First Curve - Am I doing this right?

[zorax2]

I've noticed the curves posted on Sonny's page usually begin around 85dB on the left side. I'm using REQ testing the subwoofer and after clicking on Set Target Levels, it sets a cutoff of 80 and a target of 73.7.

My graph shows that REW suggests cutting two peaks at 23.1 and 35.6 Hz and the anticipated response after equalization. I've got the MIDI cables set up with the BFD. Do I store these cuts and then re-equalize to fix the small peak at about 52 Hz and so on?

Additionally, the set input volume establishes the line in setting at .998 or occasionally 1.0 and then gives an error message. I wasn't able to get the soundcard calibration test of the left channel to work right as best as I could tell.

I've spent hours reading through the AVS Forum, Next Level Forum and here but in spite of generally feeling "tech-savvy", I find myself wondering if I'm doing things right. Your feedback is appreciated.

I've also read a bit about house curves and find myself confused there as well. Can a suggestion be made for a house curve based on the measured response shown here?

[zorax2]

I forgot to note that my first graph was done using a 1/6 octave sweep since that is what I saw posted on a lot of graphs and since I thought a slower sweep might be more accurate.

This graph is done using a regular 10 - 200 sweep and seems to have a lot of dips and valleys compared to the other graph. Which setting should I use for REW to set my BFD filters? Obviously - this graph requires a lot more filters and shows a ton of dips. Is there anything that can be done to fix dips?

[Sonnie]

Looks like you are headed in the right direction. You have pretty good extension down below 20hz too.

First I would suggest reading post #2 and #3 here. That will get us started.

Then show your graph with no suggested filters to start with.

For a house curve, it really depends on what you like but you could try a few and see if you like the sound. See this post for detailed instructions on how to load them and a few house curve files you can download to try out.

Many times what we'll do is get our intial measurement, let REW make suggestions for filters, but then we'll treak the settings to get it like we want.

[zorax2]

Ok Sonny - I've made the corrections for the graph. I thought I had done that but perhaps REW doesn't keep the changes as defaults when entering / exiting the program.

One problem I had is setting the width to 800. I'm doing this in my A/V area which has a 61" Samsung DLP that I'm using as my monitor at 1024 x 768. I'm being limited to a width of 632. I'm not sure whether this distorts the graph for comparison purposes. If you think it does distort, I could probably haul in an external monitor if necessary - or maybe John can comment if there is another workaround.

I hope I've saved the graph as you desired. Thanks for the help!!!!

[Sonnie]

That's fine like you have it. We mainly say 800 pixels wide so that we limit the storage space on the database.

If you wanna emal me your .mdat file I'll play around with it tonight and post the filters and graph to kinda show you (and others that may be interested) how we fiddle with it. Or you can post it as an attachment here.

[zorax2]

I also have a question relating to integrating house curves in REW. It appears to me that many of the graphs posted here are using the Excel spreadsheet rather than REW. Perhaps that is why the spl of the desired response starts in most cases around 85 dB rather than the 75 dB of those posted with REW.

When I loaded some of the parameters listed on this forum for house curves into REW, the desired house curves all started quite higher on a dB basis in REW effectively creating more spots below the desired curves (boosts needed) rather than response above the curve requiring cuts needed. For example, I tried the house curves in this thread: http://www.hometheatershack.com/foru...hread.php?t=37

It didn't seem to make sense. Are the curves entered differently for REW, or alternatively do I disable the Radio Shack meter calibration or instead try running REW at 75dB go to 85 dB?

[brucek]

I'm using REQ testing the subwoofer and after clicking on Set Target Levels, it sets a cutoff of 80 and a target of 73.7.

It's up to you to set the cutoff that you want - not REW. Set it to the crossover frequency of your processor/receiver.
The target will be approximately at the level you originally set when you adjust the A/V volume during the setting of the Measurement Level on the SPL meter.

Do I store these cuts and then re-equalize to fix the small peak at about 52 Hz and so on?

