Jump to content

Using REW to Find Parametric Equalizer (PEQ) settings


Chris A

Recommended Posts

23 minutes ago, Chris A said:

 

PEQs that are generated using very small values of bandwidth (BW) (conversely said, high values of "Q") are the first filters that I look at omitting.

 

I think I've started to chase my tail a bit the last few days. I've got it sounding very good, but trying to get the couple of group delays that were still above 2ms to go away and it seems to impact other things like the Spectrogram and phase a lot.

 

Next I started to read about Group Delay and Excess Delay and minimum phase on the REW site:

https://www.roomeqwizard.com/help/help_en-GB/html/minimumphase.html

and while I don't exactly understand everything I seems like you don't want to try and "fix" Group Delays or SPL spikes/dips anywhere the Excess Group Delay plot is not flat....which I take it to mean "minimum phase". Or maybe a better way to put it....Use the Excess Group Delay option, from the Group Delay section, to see where the line is relatively flat, which is Minimum Phase. Only try and fix things there (flat/minimum phase), not everywhere. Is that a decent rule of thumb?

 

I think most of my tail chasing could be a result of trying to use PEQs with small numbers (High Qs) to try and pinpoint spikes and dips that can't easily be addressed.

Link to comment
Share on other sites

1 hour ago, rplace said:

I think I've started to chase my tail a bit the last few days. I've got it sounding very good, but trying to get the couple of group delays that were still above 2ms to go away and it seems to impact other things like the Spectrogram and phase a lot.

This is the crux of the issue: whether you are willing to continue measuring and making updates, while knowing when you're at the 99% level of performance that you can extract out of what you've got.  This state of realization of what you can achieve isn't always self apparent as evidenced by my zero-phase-growth crossover filters and phase flattening efforts over the past year or so, which came as a complete surprise. You can sometimes realize that there is something that you can do or mitigate to increase performance even further. 

 

If you're on-site and can take measurements yourself, you can go quite a bit further than I can help you remotely via email into the last tenths of a percent of performance (on the nonlinear scale of subjective listening performance vs. objective measurement performance attained). 

 

For me, taking measurements, reading those measurements, and dialing in DSP crossover(s) has been a voyage of discovery of new tricks/approaches and understanding of what is occurring...and how it all relates to that subjective listening scale of performance.  This voyage has spanned 12 years of using DSP crossovers--with particular progress in the past 5-6 years on increased knowledge and resulting performance gains.  What's amazing about this journey is that I find myself looking closer at the measurements, updating the DSP crossover settings, making adjustments to the acoustic room treatments, and updating drivers/horns, etc.  instead of the more common path of swapping out of electronics that most other "audiophiles" embark upon.  What I try to do is share these "aha" moments so that others don't have to wait 12 years before finding it themselves (if ever). 

 

So the bottom line is: save your settings and don't be afraid to step back a little if you go down a DSP settings path that doesn't pan out well.  The good thing is that it's only time, and the experience gained is invaluable.  (The bad news is that the wife may not be as patient as you are... :rolleyes: ) Whatever the case, the performance levels of the system that I've personally achieved are presently far beyond what any fool with a checkbook could achieve.  If I can achieve these things, others can, too.

 

Chris

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

44 minutes ago, Chris A said:

and understanding of what is occurring...

This is where I am lacking. I whish I really understood what I was doing and why. I started off as a Mechanical Engineer in school and switched to Computer Science. I only wish I had a bit of EE and signals background.

 

Am I correct about minimum phase off of the group phase and only try to correct things in those "flat" minimum phase regions? I think I'm starting to understand where the speaker ends and room takes over. Often my measurements with the same settings are not exactly the same. I've got a pretty "dead" room. Perhaps my treatments are not in the optimal location.

Link to comment
Share on other sites

13 hours ago, rplace said:

This is where I am lacking. I whish I really understood what I was doing and why. I started off as a Mechanical Engineer in school and switched to Computer Science. I only wish I had a bit of EE and signals background.

That's pretty much my background subject areas in terms of formal education. 

 

It's pretty straightforward-- reading and absorbing, asking questions.  Having a STEM or business analysis background helps, but its certainly not required.  Just like formal education, it's usually individual interest and drive that is the difference in absorbing the material, not sitting through lectures, etc.  Interest is the key factor--there are few people walking around with formal education in this particular subject matter (using DSP crossovers in small listening rooms, and  acoustic/psychoacoustic theory).  

