Home Theater Forum and Systems banner

1 - 20 of 21 Posts

·
Registered
Joined
·
16 Posts
Discussion Starter #1
Hi there,

I'm new to REW, have read the basic help + tut's and trialled the software but having a little trouble getting any results. Basically, I'm having trouble joining the dots and would like input from anyone spell out the steps to set eq filters in a second software app via a plugin

Lets start with my setup and what I want to do;
- I'm using MC17 as my Media center
- I have a 5.1 Setup with a MAudio Profire 610 as the audio device (8 analog outs going direct into a poweramp)
- I have Voxengo GlissEq as a plugin in MC17 which is 5.1 capable and wish to use this to EQ my setup with its parametric eq's.
- I also have a Behringer ECM8000 to measure results which I can plug into the Profire 610 mic preamp
- I expect, I should be able to apply filters in Glisseq (manually) to correct my speakers and room and wish to use REW to determine what filters and parameters to apply.

How am i going so far?


In theory this is what I think I'm meant to be doing to achive this.
1. Measure a said speaker in REW - eg Centre (not quite sure how to)
2. Have REW determine the required filters (not quite how to)
3. Manually load these into GlissEQ for the relevant channel (No issues there)
4. Repeat the process until all speakers have custom filters set (iterate)

Have I got the process right or is there a better way?
Any tips on the steps required to do in REW itself as im pretty comfortable working MC17 and GlissEq:wave:

Any advice appreciated.

Cheers,

bassmann
 

·
Premium Member
Joined
·
1,400 Posts
bassmann said:
Any advice appreciated.
( Advice ?? ) Yes , allow yourself about half a year ( with about 3 hours a week of applied study time ) to truly get competent/familiar with all the concepts ( that you just managed to fly by ) .

You can start this ( fairly lengthy ) process by creating ( and posting a screen capture pic ) of a soundcard calibration ( captured by you ) for your Profire 610 ( details on how to do create a SC calibration are found within the REW software ) .

:sn:
 

·
Super Moderator
Joined
·
9,273 Posts

Hey bassmann,

In theory this is what I think I'm meant to be doing to achive this.
1. Measure a said speaker in REW - eg Centre (not quite sure how to)
2. Have REW determine the required filters (not quite how to)
3. Manually load these into GlissEQ for the relevant channel (No issues there)
4. Repeat the process until all speakers have custom filters set (iterate)

Read more: REW basics for 3rd party software eq - Home Theater Forum and Systems - HomeTheaterShack.com
You basically have the right idea. But first, did you have a custom calibration done for your mic? You’ll need one if you intend to EQ main channel speakers. The ECM doesn’t have flat response, and any measurements you take with REW will be a combination of the room response + the mic’s. Obviously, you don’t want to employ any full-range equalization based on inaccurate measurements.

2. Have REW determine the required filters (not quite how to)
REW only determines filters for bass measurements for subwoofers. For main-channel speakers, you’ll have to do the filters yourself.

Regards,
Wayne
 

·
Registered
Joined
·
16 Posts
Discussion Starter #4
Ok guys,

I've had a crack at this.
- I've created a soundcard calibration file at 96kHz, 24bit using ASIO drivers for my Profire610 (this is how I run it typically).
- I've downloaded the mic calibration file for the ECM8000. Wayne I'll go back and create a custom one for my MIC when i get a chance.
- I've set levels as help files suggest
- Then I've taken measurements for my speakers.

I've attached my soundcard cal and the centre speaker measurement graph
- The soundcard cal looks good to me. Any comments
- Any comments on the response of the centre seapker measurement?

My Thoughts are;
1) With no EQ applied there's a natural rolloff, maybe 12db from 100Hz to 20kHz, but this steepens after 6k so I might want to boost after 6kHz with a shelve
2) There's mid range drops between 1.5 and 3kHz which I might want to boost since its a centre speaker and in the speech domain
3) Lower mids could also benefit from a boost in the 250-400Hz range
4) Theres a couple of bass peaks to tame at 95 and 200Hz, the sprectrgram shows this well. Is the aim to pull out the yellow and red spots?
5) I'm not too concerned below 100Hz since its a centre speaker

The only thing I'm curious about is the dynamic range im seeing here with a max of 78db at 95Hz and a min of say 10db at 6k. Is this typical for a non-eq'd speaker or should I be concerned?

Bassmann
 

Attachments

·
Registered
Joined
·
16 Posts
Discussion Starter #5
Guys - I'm starting to get the hang of this (i think).
I've developed some eq filters for my centre speaker and am just trying to dial into gliss eq.
I'm having a little trouble transposing the REW q value into glisseq b/w/ parameter.