No, you accept and use the suggestions and then add your own by entering them into REW at that time to see how they look on the REW screen and then enter them into the BFD.
Now you can remeasure and tweak.

I wasn't able to get the soundcard calibration test of the left channel to work right

You did use a loopback short from line out to line in?

my first graph was done using a 1/6 octave sweep since that is what I saw posted on a lot of graphs and since I thought a slower sweep might be more accurate.

No, the full sweep from 10-200 is the most accurate.

Obviously - this graph requires a lot more filters and shows a ton of dips. Is there anything that can be done to fix dips?

Yep, you can use a filter with gain. But realize it eats headroom from your subs amplifier. Small gains are fine if they work. Some dips defy any repair as revealed by the fact they won't respond to gain.

You can also apply some trace smoothing to remove sharp spike dips that can usually be ignored anyway.

but perhaps REW doesn't keep the changes as defaults when entering / exiting the program

It stores them in the Axis button that can be applied any time. As soon as you measure outside the setting, the graph axis will change to accomodate. Simply press the Axis button to return your default.

Perhaps that is why the spl of the desired response starts in most cases around 85 dB rather than the 75 dB of those posted with REW

Best to use 80db since the SPL meter movement will be centered at 80.....

When I loaded some of the parameters listed on this forum for house curves into REW, the desired house curves all started quite higher on a dB basis in REW effectively creating more spots below the desired curves (boosts needed) rather than response above the curve requiring cuts needed

Just adjust your measurement level higher and then cut down to the target.

Are the curves entered differently for REW, or alternatively do I disable the Radio Shack meter calibration or instead try running REW at 75dB go to 85 dB?

Simply put a simple housecurve file into the REW directory. Call it housecurve.txt
A simple one that rises from 80Hz by 5dB until it hits 30Hz would have two entries in the file.
30 5.0
80 0.0
Start REW, select FILE and Load Housecurve. That's it. It has nothing to do with the RadioShack calibration file (other than both files are stored in the REW directory)....
Once the housecurve is loaded you'll see it in the target response.

brucek

[zorax2]

Thanks for clarification on so many things. I didn't realize that cutoff = crossover though it makes sense now.

It appears to me (correct me if I'm wrong) that once I import the house curve (30 5.0 and 80 0.0) into REW that I move it up or down within REW using the "Target" control on the lower right side until it intersects nicely (more peaks than dips) with my measured sub response. I then use the "Find Peaks" and "Assign Filters" to adjust my subs response to follow the house curve. Finally, I export the filters to the BFD and now adjust the volume control (not the trim control on receiver) on my subwoofer to equalize the sound level between speakers.

As far as the soundcard calibration goes - I'm thinking I'm following things to literally. The help guide states: "The default is to use the Right input, but either input can be selected on the SPL Meter panel. Both of the soundcard output channels carry the test signal, one (typically the Right) must be connected to an input channel on your AV processor or to the input of your equaliser." Therefore, when I got to the help page and all of the samples shown showed examples with the left channel I was a bit confused. In the end, I still haven't verified the response of my audio card calibration as when I click on the Left tab and do a sweep I get an inverse copy of what appears to be my MIC calibration file. Nevertheless, I believe the soundcard is accurate and working properly and my results are OK. (Yes I did a loopback - removed the initial subwoofer to line-in and line-out to AV reciever and used a cable to jumper between the RCA plugs on my USB audio card.)

Is the standard sweep more accurate as it picks up the reflected, transient sounds more quickly than the 1/6 octave sweep? Why do many people post the results of the 1/6 octave sweep instead of the standard sweep?

As far as the SPL, the 80 setting works nice with the needle centered. Again, I was following the REW directions more literally using 75 dB. I found that I had no problems with REW setting the input volume once I did this as well. Previously, the input volume would max out until I lowered the SPL level to 60 dB on the sound meter.

Thanks everyone for all of your help. I'm learning a lot and believe I'm on track to figuring all of this out.

