[DaveNagy]

Hi, everyone. I've decided to create a single thread, mostly so I can keep all my silly questions confined to a single, easy to ignore, spot. The trade-off is that the thread's title is pretty unhelpful. Sorry about that.

First off, greetings! Great program, great forum, and pretty good docs. (They actually exist, which is 80% of the battle.) Everyone involved are to be commended.

Okay, here's my story. I'm a computer geek and I have done some live sound and studio recording work in my younger days. I've just purchased (but have not yet received) a nice new sub. I do software QA for a living. So, I should be right smack dab in the middle of REW's target market, and one would think that I'd need very little hand holding. It appears that one would be wrong. :whistling:

Weekend before last, I installed REW on my "TV computer" and started working my way through the "Getting Started" instructions. Things started well, but then got a little inexplicable, so I wasn't actually able to take any measurements. Since the problems all seemed like they might be sound card (or driver) related, I'm in the process of installing REW to my work laptop as well. If that ends up working better, then we can all just pretend that my earlier attempt never happened. Still, out of curiosity, and in the interests of full disclosure, I'll describe the weirdness that I encountered. Perhaps there's an easy and well known fix.

(Again, I don't have my sub yet, and my BFD Pro only just arrived, so I'm just doing proof of concept experiments right now. If I can get to the point where I can do a full range graph of my right front speaker, I'll be happy.)

My TV computer is running Vista Home Premium (SP2, I think). It has motherboard-based Realtek audio. It's a pretty recent mobo. "Realtek HD", maybe?

Here's the short version. The sound card calibration went well, and I ended up with a nice flat response. There was absolutely no indication of feedback, which is surprising given my later experiences.

The "Checking Levels" step also seemed to go as expected. A "-30" setting on my AVR resulted in 73dB of pink noise coming out of my right speaker. (This according to my analog Rat Shack meter.) Since -30 was a nice and easy to remember setting on my AVR, I decided that 73dB was close enough for government work.

I then moved on to the "Calibrating the SPL Reading" step. Not surprisingly, my Rat Shack meter was still (again) hearing about 73dB, so that's what I told REW. Supposedly, I was now ready to take some measurements. (Note that at this point I have not yet hooked up my measurement mic. The only thing that was ever connected to the computer's line-in was the loop-back cable from the line-out, and at this point even that is no longer connected.)

Here's where I get all confused.

I went ahead and hooked up my Behringer mic and mixer, even though the instructions seemed never to mention that this needed to be done. (Did I miss a crucial step, earlier on?) At this point, at least two things seemed "not right".

1. The SPL meter at the top of the REW window was reading "73dB" pretty much all the time. Even when no test tones were being generated, and when the room was dead quiet. Fiddling with the output levels of the mic preamp doesn't seem to make any difference. Turning on the pink noise doesn't cause the SPL meter to react. Nothing did, except...
2. If I tapped or blew on the mic, I could hear the sound come out of my right speaker. (Not very loudly, however.) "Yelling" into the mic would absolutely peg the meters on the mic mixer, and would cause the SPL meter in REW to rise above 73dB, briefly.

That's not right, is it? It seems like REW isn't really getting much level from my test mic, and REW's SPL meter seems to be interpreting no input (or output) as "73dB", and the (kinda quiet) mic signal is feeding back into the speaker.

I poked around in every control panel I could find (both Window's and Realtek's), attempting to turn off the "monitoring" of the line-in input. Nothing seemed to make a difference. Muting the line-in made no difference.

I quickly became baffled, then annoyed, and ended my experimentation soon after that. I've installed REW onto a different computer, but in the mean time, have I fallen into a classic newb mistake? Ignoring the feedback weirdness for a moment, it seems like at no point did I set input levels from the mic mixer.

Do I understand the purpose of the following early steps correctly?

