Jump to content

Using REW to Find Parametric Equalizer (PEQ) settings


Chris A

Recommended Posts

Assuming that you have taken a measurement using REW (Room EQ Wizard freeware), you probably have a plot that looks something like this:

 

Typical REW Frequency Reponse Plot.png

 

What you have is a frequency response that's not quite as flat as you'd like.  So for that particular plot, you'd like to find the digital crossover equalization filters to make it flat--but without a lot of cut-and-try and doing a bunch of measurements (upsweeps) along the way.

So you look at the REW window and you see the function bar across the top of the window:

 

REW Function Bar.GIF

 

The next-to-last button is "EQ", which you push, and the following window pops up:

 

REW Equalizer Window.GIF

 

The window has two plots: the plot that the started with (frequency response) and a second plot area (blank) which can provide a waterfall, impulse, or Pole-zero plot.  There are two arrows next to the left border of this window.  If you punch the arrow pointed down, the bottom plot is replaced by the upper plot whose size now occupies the space that both windows did.  (Do it now.)

 

On the right top side of the window, there are 5 bars with text embedded in them, namely: Equalizer, Target Settings, Filter Tasks, Modal Analysis, and Resonances. 

 

If you click on the first bar, it opens a menu downward and shows you a list of equalizers that are supported directly by REW. 

 

Equalizer type menu.GIF

 

If you don't see your equalizer model in that list, fear not.  You can pick one that's close in terms of your options.  For example, using the EV Dx38 or Yamaha SP2060, I used "miniDSP", and it works.  If using Xilica, then select "XP2040", etc.  If you sequentially choose each equalizer, and look at the PEQ filters (which we'll generate in a moment), you'll find one that give you the best fit to the features of your crossover.  Once you select a crossover type, minimize the Equalizer sub-window by clicking on its top bar again.

 

Now click on the next menu bar - Target Settings:

 

 

Target Settings menu.GIF

 

The important settings here are "Full Range" if using you're equalizing a full range loudspeaker.  REW can also help you with subwoofers and "bass limited", i.e., HT surround loudspeakers that are intended to cross above a certain subwoofer crossover frequency.

 

LF slope indicates intended the slope of the roll-off of the low frequency end (12 or 24 dB/octave) corresponding to the crossover filter slope used between the loudspeaker and the subwoofer.  If you look at the measurement plot that you are using to equalize, it will show you the lowest crossover frequency that you should use.  For my Jubilees and K-402-MEH, that frequency is about 30 Hz.  If you're using HT loudspeakers, it may be as high as 100 Hz.  Set that LF roll off point to match your loudspeaker.  The same settings are used for the high end. 

 

All of these settings are settable to new default values within the "Preferences" menu on the top bar of the REW main window.

 

This is the place to put in your "house curve" if you choose to have non-flat loudspeaker response.  You can boost or cut highs or lows using the above controls, and the resultant curve is visible, so you can see what your "goal curve" is.  I use flat response everywhere since I unmaster my recordings, but you may choose to use a "one size fits all" house curve to compensate for non-flat EQ used on your recordings.  In the context of home theaters, that's what a "house curve" is doing.

 

The last item on this menu is extremely important.  It tells REW what SPL (loudness) to aim for overall when it optimizes the equalization PEQs.  You want to get this right, or, not only does REW yell at you, it also will produce too many PEQs to try to flatten the response.  For the measurement above, 70 dB is the right answer.  If you use a higher value, REW will try to boost all frequencies below your target level.  If you use a setting that's too low, REW will attenuate everything to make that lower SPL.  It's better to set this value 1 dB too low than 1 dB too high.  It's better to attenuate using PEQs than to boost. Use the overall channel gain setting on your digital crossover to generally boost or attenuate one driver channel of your loudspeaker.  It will be apparent when you need to change the gain of a channel.

 

Next up: Creating your optimized PEQs using REW

  • Like 7
  • Thanks 1
Link to comment
Share on other sites

Assuming you followed everything above, you should have something that looks similar to this after you select the next menu bar, called "Filter Tasks":

 

Filter Tasks window.GIF

 

Note that the "target curve" on the left-hand plot is going right through the middle of the curve.  Once you've set up your "house curve" preferences in the Preferences dialog, the only thing that needs to change from loudspeaker to loudspeaker is the "Target Level (dB)" setting after each PEQ optimization or measurement sweep.  That value will reset itself after every measurement to 75 dB, and you'll have to manually set it to the right level after each measurement by looking at the measurement and deciding what the target level should be.

