrWho]

tenzip wrote:

Mike, I'm not going to argue any of your response, but I will say that lots of people seem to be very happy with house curves. I've never seen one that looked like an equal loudness curve or F-M curve.

If you can show us a better way, lead on, man, lead on!

Man, I just keep coming off negative in this thread...

Why try to validate house curves by referencing the F-M curves, when it is quite clear the house curves aren't addressing equal loudness? That was my original point...and I only commented on it because it can lead to future misinformation.

As far as providing a better way, it's way easier to be negative I believe there was actually a recent article in AES on this subject, but I haven't gotten around to finding it at work (bloody deadlines). I think it will actually end up being source material specific and therefore some compromise will always be required....so then it starts to become purely subjective.

[tenzip]

DrWho wrote:

Man, I just keep coming off negative in this thread...

Mmm, maybe 'firmly assertive' would be better.

Why try to validate house curves by referencing the F-M curves, when it is quite clear the house curves aren't addressing equal loudness? That was my original point...and I only commented on it because it can lead to future misinformation.

I am no expert, spewing misinformation from time to time is my lot, I fear. Maybe I should have engaged brain before applying fingers to keyboard, and let someone else answer the OP's question. I am willing to learn, and I have no doubt that you've forgotten more of the stuff from your school and work than I have time or energy to learn. That pesky life stuff just keeps 'asserting' itself.

As far as providing a better way, it's way easier to be negative I believe there was actually a recent article in AES on this subject, but I haven't gotten around to finding it at work (bloody deadlines). I think it will actually end up being source material specific and therefore some compromise will always be required....so then it starts to become purely subjective.

And isn't all life? Shoot. I hoped you had a silver bullet.
I recall your information was always very interesting and informative when I read the Klipsch forums, I haven't had time to stop by for quite a while, unfortunately.