• "Calibrating the Soundcard" characterizes both the input and output stages of the sound card, and corrects for any non-linearities.
• "Checking Levels" is just a sanity test to verify that REW is able to output test tones at a usable volume.
• "Calibrating the SPL Reading" is a way to tell REW that, "That sound you are making right now? It's about this loud at my measuring position." (But does not assume that REW is "hearing" that sound yet.)

 

[brucek]

hooked up my Behringer mic and mixer, even though the instructions seemed never to mention that this needed to be done

It's assumed that this is necessary to do any measuring.

"Calibrating the Soundcard" characterizes both the input and output stages of the sound card, and corrects for any non-linearities.

Yep, its purpose is to create a soundcard calibration file, that is saved and automatically applied to the measurements when REW creates a graph.

"Checking Levels" is just a sanity test to verify that REW is able to output test tones at a usable volume.

Nope. The purpose is to set up the output and input levels of the soundcard. The setting of input and output levels done during the soundcard calibration routine have no bearing on the Check Levels test used when getting ready to measure..

"Calibrating the SPL Reading" is a way to tell REW that, "That sound you are making right now? It's about this loud at my measuring position." (But does not assume that REW is "hearing" that sound yet.) .

The purpose of the Calibrate routine (that is run directly after the Check Levels routine), is to match the real SPL meter at the listening position to the REW internal SPL meter. If you're not seeing the input level VU meter moving when you set Check Levels or when you run Calibrate, then the mixer isn't set properly. A typical dail setup of the 802 mixer is here.

If you were able to produce a soundcard calibration file and save it and then take a measure of that loopback cable and get a flat response, then you should have no problems taking a measurement.

Here's the little list I usually give people:
You need to first create a soundcard calibration file.

Until that is accomplished there are no connections to a mic/mixer or receiver required.

Simply connect a loopback cable from the right channel line-out to the right channel line-in. Be sure to select the right channel for the input channel on the REW Settings page.

Run the soundcard Calibration Measure routine and save the file with the Make Cal button.

Now, with the loopback still in place, run a Check Levels routine (select Main speaker to Check/Set Levels), and when done, exit the Settings page and run a Calibrate routine (select Use REW speaker Cal signal) and set the level to 75dB.

Now run a Measurement from 0-20000Hz and the result should be a flat trace.

Remove the loopback and use that calibrated channel for the mic/mixer and connection to your receiver.

Hook up the mic/mixer to the line-in (be sure to load the mic calibration file).

Hook up the receiver from the line-out. Use the AUX or CD input on the reciver with a Y-splitter so you can check both left and right speakers and sub.

Run Check Levels routine (with the mic/meter at the listening position, and set a hand held SPL meter placed close to the mic to 75dB).

Run Calibrate routine.

Measure.

If you feel you've got a monitor mode turned on in your soundcard, then check the line-in of the Recording Devices Mixer (not the Playback devices Mixer). It should be disabled in Record.

brucek

 

[DaveNagy]

Okay, I think I've answered some of my own questions. Way down at the bottom of the "Calibrating the Soundcard" instructions is the following important line:

"8. Remove the loopback connection and connect the soundcard to the SPL meter and AV processor/equalizer"

Well, that's pretty crucial! :unbelievable: I'm not sure how I missed that the first 76 times I read that page. I might humbly recommend that this crucial step be called out a bit more enthusiastically and fully. I probably blipped over it because I don't have a "processor/equaliser"[sic] yet, and my SPL is not ever going to be used as an input to REW. (I'm not blaming the documentation for my own idiocy, that's all me, I'm just saying that it could be made a little more idiot-proof.) I might recommend something like:

"Remove the loopback connection now. Connect your SPL meter (or measurement microphone) to your sound card's line-in, and connect the sound card's line-out to the appropriate input(s) of your audio system. (You may be connecting to the input of a receiver, a preamp, an equalizer, or directly to a powered sub, depending on your application.)"

I might also recommend that this instruction be moved to the top of "Check Levels" section, since it will be both easier to notice there, and will be more appropriate to that task. (The instruction has nothing to do with the act of calibrating a soundcard. Quite the opposite.)