 

The next thing to look for is the "Match Range  XXX to YYY  Hz" . If your setting the bass PEQs first, then set these two frequencies to:

 

1) the lower frequency should be the lowest frequency that you want to try to equalize.  I wouldn't set this more than 1/2 octave below where it naturally starts to fall off.

 

2) the upper frequency is set to about 1/3 octave below your woofer-midrange crossover frequency.

 

The next item is to set your "flatness target".  I generally use 4 dB (implying the optimizer will stop optimizing when the predicted response gets to plus or minus 4 dB from your target level, between the two frequencies that you selected above.  Try setting it at 4 dB.

 

EDIT 23 Oct. 2018: REW uses the smoothness settings for the measurement frequency response plot (that you set under the "Graph" menu) to determine the optimized PEQs.  I recommend using "Psychoacoustic" smoothing before running the EQ filter optimizer, so that you're not trying to create a lot of high-Q (i.e.,very narrow) filters that are probably trying to correct for high-Q room modes. 

 

Now you are ready to generate the PEQs...

 

Punch "Match Response to Target", then watch the show go on.  Filters are created, optimized, and consolidated before your eyes.  The optimizer stops.

 

When you do this for the example above, and then punch the "EQ Filters" button just above your plotted data, you should see something like this:

 

Optimzed PEQs window.GIF

 

The corrected curve is shown, along with the individual PEQ curves and the overall PEQ correction curve (you can turn the curve traces on and off using the check boxes below the plot to see everything).  The overlay window are the PEQs that you can type into your active crossover settings.  You just transfer these values to your crossover control software, or you can punch them in using the front panel of the crossover itself.  For the EV Dx38 and the Yamaha crossovers, this is easy to punch them in the front panel.  For Xilica, I strongly recommend using the XConsole application on your computer, then synchronizing with the crossover using the "Tools" menu within XConsole.

 

If you generated too many PEQs, then hit "Reset Filters for Current Measurement" and okay the dialog box that pops up.  Increment the flatness target by one or two dB upwards, then punch "Match Response to Target" again.  If too few PEQs are generated (or none), decrement the flatness target by 1 dB then reset and run again.

 

Repeat this process until you get something that looks reasonable in both the predicted response curve and the number of PEQs.  Once you transfer those new values into your digital crossover, run a sweep again to verify the results.  I think that you'll be very close to your goal - in one try!  If not, then go back to the beginning and change the gains of the channels, the target level used, and/or the flatness level, then try again.

 

If your resulting PEQs are outside the values allowable for your crossover, then go back up to the "Equalizer" type menu, and select a different equalizer type until you find one that results in the PEQs that you can use.  The EV Dx38 has limitations on attenuation of PEQs that the Yamaha and the Xilicas do not.  You'll find that having the ability to do big attenuating PEQs with your crossover will save having to double up PEQs to get enough attenuation.  This is a big advantage of the newer digital crossovers - so you don't have to use shelving filters. 

 

You'll notice that REW doesn't use shelf filters.  If you want to use them, simply dial in what you want and take a sweep, then use that sweep's data to continue to build your PEQs. 

 

Voila!  Instant success (usually).

 

Chris

  • Like 4
  • Thanks 1
Link to comment
Share on other sites

Here's a plot of the "predicted response" of using the PEQs above:

 

Example target response.png

 

Repeat this process for your high frequencies (tweeter and/or midrange) and lay in your PEQs for those frequencies into the output channels of your digital crossover.

 

If you have frequency response issues in the crossover regions, then use the resulting PEQs from the optimization on the INPUT side of the crossover, or if you start to run out of PEQs on the output channels, switch to the channel inputs and keep going.

 

If you're dissatisfied to the flatness of the response after all  is done, and you still have available PEQ filters not being used in the crossover, then you can run the optimization again using the updated equalization PEQs embedded to see what more you can add to flatten response.

 

Chris

  • Like 3
  • Thanks 2
Link to comment
Share on other sites

Note that by using REW equalization in coordination with a digital crossover like the Xilica-8080, you really don't need Audyssey, YPAO, ARC, etc. room correction software--at all.  And you get to see and control what it is doing. 

 

When you consider that the price of the professional version of Audyssey is on the order of a used Xilica or Yamaha digital crossover, you might realize that the real value is going this approach rather than the "push a button and stand back" approach used by canned room equalization firmware--and that typically leave much to be desired.

 

I find that the amount of time required to do it this way or by using Audyssey are about equal. 

 

The only thing that I use Audyssey for is setting channel delays for the 5.1 (i.e., setting 6 channel delays).  That's it.  I use my ears and a hand-held SPL meter to set channel gains manually.  It takes about 60 seconds to do it.