 

13 hours ago, rplace said:

Am I correct about minimum phase off of the group phase and only try to correct things in those "flat" minimum phase regions? I think I'm starting to understand where the speaker ends and room takes over. Often my measurements with the same settings are not exactly the same.

Generally you're trying to collapse down the excess phase and excess group delay in the measurements via selection of crossover filter types and careful setting of channel delays  (and positioning the drivers/horns as close together as possible to each other). This is the so-called "minimum phase condition".

 

Measurements are sensitive to early reflections in--room, which is the reason why I recommend lot of absorption during the measurements, and finding a good microphone position and staying with that position--works.  The "dense mode region" of acoustic reflections above the room's Schroeder frequency (usually ~200 Hz) will lead to differences in measurements even though nothing ostensibly changed.  I'd pick slightly different microphone positions until you find something that gets a bit more stable measurements.  Just be aware that you don't want to be on fractional multiples of room boundary length, width, height...and symmetrical distances from two or more room boundaries with the microphone.  Even moving the microphone a couple of inches can make all the difference in the world.

 

You need to be aware of 1/4 wave microphone cancellations with floor, walls and even the ceiling (if you have 8' ceiling or lower) by calculating where those frequencies occur and not trying to EQ out frequency response dips at those 1/4 wavelength boundary cancellations.

 

Chris

  • Thanks 1
Link to comment
Share on other sites

  • 1 year later...
On 8/2/2018 at 12:28 PM, Chris A said:

Bump...

 

Chris - First of all, thanks for taking the time to break this down for people like me! REW is powerful software in the hands of an expert,  but to me it is daunting.  So I am starting from scratch on a tri-amp speaker setup and want to get the order correct for measuring and programming a digital active crossover. In my case I am using REW on Windows 10, UMIK-1, MiniDSP 4x10HD, pre-amp, three power amplifiers, and La Scala speakers. Hopefully this will evolve into a check list at some point, which is why it's written this way. Any help here is tremendously appreciated! 

 

*Initial measurements*

1.1) Load UMIK-1 calibration file. (Not the 90 degree file)

1.2) Place UMIK-1 microphone 1 meter from speaker, at the level of the horn and tweeter. Aim mic at 30-45 degree upward angle. 

1.3) Measure speaker's SPL with all three drivers powered (sub,mid,tweeter) and without any crossover or EQ settings in place. In REW - 'Pink Noise' generator use 'Full Range' setting. This measurement should be taken with the SPL meter's 'Z' weight?? Increase volume until 75dB is reached, then mute all drivers.

1.4) Turn on one driver at a time and initiate sweep. Each sweep should be 20Hz - 20000Hz? 512k length. No timing reference needed? Should mic channel be the same as the speaker??

1.5) Save all of these sweeps, I use the 'Save all' feature that creates an .mdat file.

 

 **REW EQ Filters** 

2.1) Load each sweep into REW 'EQ' filter. Choose the correct Eq, or generic, from the first menu. 

2.2) Set speaker type to 'Full Range' for each driver in 'Target Settings'. Leave all other settings alone? Click 'Match response to target'. 

2.3) I have no idea what to do in 'Filter Tasks' ... so I use 20Hz -20000Hz in 'Match Range' and leave the other settings alone. Finally I press 'Match Response to Target'. If no gripes from REW, click 'Save filter coefficients to file'. Do this for each driver.

2.4) Load the files that were created in step 3 into active crossover. Mine happens to be a MiniDSP 4x10HD, details below.

 

 ***MiniDSP PEQ Import***

3.1) Load the corresponding files into the correct channel in the OUTPUT EQ 'PEQ' settings, using the advanced section. Leave Xover settings alone.

3.2) The mic should still be placed per step 1.2 at this point??

3.3) Using the REW pink noise generator (see step 1.3) adjust gain of each MiniDSP OUTPUT channel so that their SPL levels match. This is done one at a time.

 

 ****Next Step****

4.1) This is all I have so far...  

Link to comment
Share on other sites

TexDrone,

Before your neighbor Chris A replies, may I suggest some changes in the initial settings measurements.