The "Check Levels" page starts off with the following instruction, which I also managed to ignore: :wits-end:

"Setting the signal level REW uses during measurement involves generating a pink noise calibration signal and adjusting the AV processor's volume control and/or the calibration signal level so that at the measurement point (usually ear height at your main listening position) your SPL meter shows a level of around 75dB. The soundcard's input volume then needs to be adjusted to get a good signal level from the SPL meter or mic preamp when the cal signal is playing." (bolding mine)

That obviously should have been my clue that input levels had suddenly become important!

Since this appears to be the first time the user is asked to connect their meter/mic to REW, shouldn't the mic calibration file need to be loaded? I seem to have missed the instructions asking me to do this. (Not that anyone should be surprised about that at this point!)

 

[DaveNagy]

Thanks, Bruce. Very helpful advice. I'm sorry I didn't get my follow-on post done quickly enough to save you some typing.

It's assumed that this is necessary to do any measuring. (Hooking up the mic. -Dave)

Right, but I never really made it to the "Making Measurements" section, because things were already obviously foobarred by the time I was doing the "Calibrating the SPL Reading" steps. It appears that I went off the rails way back in "Calibrating the Soundcard". This is ironic, since calibrating my soundcard was the only thing that went as expected!

The purpose of the Calibrate routine (that is run directly after the Check Levels routine), is to match the real SPL meter at the listening position to the REW internal SPL meter. If you're not seeing the input level VU meter moving when you set Check Levels or when you run Calibrate, then the mixer isn't set properly. A typical dail setup of the 802 mixer is here.

I'm fairly sure that my mixer is set up right. Like I said I was pegging the meters deep into the red at some points. Still, I will double check next time. Maybe I'm reading channel level rather than output level. I think it's more likely that it's the input settings of the card that need to be tweaked. I never even touched those. Didn't realize I was even supposed to at that point. (, I didn't even connect the mixer to the sound card until well after the Calibrate routines.)

If you were able to produce a soundcard calibration file and save it and then take a measure of that loopback cable and get a flat response, then you should have no problems taking a measurement.

That's what I was thinking! :bigsmile: How the could could the loopback calibration steps go so flawlessly, and yet the card can't seem the "hear" my mic mixer, and everything it does hear is being sent to the line-out. Those two sets of behaviors do not seem like they could both happen simultaneously. (I suppose something must have changed in the interim.)

Here's the little list I usually give people:
You need to first create a soundcard calibration file.

Until that is accomplished there are no connections to a mic/mixer or receiver required.

Simply connect a loopback cable from the right channel line-out to the right channel line-in. Be sure to select the right channel for the input channel on the REW Settings page.

Run the soundcard Calibration Measure routine and save the file with the Make Cal button.

Now, with the loopback still in place, run a Check Levels routine (select Main speaker to Check/Set Levels), and when done, exit the Settings page and run a Calibrate routine (select Use REW speaker Cal signal) and set the level to 75dB.

Now run a Measurement from 0-20000Hz and the result should be a flat trace.

Yep, all of that went without a hitch, and it all made sense to boot.

Remove the loopback and use that calibrated channel for the mic/mixer and connection to your receiver.

Hook up the mic/mixer to the line-in (be sure to load the mic calibration file).

That's where it all went wrong. :bigsmile: I've since found the single sentence that (sort of) tells you to remove the loopback and connect the mic, but I still don't see anything about loading a mic calibration file. Seems like a strange omission. Perhaps it's in the actual calibration wizard that REW pops up at that point:

e.g. "Press the Check Levels... button and follow the instructions on screen."

If you feel you've got a monitor mode turned on in your soundcard, then check the line-in of the Recording Devices Mixer (not the Playback devices Mixer). It should be disabled in Record.