[Sonnie]

It sounds like you are understanding everything pretty well. You'll be able to teach us before long.

I think a lot of people post that 1/6 octave smoothing so that the graph looks prettier... it smooths the response line and eliminates all those little jaggies. I typically prefer to see the full resolution sweep myself.


EDIT: I just noticed JohnM posted about something similar on your soundcard issue here. It might help you, but I think you are okay anyway.

[Sonnie]

Okay... I tinkered with your .mdat file a little. This is probably one of the tougher ones I've fiddled with. This is the extreme, you don't have to really go this far with the filters, but mainly this is to show you how much adjusting can be done.

Here's your original unfiltered response:



Here's a filtered response:



Here's what the filtered response looks like with 1/6 smoothing applied:



Here's the filters I used:



Here's that filter file:

Zorax2.req

Again, this is the extreme merely to show the potential of tweaking the filters.

Keep in mind too that most of the boosting is not really boosting, it is simply compensating for the excessive bandwidth cuts. Those boosts never exceed your original response. Only the area between 40-50hz actually has any real boost.

[zorax2]

Phenomenal! Sonnie, what is your approach to tweaking? I've been playing with the automatic filters on REW which does a good job but doesn't come close to the smoothness here. Do you have a systematic method that others could duplicate? Are you focusing all filters on frequencies below 100 Hz because of the crossover at 80 Hz? I wasn't sure how all the wacky spikes and dips higher than 100 Hz would effect overall sound. I'm assuming the AV receiver doesn't send a lot of information to the sub above 80 Hz - therefore those frequencies don't matter as much.

Thanks for demonstrating the power of the program!

[Sonnie]

There's really no set rule to accomplish it... I just start sticking filters in there. Sometimes I'll let the Wizard suggest filters but I limit it to finding it to finding peaks between 20hz -70hz so that it doesn't give me a bunch of suggestions above the x-over point. Then I'll tweak from its' suggestions until I get what I want.

As far as the response above your crossover point... I don't mess with it unless there's a huge peak out there for some reason. You shouldn't notice those in that response above when listening.

Keep in mind that if you connect the output of your soundcard to an input on your receiver and engage your x-over, the response you then measure will be reflective of what your system is really producing. What you see above the crossover point would then be what you will actually get with the x-over engaged.

[Ayreonaut]

Sonnie wrote:

There's really no set rule to accomplish it... I just start sticking filters in there...

True, you just have to tweak a bit. But I had great success by first assigning a filter at the frequency of each major peak and trough. It may help you to identify the "major" ones by applying some smoothing.

The filters at the throughs aren't really used to boost much, they are used to couteract the effects of the neighboring peak filters.

[media2368]

Sonnie wrote:

Here's the filters I used:



Here's that filter file:

Zorax2.req

Again, this is the extreme merely to show the potential of tweaking the filters.

Keep in mind too that most of the boosting is not really boosting, it is simply compensating for the excessive bandwidth cuts. Those boosts never exceed your original response. Only the area between 40-50hz actually has any real boost.

Hi Sonnie,

May I know how did you calculate the above 'filter Setting' by using Zorax's mdat file? And how did you manage to preview the final curve after you key in the setting without playing the pink noise at Zorax's system setup?

Is it something to do with the 'Convolver' Plug-in?

I am really keen to learn about this, hope you can teach me how did you do it?

Thank you in advance.

[brucek]

media2638,

You can simply load any mdat file and start entering filters and the response line will dynamically change as you change filter settings.

You can begin with the suggested filters using the (FIND PEAKS and ASSIGN FILTERS and OPTIMIZE) if you want, or just begin with your own.

Play with the filter settings until the response line is at your target. This "off-line" method usually comes fairly close once those filters are entered into your BFD.

The one thing that may not work is when you add a filter with gain. Sometimes dips don't respond to gain, but most do.

What is Sonnie's secret? Trial and error and he's been doing it for a long time.

brucek

[media2368]

Thanks Brucek. I will test it out my own.