 

 

  • Like 2
Link to comment
Share on other sites

Chris, as always your posts are very interesting. 

 

The headache I always have is microphone placement. When dialing the K402 with a different driver (in my case a JBL 2446). The mic was placed fairly close to the face of the horn mouth and centered. I am okay with that measure.

 

 For the bass bins I do a funny combination. The first is to place the mic at the mouth of the bass bin and second at some distance away, perhaps a few to several feet. By necessity the measures must be made in my listening room. However the variability and room interactions when the mic is placed AWAY from the mouth of the bass bin are huge. Hence, my headache.

 

I am not telling you anything you do not already know, however I am curious on your strategy when you do your measures. I am guessing that there is no simple answer or the answer starts with the words: "it depends ..."

  • Like 1
Link to comment
Share on other sites

I use a microphone measurement at one metre (~39.4 inches) from the front face of the loudspeaker under test, centered on the high frequency horn (the K-402 in my case)--with the microphone pointed on-axis (not at 90 degrees incidence), with a lot of absorption material on the floor between the microphone and the loudspeaker and about 1 metre wide.  Then I run the PEQs in REW, looking for PEQs that look like they're trying to boost room nulls.  I simply ignore those (high Q) PEQs that are generated by REW.

 

I find the results are quite good and result in usable full-range phase curves (after adjusting for impulse response delay).  Taking measurements further back makes the phase measurements useless--and those phase measurements are pretty important, I've found (delays using phase and excess group delay curves--which I find correlate to overall sound quality fairly closely). 

 

I find those EQ and delay results to be superior to other more complicated schemas, however technically accurate they may be.

 

Chris

  • Like 1
Link to comment
Share on other sites

It sounds like you are using this on your "new-fangled" horn with drivers in the K-402. 

So I guess it is partly a philosophical question. When you do this with the traditional version of a Klipsch jubilee, there going to be peaks & dips at the lower frequencies due to a smallish mouth on the bass bin. I have always bee liberal about flattening the peaks but conservative about boosting the dips too much, especially at lower frequencies.

 

With regards to the measures, if I place the mic at the mouth of the bass bin (or even slightly inside) the measures are quite similar to those that Roy Delgado has provided (certainly at the lower frequencies). However, when I pull the mic back and away from the bin, then  I start to see all sorts of "room effects". One can then get into this back and forth about "correcting" the speaker itself  vs  "correcting" the speaker-in-that-room. Some are opposed to correcting for the room interactions and others are not. Those opposed frequently have some soft-headed thinking on this issue. When I read Toole's book he seems a bit more lenient and encourages a degree of correction. 

 

When I am not worrying about this stuff, I am actually enjoying the sound. Will it never end?

  • Like 2
Link to comment
Share on other sites

17 minutes ago, PrestonTom said:

I have always been liberal about flattening the peaks but conservative about boosting the dips too much, especially at lower frequencies.

That's my inclination,too.  Boosting room nulls doesn't work very well.  I caught myself doing it in the past: I try not to do that anymore.  Sometimes, however, the width of those nulls can fool you.  I look at the excess group delay curve now to see those room nulls - they're not minimum phase.

 

17 minutes ago, PrestonTom said:

However, when I pull the mic back and away from the bin, then  I start to see all sorts of "room effects".

Yes, but you're also measuring the lobing behavior, which of course changes with frequency, and that's real in the midbass region.  The K-402-MEH really doesn't have that problem, while the Jubilee bass bin above ~200 Hz has lots of lobing, so it's really a mid-bass issue.  You see that in the group delay plots, too. 

 

The MEH is much cleaner in that regard (low group delay, and GD curve smoothness with frequency--all the way up and down). That's one of the audible differences, IMHO.

  • Like 1
Link to comment
Share on other sites

Chris, this should be a stick with your Bi amping thread maybe a sub section.  This should also be a stick over at HTS.   I have searched in the past for such thread and had no luck, it seems most of the tutorials cover sub integration.  Thanks again for your valuable time. 

 

Would one use your method of mic placement for equing the speaker as it self, then move it to the listening potion when integrating the sub?  I guess in your case, since you run a sub with each main (minus your center).  You may be able to eq that with the main at the position.

  • Like 1
Link to comment
Share on other sites

On 11/13/2016 at 5:19 PM, The Dude said:

Would one use your method of mic placement for equing the speaker as itself, then move it to the listening position when integrating the sub?