I looked, but I will certainly look again. Do I need to uncheck one or both of the "Control xxput mixer/volume" check boxes to make sure that REW isn't overriding these settings? Do you find it odd that the loopback calibration worked perfectly, given that I apparently have some sort of monitor mode activated?

Thanks for all the feedback. On my next attempt, I should have a better idea about what I'm trying to achieve at each step.

 

[brucek]

Connect your SPL meter (or measurement microphone) to your sound card's line-in

This is why I created the REW Cabling and Connections Basics to expand on the Help Files with regard to connections of meters or mics.

shouldn't the mic calibration file need to be loaded?

See Mic/Meter Settings from the REW HELP FILES.

brucek

 

[johnmail76]

thanks

 

[DaveNagy]

Hello again.

I just repeated my attempt to "measure" one of my speakers, and I had far better luck this time around. I had no trouble getting a plausible looking response graph. I used a laptop this time, as opposed to my HTPC, and was able to avoid that baffling feedback issue. (The laptop is running Windows XP, and has an M-Audio Transit attached via USB.)

I just realized that, once again, I managed to overlook the instructions that told me when to load a mic calibration file. I shall go read the instruction page that Bruce references in the post above.

But in general, I think I'm ready to start looking at the docs for my FBD Pro, and continue to anxiously await the arrival of my new sub. Thanks again, Bruce.

 

[DaveNagy]

I played with my FBD Pro a bit this evening. I was able to easily transfer filter settings from REW to it via a MIDIMan Uno interface. It worked really well. (Yes, I was a little surprised!)

I would kinda like to "play with" the EQ filter settings in REW. You know, just experiment with the various automatic and manual ways that you can flatten a jagged response, experiment with different house curves, minimizing the number of filters used, etc. The trouble is, I don't have a working sub at the moment, so I can't generate a realistic response graph to play with. I should be able to download and load one into REW, right? Do any of you have a graph I could work on? Assign me some homework, please! :bigsmile:

Changing the subject: I'm going to assume that my bass response graphs are going to differ considerably, depending on the location of my listening/measuring point. I plan to measure at several points, just to see if and how much they vary. But, I don't quite understand what I should do about those variances. Should I just pick a single graph from prime seating spot, and equalize to that? I was wondering if will REW let you average several response graphs together, so you could devise some kind of a compromise solution. Is that possible? Desirable?

(My room is super irregularly shaped, and is open to most of my (town)house. 6000+ cu/ft, I estimate.)

 

[thewire]

Do any of you have a graph I could work on? Assign me some homework, please! :bigsmile:

The only way to practice with filters is to actually load them into the equalizer and measure the subwoofer/subwoofers. Too many variables are taking place to predict the result but it can come close. You can look at the features of REW better by having a measurement to work with so here is a measurement before filters. The filters were done prior to some minor room treatment changes. I still may need to dial them in better yet but not really felt I had the need to. Part of knowing you have the correct results and the most important is to actually listen. These were done with an 85 target but most here seem to use 75dB. The result in headroom after filters brought the low end response to 20.9Hz and the waterfall around 300ms. The laptop was using the adapter also so might be some slight noise/distortion in the response.

 

[brucek]

I plan to measure at several points, just to see if and how much they vary. But, I don't quite understand what I should do about those variances. Should I just pick a single graph from prime seating spot, and equalize to that? I was wondering if will REW let you average several response graphs together, so you could devise some kind of a compromise solution. Is that possible? Desirable?

Strangely enough, the Averaging feature it's located under the Averaging tab.

Take up to 8 measurements and average (by checking the box) of any number of those eight measures to produce a new measurement that you can then create filters for your EQ.

brucek

 

[DaveNagy]

The only way to practice with filters is to actually load them into the equalizer and measure the subwoofer/subwoofers. Too many variables are taking place to predict the result but it can come close. You can look at the features of REW better by having a measurement to work with so here is a measurement before filters. [...]

Hey, thanks for that file! I should have explained better: For now, I just want to simulate the creation of some filters. I'll just let REW predict the results for now. I just want to get some practice in. It seems like there might be a bit of an art to it. (Assuming I don't just let REW create the filters automatically. But where's the fun in that?)