 

When I did this for my KHorns, I was very concerned that I might fry a driver, so before I started with the KHorns I hooked up a pair of old AR bookshelf speakers, one to the miniDSP channel I had designated as bass and one to the channel I had designated as mid and familiarised myself with the miniDSP and confirmed that there would be no earth loops or loud clicks. I continued to do this when I started doing the crossovers as well.

 

Suggestions on Initial measurements -

1.3 - I suggest that you do this using the passive crossover to get a base measurement of the speaker response.

After you place the microphone, measure the height above the floor and the distance from the speaker so it is easy to measure later from the same place.

1.4 Use REW to limit the frequency - Bass 20 ~ 2,000, Mid 200 ~ 20,000, Hi 2,000 ~ 20,000.

Save each sweep.

 

Here is the write up I did, it needs some editing and fixing up, but starting January 27 I added pdf files where I went into a lot more detail.

https://community.klipsch.com/index.php?/topic/195572-bc-dcx464-as-the-mid-hi-drivers-for-a-k-horn/&ct=1620012789&_fromLogin=1

 

  • Thanks 1
Link to comment
Share on other sites

5 hours ago, Wirrunna said:

TexDrone,

Before your neighbor Chris A replies, may I suggest some changes in the initial settings measurements.

 

Suggestions on Initial measurements -

1.3 - I suggest that you do this using the passive crossover to get a base measurement of the speaker response.

After you place the microphone, measure the height above the floor and the distance from the speaker so it is easy to measure later from the same place.

1.4 Use REW to limit the frequency - Bass 20 ~ 2,000, Mid 200 ~ 20,000, Hi 2,000 ~ 20,000.

Save each sweep.

 

Here is the write up I did, it needs some editing and fixing up, but starting January 27 I added pdf files where I went into a lot more detail.

https://community.klipsch.com/index.php?/topic/195572-bc-dcx464-as-the-mid-hi-drivers-for-a-k-horn/&ct=1620012789&_fromLogin=1

 

 

Chris has helped me so much I am halfway expecting an invoice. 😁 In fact I may be wearing out my welcome! I am trying my best to document the basic steps on setting up an active 3 way crossover so that I don't have to ask Chris the same questions again next time... and hopefully it will help someone else later on. 

 

Thanks for the input, that is exactly the type of info I am looking for. Also, feel free to fill out section 4.2 thru the final section. 😅 

Link to comment
Share on other sites

Generally, the comments below can be found in the existing REW tutorial in pdf format.
 
Here are my specific comments based on your questions and process steps, above:
_________________________________________
"1.3)...This measurement should be taken with the SPL meter's 'Z' weight??" (Yes.)
 
You can take a sweep at higher SPL than 75 dB.  The 75 dB number is actually a minimum level, and in my experience, it needs to be taken at a higher level.
 
I actually recommend taking initial sweeps at 90 dB @1m on-axis.  The 75 dB number comes from those people using direct radiating loudspeakers that experience significant amounts of compression and modulation distortion.  Sometimes you won't see problems in the measurement or amplification circuits until they are turned up a bit to let them "breathe" a little.
_________________________________________
 
"1.4)...Each sweep should be 20Hz - 20000Hz? (In general, yes.) 512k length. No timing reference needed?" (Yes, not at this point.)
 
Should mic channel be the same as the speaker??
 
I don't know what you're asking here.
_________________________________________
 
"2.1)...Choose the correct EQ, or generic, from the first menu."
 
I believe you are saying that you should select the correct DSP crossover model from the first drop-down menu within the REW EQ facility.  If using a Xilica, select the "XP2040" button, if using a miniDSP 2x4 HD, select the miniDSP 2x4 HD button, etc.
_________________________________________
 
"2.2) Set speaker type to 'Full Range' for each driver in 'Target Settings'. Leave all other settings alone?"
 
image.png.e24276f9998d897959eace37c6e87427.png
 
The default values for the EQ facility need to be set within the "Preferences:Equaliser" menu from the main menu.  "Full Range" should be chosen, the Crossover HP and LP type should be set to "None", and the room curve set to a starting value of 200 Hz and a slope of 0.4 dB/octave...or zero if you're not using room curves, which is a good starting point.
 
The "target value" setting is extremely important--it sets the default overall SPL level that REW will try to EQ to tell the application where to stop EQing. 
 