Yes.  You're in the pressurization zone near field at subwoofer frequencies.  Near field acoustic measurements at less than a quarter wavelength after it leaves the horn's mouth are subject to errors.  That's what Tom was referring to with his Jub bass bin acoustic measurements.

 

So for subwoofer measurements, the only thing that you can really do is to measure these frequencies at your listening position(s) in room, and arrange the subwoofers and your listening positions for maximal transfer of SPL below the transition frequency of listening in the near field.

 

Chris

 

 

  • Like 1
Link to comment
Share on other sites

Chris, how do calculate if any attenuation is required between each driver on a 2-way speaker, HF vs LF?  For example the -3db between the Jube Bass Bins and the K402.  I suspect that if it was close and you ran your full range measurement it would compensate, but what if it was way off? 

 

Great Write-up!

Joe

  • Like 1
Link to comment
Share on other sites

Joe, it depends on the starting frequency response of the 2-way.  I've used as much as 10 dB of attenuation on the Belle midrange, and as little as zero on the K-402-MEH.  It depends on how you structure your PEQs and shelving filters.  I've found that attenuation using PEQs is much more preferable to boosts (the overall architecture of the crossover can run out of local gain otherwise, and produce strange results--something that I found out the hard way with the Dx38 crossovers). 

 

I usually just look at the bass channel first (turning off the HF channel), then the compression driver (HF) channel is EQed flat. Then I set the relative gain to "join" the two parts.  It's pretty straightforward.

 

Chris

  • Like 2
Link to comment
Share on other sites

On 11/13/2016 at 11:52 PM, Khornukopia said:

...which led me to discover that I needed to install the free REW software update.

Yes--the upgrades to REW are sometimes significant I've found.  The REW application itself can be set to check for updates when you start up.  I've been pretty impressed with what's happened to this application in the last 4-5 years in terms of improvements, so I stay up to date on latest releases.  It's one of the very few applications that I can say that about.

 

Chris

Link to comment
Share on other sites

On 11/13/2016 at 5:19 PM, The Dude said:

Chris, this should be a stick with your bi-amping thread maybe a sub section.  This should also be a stick over at HTS.

I've placed a link to this thread and the Xilica integration thread in the bi-amping/tri-amping FAQ. 

 

I'm not sure about where to place a link in the home theater section...or the two-channel section for that matter (assuming any mossbacks over there would consider--for a moment--moving out of the stone age. ;) )

 

Thanks for the idea.

 

Chris

  • Like 1
  • Haha 1
Link to comment
Share on other sites

  • 2 weeks later...

I bought an ECM mic and a phantom power source mixer when I first started using REW a few years ago, but the mixer contributed too many potential variables, so I replaced it all with a simple USB microphone. I attached a one meter long string to the mic tripod mount to save from needing a tape measure at every different speaker test.

 

 

 

ECM.JPGUSB microphone.JPG

  • Like 3
Link to comment
Share on other sites

By the way, REW uses the smoothness settings for the measurement frequency response plot (that you set under the "Graph" menu) to determine the optimized PEQs.  I recommend using "Psychoacoustic" smoothing before running the EQ filter optimizer, so that you're not trying to create a lot of high-Q (i.e.,very narrow) filters that are probably trying to correct for room modes. 

 

Once you select "Match Response to Target", the PEQ filters are created and optimized before your eyes, and you can see them numbered as they are created, and even deleted as the optimizer continues to skinny down the number and strength of the individual PEQs. 

 

You can also use the "Modal Analysis" and "Resonances" menu bars in the right hand side of the EQ window to plug in your room's dimensions and calculate likely room resonances and cancellation frequencies based on those room dimension inputs.  I find that somewhat useful in deciding if midrange and midbass PEQs that are generated that also look fairly narrow and steep--to decide if they are in fact trying to boost or cut room resonance frequencies.  I find this is a useful feature to have for all loudspeaker PEQ optimizations. 

 

Chris

  • Like 1
Link to comment
Share on other sites

  • 1 month later...

Hi Chris,

 

Is it better to do the REW sweep at say 90dBC and then attenuate all the dips out completely or to have an individual max boost at a high number with a target mid curve (15dB like you do above)?

 

Another PEQ filter tutorial I was reading said to set the max boost to 0dB (sounds overly cautious)?!?  I guess this is in fear of over driving a driver or causing phase distortion?

 

I'm thinking I used a target of 8dB when I tried this on my XMC-1 (has 12 band PEQ for each channel), but then when I rerun my measurement at 90dBC SPL, I noticed THD at some frequencies had gone up (while most THD remained the same).  Am I correct in thinking that if I use an individual target set of 0dB THD my driver's distortion would not have increased?