Strangely enough, the Averaging feature it's located under the Averaging tab.

How unintuitive! How was I supposed to to know to look there?!? :nerd: I guess that means there will be an associated help page to read as well. Sweet. I'll also poke around in the forum to see if I can find some discussion of this whole "correcting to an average" concept.

Thanks again for the help. If anybody else wants the send me a "before" graph, I'd appreciate it.

 

[DaveNagy]

To: "thewire",

I think I'm going to fail your homework assignment! :mooooh: That graph looks kins of un-fixable to my untrained eyes. There's that huge hump from 50Hz to 105Hz, and below that, things drop up off pretty steadily. Even if I apply a truly heroic (and unwise?) boost down around 42Hz, I still end up with something that looks more like a woofer than a subwoofer.

And if I attempt to tame the hump centered on 73Hz, (which is difficult since it's not shaped at all like my filters) I'm still left with no extension to speak of. Plus, now I've cut the heart out of your bass. You'd have to turn way up in order to compensate. I guess that's the way to go, however. Let me give that approach another try and then I'll post my results.

 

[DaveNagy]

Okay, here we go. You folks that actually know what you are doing will probably get a good laugh out of this. I'm going to sort of "think out loud" here, so perhaps someone can correct me when I go astray. First off, here's the measurement that "thewire" sent me:



Notice that I've cranked my target line up as high as it will go. I thought it looked like it would be a better "fit" up there. This is a loud measurement! I decided that I won't be touching that dip (suck-out?) at 154Hz. That's way above the crossover, and it looks kinda un-fixable anyway. (Because it's so narrow.) The same goes for that hump "down" around 135Hz. (That one looks like it could be fixed, but it's still up too high.)

I decided to attack that little dip at 42Hz first. Here's 6dB boost, centered there. We're still left with poor bass response, but at least it falls off more smoothly now:



Lovely. I don't dare try to "extend" the bass any further, do I?

Now I'm going to attempt to tame the part of that broad hump that's above my target line. I centered my cutting filter at 94Hz (so high!) because, well it seemed to work best there. Check it out:



That was a 9dB cut, by the way. Wide too. Seems a bit extreme to me. But, hey. Nothing ventured, nothing gained. I could continue and try to knock down those two humps at 102Hz and 134Hz, but I think I've done enough damage already.:time-out:

Here's the resulting (predicted) curve in all its glory.



Be gentle.

 

[monk]

1. The SPL meter at the top of the REW window was reading "73dB" pretty much all the time. Even when no test tones were being generated, and when the room was dead quiet. Fiddling with the output levels of the mic preamp doesn't seem to make any difference. Turning on the pink noise doesn't cause the SPL meter to react. Nothing did, except...
2. If I tapped or blew on the mic, I could hear the sound come out of my right speaker. (Not very loudly, however.) "Yelling" into the mic would absolutely peg the meters on the mic mixer, and would cause the SPL meter in REW to rise above 73dB, briefly.

I use Vista Home Premium as well, also with a Realtek soundcard. I have also experienced this behaviour on more then one occasion. Usually i restart REW, and all is well again. However, it's hard for me to ascertain why this behaviour occurs. I might be the flexibility of the Realtek chipset to assign the physical inputs on the fly to "line-in", "microphone", etc. Perhaps (under some conditions?) it opens a monitoring function when plugging the mic. or SPL in.

Just reaching of course...

J.

 

[brucek]

Notice that I've cranked my target line up as high as it will go.

Completely incorrect. Put your target back at 75dB. The plot looks like it has a very high crossover used, so you need to know the crossover before beginning. You also need to know what type of sub it comes from. Looks like an IB. You really should be playing around with your own plots, where you know all the facts of the situation.

I decided to attack that little dip at 42Hz first. Here's 6dB boost, centered there

No, you don't add boost to that.