******[Before starting REW EQ optimization, you need to set the smoothing of the measured SPL response to something like "Psychoacoustic" or "1/6th octave" smoothing in the main window, under the "Graph" menu.  I'd set Psychoacoustic or 1/6th octave smoothing as the default value in the Preferences menu.  Otherwise, REW will try to correct all the little peaks and dips of the measured SPL response, to no avail.  You really have to smooth the curve first before trying to find the PEQ filters to flatten the response.]*******
 
The "Target Level (dB):" control moves the"room curve up or down to lie on top of the measured SPL response.  It's best to set this to match within 0.5 dB of a "happy medium" of SPL values.  By "happy medium", I mean that the target curve is matching or somewhat lower than most of the peaks in response is applying attenuating filters to the response--instead of boosting filters (more on the "why" of this later).  Otherwise, you're going to be using your DSP crossover PEQ filters to boost overall response, and that will generally run you out of overall PEQ boost range before the job is done--which is something to avoid. In general, set this value to maximize the level of overlap of the room curve to the measured SPL response of the driver or drivers taken beforehand.  You'll get a feel for this after using it a little.  If REW generates too many PEQs to correct the SPL response, you run out of PEQs before the job is done.  The ear can't hear those small perturbations of the raw measurement anyway. 
 
_________________________________________
 
"2.3) I have no idea what to do in 'Filter Tasks' ... so I use 20Hz -20000Hz in 'Match Range' and leave the other settings alone."
 

(Here is a screen shot of the "Filter Tasks" dialog:

 

image.png.e46a7c23b6d41d9721fe1b5a6a1b9ec3.png

 
The "Match Range" settings determine the highest and lowest frequencies that REW will try to correct to the target curve level (the "room curve", above). 
 
If flattening the response of a woofer, then I'd set these two values to something like 10 Hz and 500 Hz--the upper limit being a little above the low pass crossover point of the woofer to the midrange/high frequency drivers.
 
For midrange, I'd set the two values to be just lower than the high pass frequency and just above the low pass frequency of the midrange crossover points, something like 300 Hz and 8000 Hz, etc.
 
For tweeter, set the first value to the high pass value (usually around 1-8 kHz), and leave the upper value at 20 kHz.
 
The default values of "Individual Max Boost" and "Overall Max Boost" are good--you can leave them where they're at.
 
The next two check boxes are also good to leave at their default values ("Allow narrow filters..." and "Vary max Q...").
 
The "Match Response to Target" is the trigger to start REW to optimizing PEQs.  Once you punch this link, REW will optimize the PEQs for the measurement selected, based on the target values and guidelines you set above.  If REW tries to create too many PEQs, you should select the "Reset filters for current measurement" to reset all the calculated PEQs, then usually increment the "flatness target" link upwards by one dB, then try again.  If you've forgotten to smooth the measurements to 1/6th octave or Psychoacoustic smoothing, you'll find out at this point that you forgot to do that.  Reset, smooth, and try incrementing the flatness target upwards to reduce the number of PEQ filters generated to a reasonable number (usually between 1-->6 filters total for each driver/amplifier channel).
_________________________________________
"2.4) Load the files that were created in step 3 into active crossover. Mine happens to be a miniDSP 4x10HD..."
 
When you've achieved a reasonable number of PEQs generated, and the optimized response simulated by REW shows a reasonable flat resulting response, then you move on to exporting these into your DSP crossover--which is a semi-automatic operation for miniDSP crossovers, and manual for other brands of DSP crossovers--worry not, it only takes a few moments to transfer the REW-optimized PEQs into your crossover's application to program it for the channel/measurement you're working on.
_________________________________________
 
"3.2) The mic should still be placed per step 1.2 at this point??"
 
Yes. Don't move the microphone between successive iterations, unless you wish to multiply the work you have to do.  In general, once you find a good place to locate the microphone, put a little piece of tape down on the floor on centerline with the center pole of the microphone stand to quickly be able to locate the correct position of the microphone--in case you inadvertently move the microphone between process steps.
_________________________________________
 
"4.1) This is all I have so far..."
 