 

Also, what are there any drawbacks to just hacking the curve away and setting the target just above a dip (instead of in the middle)?

 

Forgot to mention: shelving shows up in the latest version of REW as it lets you set crossover slopes for roll offs (but uses up some of the PEQs).  Any advantage to doing this on the Yamaha instead of using a normal shelf filter?

 

PS: you were right about the Ashley XR1001's.  When I used 12 watt resistors to make the impedance of the MF and HF close to 10 ohms each (took a 4.5 and 6 ohm resistors; amp really was seeing a 3.2 ohm impedance on one of the drivers forgot which now), but I noticed there was only a slight decrease.  This didn't make sense as 10 ohms is a nice load the amp should like, and the hiss was still there even with my processor (XMC-1) turned off!

 

Then I unhooked the Ashley and the hiss was at a very low level as before: very hard to hear at the MLP.  Will the Yamaha SP2060 I ordered give me any trouble?  I'm hoping there are adjustable level controls that will fix any issue like this on the Yamaha?

 

On 11/13/2016 at 11:44 AM, Chris A said:

The EV Dx38 has limitations on attenuation of PEQs that the Yamaha and the Xilicas do not.  You'll find that having the ability to do big attenuating PEQs with your crossover will save having to double up PEQs to get enough attenuation.  This is a big advantage of the newer digital crossovers - so you don't have to use shelving filters. 

 

You'll notice that REW doesn't use shelf filters.  If you want to use them, simply dial in what you want and take a sweep, then use that sweep's data to continue to build your PEQs. 

 

Link to comment
Share on other sites

On 1/28/2017 at 4:57 PM, etc6849 said:

Is it better to do the REW sweep at say 90dBC and then attenuate all the dips out completely or to have an individual max boost at a high number with a target mid curve (15dB like you do above)?

It depends.  Generally you see boost on the LF--usually below Fc of a woofer, and cuts almost everywhere else. 

 

On 1/28/2017 at 4:57 PM, etc6849 said:

Another PEQ filter tutorial I was reading said to set the max boost to 0dB (sounds overly cautious)?!?  I guess this is in fear of over driving a driver or causing phase distortion?

If that were true, then there would be no filters with boost capability - like that of passive crossover and equalization circuits--attenuation only.

 

In my experience with active EQ, the goal is to have more attenuation than boost, usually much more. That way it typically ensures that non-minimum-phase room and driver/horn nulls aren't boosted, and each electronics box is assured to have enough gain, i.e., they won't "run out of gain" during musical transients.

 

On 1/28/2017 at 4:57 PM, etc6849 said:

Am I correct in thinking that if I use an individual target set of 0dB THD my driver's distortion would not have increased?

If you're aiming for flat frequency response and the crossover is adding very little to the THD (a reasonable assumption for digital crossovers), any way you achieve flat FR will result in the same levels of harmonic distortion (HD) at equal loudness on axis (SPL). The only time when THD rises is when boost is used below or above the natural frequency response of the driver+horn, i.e., when trying to extend the FR. 

 

If using net boosting across the board, you're simply using the crossover to be a preamp in addition to an EQ/crossover.  This is the condition that I try to avoid, since the filters aren't really meant to be optimal to do across-the-board boosting.

 

On 1/28/2017 at 4:57 PM, etc6849 said:

Also, what are there any drawbacks to just hacking the curve away and setting the target just above a dip (instead of in the middle)?

I've found that the goal is to use as few filters as possible, then they aren't fighting each other.  Putting a single attenuating PEQ right on top of a response peak, and using the negative gain of that filter to pull that response down usually works the best to minimize filters.

 

When one end of the of driver+horn SPL response is greater than the other end, the shelving filters usually work best, and using attenuation on the high end is preferred to boosting the low SPL end.  Then boost the gain for the channel to level everything back to flat SPL.

 

On 1/28/2017 at 4:57 PM, etc6849 said:

Will the Yamaha SP2060 I ordered give me any trouble?  I'm hoping there are adjustable level controls that will fix any issue like this on the Yamaha?

My Yamaha SP2060 has a little hiss--incrementally more than the Dx38s.  On the TADs, I could hear the hiss when I stuck my head inside the K-402s. I was using an XLR-RCA cable from the Yamaha to the First Watt F3 - and this might have increased the hiss.

 

Setting the gain structure of the amp, crossover and preamp will help.  Generally, using a little more overall gain on the crossover and less on the amplifier will suppress the DSP noise floor hiss.  (The Xilica XP-8080 is dead quiet.)

 

Chris

  • Like 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...