I could continue and try to knock down those two humps at 102Hz and 134H

No, they're probably above the crossover and aren't touched until the mains are added to see their resultant effect - which may be zero.

Simply place the trace around the target line at 75dB and remove as much of the problem, which is a peak between 50Hz-100Hz. That's it.

brucek

 

[DaveNagy]

I use Vista Home Premium as well, also with a Realtek soundcard. I have also experienced this behaviour on more then one occasion. Usually i restart REW, and all is well again. However, it's hard for me to ascertain why this behaviour occurs. I might be the flexibility of the Realtek chipset to assign the physical inputs on the fly to "line-in", "microphone", etc. Perhaps (under some conditions?) it opens a monitoring function when plugging the mic. or SPL in.

You may well be correct. I noticed that the soundcard didn't "create" the ports until I plugged something into them. And it would destroy the ports when I unplugged. This meant that I couldn't do any re-patching while REW was running. REW wouldn't pick up pick up the changes. I had to restart REW a few times. It's very possible something got confused or changed in the process. I'm happy with how the laptop and USB sound interface is working for my now. So I'll just stick with that.

Completely incorrect. Put your target back at 75dB. The plot looks like it has a very high crossover used, so you need to know the crossover before beginning. You also need to know what type of sub it comes from. Looks like an IB. You really should be playing around with your own plots, where you know all the facts of the situation.

That would certainly be preferable. Hopefully my sub should arrive in a month or so. As I mentioned above, I didn't really know how to approach thewire's graph -- it seems quite unlike what other people have posted. It almost seemed like a trick question to a newb like me. :rubeyes:

Your approach makes sense. If you can knock down that big hump between 50Hz and 100Hz, the rest of the graph starts looking better all by itself. In fact, that's the first thing I tried. But I decided I must have been going about it wrong, because the hump was so huge (20dB+ above the target, and really wide) that my puny filters only seemed to be making it mad. :scratchhead:

Basically, I just wasn't sure how approach such a wide, tall and flat topped hump. Is is better to use a series of not-particularly-wide cutting filters, arranged side by side? Or perhaps use one huge cutting filter to "hollow out" the center of the hump, and then come back with boost filters to fix collateral damage caused by the first overly wide cut?

Or does it matter?

 

[brucek]

I didn't really know how to approach the wire's graph -- it seems quite unlike what other people have posted. It almost seemed like a trick question to a newb like me.

Yeah, exactly - it wasn't like the typical response you'll encounter.

Below are a bit more typical before and after response.

I show a Servo15 sub I use, and you can see a typical peak at 20Hz, 35Hz and 60Hz, and a dip at 28Hz.

Not much you can do with the dip, so you work on the peaks and the result I used is shown in the second trace. Perfectly acceptable.

I used a 60hz crossover and a house curve of [email protected] and [email protected]

Go ahead and play with the mdat file I attached. It is zipped becuase the size was large.




brucek

 

[JohnM]

Go ahead and play with the mdat file I attached. It is zipped becuase the size was large.

If you tick the "Decimate IR" box in the Analysis settings REW will automatically downsample low frequency measurements (when they are first measured) and produce much smaller measurement files.

 

[brucek]

If you tick the "Decimate IR" box in the Analysis settings REW

Yep, good tip John.

This was an old measurement from V3 days that I pulled up to give the OP an example. Unfortunately it was a bit large...

brucek

 

[thewire]

Good anylisis Dave. You are in fact correct that the filters needed might have been considered extreme. This is typical of dedicated rooms were achieving a flat response (if that is your goal) can be more challenging espicially in smaller rooms. The result using mulitple filters instead of a broad filter or boost results in more precise control over a larger range were one filter may have been more suited if one room location was considered instead of many. Equalization in a non dedicated room will indeed very non typical of my result. The gain near 50Hz is typical of that in small rooms and by no means atypical for results. :no: Study with what you are comfortable with and expect to acomplish. :T