This is where you combine the output of all drivers/horns together to fine tune the individual channel gains for flat response, then you can begin to time align the drivers.  I don't use loop-back within REW to set the channel delays.  Instead I use all drivers playing actively during the measurement sweep, then adjust the output based on the resulting spectrograms and the group delay (actually "excess group delay") plot.  More on that subject can be found here:
 

 

The subject of setting the crossover filters to achieve zero phase growth through the crossover interference bands can be found here:

 

https://community.klipsch.com/index.php?/topic/182419-subconscious-auditory-effects-of-quasi-linear-phase-loudspeakers/page/4/&tab=comments#comment-2388972

 

Chris

  • Like 3
Link to comment
Share on other sites

7 hours ago, Chris A said:
"1.4)...Each sweep should be 20Hz - 20000Hz? (In general, yes.) 512k length. No timing reference needed?" (Yes, not at this point.)

 

Should mic channel be the same as the speaker??
 
I don't know what you're asking here.

 

In the REW 'Preferences:Soundcard' menu there is an option for L or R under the input.

 

8 hours ago, Chris A said:
The default values for the EQ facility need to be set within the "Preferences:Equaliser" menu from the main menu.  "Full Range" should be chosen, the Crossover HP and LP type should be set to "None", and the room curve set to a starting value of 200 Hz and a slope of 0.4 dB/octave...or zero if you're not using room curves, which is a good starting point.
 
The "target value" setting is extremely important--it sets the default overall SPL level that REW will try to EQ to tell the application where to stop EQing. 

 

My Crossover HP and LP types were set to something different, I corrected that.

 

8 hours ago, Chris A said:
"2.4) Load the files that were created in step 3 into active crossover. Mine happens to be a miniDSP 4x10HD..."
 
When you've achieved a reasonable number of PEQs generated, and the optimized response simulated by REW shows a reasonable flat resulting response, then you move on to exporting these into your DSP crossover--which is a semi-automatic operation for miniDSP crossovers, and manual for other brands of DSP crossovers--worry not, it only takes a few moments to transfer the REW-optimized PEQs into your crossover's application to program it for the channel/measurement you're working on.

 

Just to be clear, these PEQs should go into the OUTPUT for each driver, correct?

 

8 hours ago, Chris A said:
"4.1) This is all I have so far..."
 
This is where you combine the output of all drivers/horns together to fine tune the individual channel gains for flat response...

 

At this point should the mic be moved to the listening area?

 

 

Thanks again Chris! 

 

Link to comment
Share on other sites

41 minutes ago, TexDrone said:

In the REW 'Preferences:Soundcard' menu there is an option for L or R under the input.

That's the input channel for the microphone--left or right channel from the viewpoint of the computer.

 

41 minutes ago, TexDrone said:

Just to be clear, these PEQs should go into the OUTPUT for each driver, correct?

Yes--output channel PEQs.

 

The only times that I use the input channel PEQs is when:

 

1) I need to correct the SPL across the crossover interference bands between two drivers/ways of the loudspeaker.

2) when I begin to run out of output channel PEQs, and need additional PEQs to do fine tuning of the SPL response.  Note that the input channel PEQs cannot correct single drivers/channels, but rather the summed (total) output of the loudspeaker --like the old preamp tone controls.

 

41 minutes ago, TexDrone said:

At this point should the mic be moved to the listening area?

Never--unless you're trying to adjust subwoofer response at the listening positions (i.e., well into the "modal region" of the frequency response where the longest dimension of the room is shorter than a half wavelength of sound, e.g., anything below ~30-50 Hz for regular-sized home hi-fi listening rooms). The problem with trying to take measurements at the listening position is separating out the effects of in-room reflections (which cannot be EQed--i.e., non-minimum phase acoustic behavior)--which apparently cannot actually be done correctly.  Even the more expensive "room correction software" applications such as Dirac, etc. cannot seem figure out what is direct arrival acoustic energy from the loudspeakers (i.e., "minimum phase") vs. in-room reflections (i.e., non-minimum phase), and they all try to boost response cancellations (nulls) when trying to take measurements from the listening position.

 

Chris

  • Thanks 1
Link to comment
Share on other sites

1 hour ago, Chris A said:
2 hours ago, TexDrone said:

At this point should the mic be moved to the listening area?

Never--unless you're trying to adjust subwoofer response at the listening positions

 

Ok, so the mic stays 1 meter away, and on axis with, the speaker being measured until the PEQ and crossovers are setup... also during the time delay measurements? 

  • Like 1
Link to comment
Share on other sites

8 hours ago, TexDrone said:

Ok, so the mic stays 1 meter away, and on axis with, the speaker being measured until the PEQ and crossovers are setup... also during the time delay measurements? 

 

Yes. The microphone doesn't move.  The dialing-in process starts with flattening the SPL response of the individual drivers/horns, then progresses to time alignment of the drivers using full-range sweeps of all the drivers together...with the crossover regions between drivers also set to their desired or natural crossover frequencies (determined via REW measurements).  Finally, the SPL response across the crossover regions (as well as setting the driver polarities) is the last portion of the dialing-in of each loudspeaker. 

 

Once this is done for one loudspeaker, the time delay and crossover filter settings (if any) can be transferred to the other stereo loudspeaker of the exact same driver/horn configuration--to save time.  Then the other loudspeaker must be checked via REW measurements with the microphone moved to be 1 m in front of the loudspeaker under test, then the EQ, delays, and individual driver channel gains adjusted as necessary to complete the individual loudspeaker dialing in process.

 

In a multichannel array (5.1, etc.) the settings can be cloned for the other channels if the loudspeakers are of the same design/model (an ideal case that very rarely, if ever, occurs).  In practice, since the center and surround loudspeakers are usually of a different configuration, they must be individually dialed in. 

 

After all the loudspeakers in the surround array are dialed-in individually, the 5.1/7.1/etc. channel gains and delays are set using something like a preamp/processor...or the DSP crossovers themselves.  Then the microphone moves to the listening position(s) to adjust ONLY the L, C, R, SR, SL, sub, (etc.) channel gains and overall loudspeaker delays in order to achieve a time aligned and SPL balanced array of loudspeakers. 

 

So this is a continuous process that first uses the microphone in one position per loudspeaker, then if the loudspeaker array is multichannel, the microphone is moved to the listening position to set only the surround loudspeaker channel gains and delays (i.e., 5.1 delays)--NOT to reset the EQ of individual drivers.

 

Chris

  • Thanks 1
Link to comment
Share on other sites

6 hours ago, Chris A said:

Yes. The microphone doesn't move.  The dialing-in process starts with flattening the SPL response of the individual drivers/horns, then progresses to time alignment of the drivers using full-range sweeps of all the drivers together...with the crossover regions between drivers also set to their desired or natural crossover frequencies (determined via REW measurements).  Finally, the SPL response across the crossover regions (as well as setting the driver polarities) is the last portion of the dialing-in of each loudspeaker. 

 

Thanks Chris! I am going to give it another go this afternoon.

Link to comment
Share on other sites

On 6/14/2021 at 10:50 AM, Chris A said:
"1.3)...This measurement should be taken with the SPL meter's 'Z' weight??" (Yes.)
 
You can take a sweep at higher SPL than 75 dB.  The 75 dB number is actually a minimum level, and in my experience, it needs to be taken at a higher level.
 
I actually recommend taking initial sweeps at 90 dB @1m on-axis.  The 75 dB number comes from those people using direct radiating loudspeakers that experience significant amounts of compression and modulation distortion.  Sometimes you won't see problems in the measurement or amplification circuits until they are turned up a bit to let them "breathe" a little.

 

More questions...

 

1) In the REW 'Preferences:Soundcard' menu, I have been using the 'Check Levels' using 'Use Main Speakers...' setting in the drop down. I then turn on all the drivers in my crossover for the speaker I am measuring. When I get a dB(Z) of ~76 I close the level checker. Is that the correct way to start off?

 

2) I then start taking the individual sweeps. Before each sweep I use the 'check level' button in the sweep menu. If a level is lower than 73dB I raise the main level on my preamp to get it at around 76dB. Is that the correct way to take sweeps?

 

3) Another question is about the warnings I sometimes get while processing a sweep in the EQ section. Let's say I run a sweep on the sub, 20Hz - 20000Hz like I have been doing on all the drivers. I load it in to the EQ processor of REW, click on 'Calculate TL from response level' (which it does) then I set a range of 20Hz - 500Hz in the 'Match Range' settings in the 'Filter Tasks' section. When I press 'Match Response to Target' I get an error about a high percentage of the 'match range being above the target'. Should I manually adjust the TL at this point? I suspect REW uses the entire sweep to get an average TL, but when I select a smaller range to EQ it throws everything off...

 

This is what I got sweeping each driver 20dB - 20000dB with no crossover settings:

LS_Plot.thumb.JPG.6dcd05ae3d6168bb8ff004fbbce7a535.JPG

 

I dinked around and ran some sweeps with limited ranges, then EQ'd them, and got this. 

Filters_limited_range.thumb.JPG.74d0716a6a0d9bc5ddf498297ef69545.JPG

 

 

 

 

Link to comment
Share on other sites

19 hours ago, TexDrone said:

1) In the REW 'Preferences:Soundcard' menu, I have been using the 'Check Levels' using 'Use Main Speakers...' setting in the drop down. I then turn on all the drivers in my crossover for the speaker I am measuring. When I get a dB(Z) of ~76 I close the level checker. Is that the correct way to start off?

 

2) I then start taking the individual sweeps. Before each sweep I use the 'check level' button in the sweep menu. If a level is lower than 73dB I raise the main level on my preamp to get it at around 76dB. Is that the correct way to take sweeps?

You can do it that way--no issues.  Like I said above, I usually take sweeps at 85-90 dB, but that's entirely up to you.  If you've got noisy appliances (or barking dogs), etc., in the room, taking measurements at a higher level will significantly decrease the effects of those type of incident noises.

 

19 hours ago, TexDrone said:

3) Another question is about the warnings I sometimes get while processing a sweep in the EQ section. Let's say I run a sweep on the sub, 20Hz - 20000Hz like I have been doing on all the drivers. I load it in to the EQ processor of REW, click on 'Calculate TL from response level' (which it does) then I set a range of 20Hz - 500Hz in the 'Match Range' settings in the 'Filter Tasks' section. When I press 'Match Response to Target' I get an error about a high percentage of the 'match range being above the target'. Should I manually adjust the TL at this point? I suspect REW uses the entire sweep to get an average TL, but when I select a smaller range to EQ it throws everything off...

 

You can reset the "target level" (TL) , up or down, to avoid the issues that ensue from setting the level too high or too low.  If REW complains, sometimes it sees a lot of boosting or attenuation about to occur, and it's making sure that you want to do that.  If so, just click "ok" to proceed.

 

If you select a frequency range that's too small, you'll end up with portions of the loudspeakers' response that's still not flat.  Usually, I find that setting the frequency limits of EQ is fairly intuitive, so I usually just iterate the frequency limits until I achieve what I want.  Sometimes, I even add PEQs by hand (manually) to get what I want, especially on the ends of the frequency response spectrum, where REW's EQ facility won't boost the response when I desire that it will boost response.

 

19 hours ago, TexDrone said:

This is what I got sweeping each driver 20dB - 20000dB with no crossover settings:

Generally speaking, one of the first things I do is to set the vertical and horizontal scales on the plots so that I can see what I'm doing.  The plots you show above have the vertical scale magnitude way too zoomed out.  I can't see anything from those plots you posted, to be honest.

 

I try to use a vertical scale so that I can see at least 2 dB increments in the vertical scale bar.  You set the vertical scale by moving the cursor into the plot, and then a little set of "+" and "-" icons show up near the top of the vertical scale in the plot area.  You click on the "+" icon to zoom in until the vertical scale shows 1 or 2 dB minor increments. 

 

Chris

Link to comment
Share on other sites

  • 1 month later...
On 6/15/2021 at 1:50 AM, Chris A said:

I actually recommend taking initial sweeps at 90 dB @1m on-axis.  The 75 dB number comes from those people using direct radiating loudspeakers that experience significant amounts of compression and modulation distortion.  Sometimes you won't see problems in the measurement or amplification circuits until they are turned up a bit to let them "breathe" a little.

 

The 90db sweep is not quite as socially acceptable as the 75db !

Seeks out the occasional rattle.

Link to comment
Share on other sites

  • 4 weeks later...

Problems with REW sound level ... no clue :( 

 

OK, thanks to @Chris A I made some progress. Have a XP-2040 hooked up to JBL 2446H horns via a preamp and a Yamaha M-45 amp (120wpc). (Later I'll do the low freq side.) 

 

So, ran a series of measurements a couple of days ago and averaged them. 3 minor EQ filters which I entered into the Xilica. Then, ran a new test measurement ... for unknown reasons (I had not changed ANY settings), the REW measurements now were about 20dB's louder. And; now it required 7 filters :(  ChrisA suggested that the SPL mismatch might be the problem. (Haha; also had the mic in the wrong position; at listening position vs center of speaker.)

 

Therefore, ran more measurements today with the mic in the correct position. "Psycho'd" these and averaged them. Curves looked nice and when I EQ'd them  (at flatness "2") I only got ONE SMALL filter from REW. :) Put the value into the Xilica :)  Then, ran a measurement to see if my changes were successful. NO; first measurement showed something like "excess headroom" and I had to lower the volume. OK; did so and got the measurement - turned out to be about 25dB louder (even after turning it down). When I EQ'd it, it now showed SEVEN major filters :( Obviously the high SPL levels (120+dB) screwed up the results. :(

 

NO CLUE why my SPL recording level is changing :( Did not even unplug my PC between initial and final measurements :( 

NO CLUE how to set SPL level ... tried it with the SPL meter, but it doesnot seem to correlate with my measurement SPL :( 

 

Appreciate your input.

 

Thanks, Emile 

Link to comment
Share on other sites

26 minutes ago, Emile said:

Problems with REW sound level ... no clue :( 

 

OK, thanks to @Chris A I made some progress. Have a XP-2040 hooked up to JBL 2446H horns via a preamp and a Yamaha M-45 amp (120wpc). (Later I'll do the low freq side.) 

 

So, ran a series of measurements a couple of days ago and averaged them. 3 minor EQ filters which I entered into the Xilica. Then, ran a new test measurement ... for unknown reasons (I had not changed ANY settings), the REW measurements now were about 20dB's louder. And; now it required 7 filters :(  ChrisA suggested that the SPL mismatch might be the problem. (Haha; also had the mic in the wrong position; at listening position vs center of speaker.)

 

Therefore, ran more measurements today with the mic in the correct position. "Psycho'd" these and averaged them. Curves looked nice and when I EQ'd them  (at flatness "2") I only got ONE SMALL filter from REW. :) Put the value into the Xilica :)  Then, ran a measurement to see if my changes were successful. NO; first measurement showed something like "excess headroom" and I had to lower the volume. OK; did so and got the measurement - turned out to be about 25dB louder (even after turning it down). When I EQ'd it, it now showed SEVEN major filters :( Obviously the high SPL levels (120+dB) screwed up the results. :(

 

NO CLUE why my SPL recording level is changing :( Did not even unplug my PC between initial and final measurements :( 

NO CLUE how to set SPL level ... tried it with the SPL meter, but it doesnot seem to correlate with my measurement SPL :( 

 

Appreciate your input.

 

Thanks, Emile 

That's the happiest and most sad I have ever seen you in a single post.   Good luck, Chris A can help you sort it out surely. 

  • Haha 1
Link to comment
Share on other sites

17 hours ago, Emile said:

for unknown reasons (I had not changed ANY settings), the REW measurements now were about 20dB's louder.

So have you changed the gain of your preamp/integrated amplifier?  I have a digital readout on my AVP, which I typically set to a movie--loud position (-12 dB in my setup presently), and which is part of my checklist of things every time I pull out the UMIK-1 and REW to measure again.

 

Also, not moving the microphone between measurements is a pretty big deal, unless you're trying to average measurements around the on-axis measurement, in which case, you'd move the microphone around by perhaps 6-12 inches or so, then average the measurements.  This is much more time consuming, however, than using just the on-axis measurement.

 

Chris

  • Like 1
Link to comment
Share on other sites

1 hour ago, Chris A said:

So have you changed the gain of your preamp/integrated amplifier?

Thanks again @Chris A 

NO; did NOT change any preamp/amp gain between the initial measurements (which came out at about 90dB) and the (attempted) check on my changes. And did NOT change the mic position. Haha, have NO CLUE. When I did the follow-up check the volume was much louder ... initially got a clipping error ... then lowered the volume substantially via  my preamp ... the REW result was still over 120dB :(  

 

Again, had made NO volume changes. Only unplugged mic and sound from PC. And, this is the second time it happened between initial measurements and checking the results :(  Spend a couple of hours reading the REW help files, but no obvious info.

 

OK; next time I will repeat this process keeping the PC connected :) 

 

Graph shows results ... 

 

REW_problem.jpg

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...