I've found in REW, that increasing the q value narrows the bandwidth, however in glisseq increasing b/w increases the bandwidth of a said frequency.

I suspect this is because the UOM is different (maybe a reciprical like 1/Q, just guessing since the relationship appears inverse). If so, is there a formula to convert. e.g. What should a Q value of 14 equal in the Glisseq B/W parameter?

Bassmann
 

·
Super Moderator
Joined
·
9,273 Posts

I've attached my soundcard cal and the centre speaker measurement graph
- The soundcard cal looks good to me. Any comments
- Any comments on the response of the centre seapker measurement?
Both look perfectly normal. :T


My Thoughts are;
1) With no EQ applied there's a natural rolloff, maybe 12db from 100Hz to 20kHz, but this steepens after 6k so I might want to boost after 6kHz with a shelve
2) There's mid range drops between 1.5 and 3kHz which I might want to boost since its a centre speaker and in the speech domain
3) Lower mids could also benefit from a boost in the 250-400Hz range
4) Theres a couple of bass peaks to tame at 95 and 200Hz, the sprectrgram shows this well.
Basically, you have the right idea. :T However, it looks like the actual upper-end problem would be pulling down the ~3-8kHz range, not boosting up on either side of it.

That said, don’t do any equalization until you get a calibrated mic.

What orientation did you use for the mic? It should be point at the speaker, angled slightly upward (about 20°).


The only thing I'm curious about is the dynamic range im seeing here with a max of 78db at 95Hz and a min of say 10db at 6k. Is this typical for a non-eq'd speaker or should I be concerned?
What you’re seeing above 400 Hz or so is comb filtering, caused by reflections in the room. Switch to 1/3- or 1/6-octave smoothing to see the underlying trend in response.


I've found in REW, that increasing the q value narrows the bandwidth, however in glisseq increasing b/w increases the bandwidth of a said frequency.

I suspect this is because the UOM is different (maybe a reciprical like 1/Q, just guessing since the relationship appears inverse). If so, is there a formula to convert. e.g. What should a Q value of 14 equal in the Glisseq B/W parameter?
Impossible to say without knowing what bandwidth designation Glisseq is using. The two most commonly used are “Q” and “octaves,” wherein a Q factor of say, 2.1 is equivalent to 2/3-octave. Here’s a handy conversion chart.


Equalizer Q and Octave Conversions reduced.jpg


Regards,
Wayne
 

·
Registered
Joined
·
16 Posts
Discussion Starter #7
Hi Wayne,

What orientation did you use for the mic? It should be point at the speaker, angled slightly upward (about 20°).
I did have it pointing slightly upward but note I took the measurements from the couch, the main listening position in the room rather than directly infront of the speaker as to account for the rooms response.

When taking the measure, I used the 1M sweeps 8 times. During this I moved the mic to the left and right of the main listening position to get an average of this location.

Is this a reasonable way to measure?


Switch to 1/3- or 1/6-octave smoothing to see the underlying trend in response.
Yes this works well and somewhat changes my interpretation
 

·
Registered
Joined
·
16 Posts
Discussion Starter #9
Ok - been playing around and starting to get some results which I've put into Glisseq.

I've got a DEQ2496 which I plan to use on the mains. I'm thinking of using its para's (10 per channel) for <200Hz to tame the bottom end which REW can work out for me and then using the graphic on frequencies above that, probably autoeq.

I did a trial with the para's and found the results clearer but lacked punch EQ'ing <200Hz. I assume this is because flat doesn't sound great and I'll need a house curve at some point.

On that, I've been reading up and I believe the process to setup my mains to be something like this.
1. Measure the response of each speaker with REW
2. Eq the speakers flat with a combination of para and GEQ on the DEQ2496
3. Re-measure with REW to confirm results
4. Develop a house curve for the room. I'm thinking doing this by checking the perceived levels in octave intervals. Then write a house curve
5. Repeat steps 1-3 with a house curve applied

Will this work? Have I missed anything?
What kind of signal should I use for step 4 e.g. pink noise?

BTW - here a measurement of my mains (left followed by right).



Thanks,

Bassman
 

Attachments

·
Registered
Joined
·
16 Posts
Discussion Starter #10
On MIC calibration i've noticed the following;

I've loaded the ECM8000 cal file from HTS into REW when taking measurements and noticed the MIC cal curve increases at around 3kHz. Consequnetly my measurements decline around the same point which one might intrepret as my MIC has a different response and is more stable over 3kHz (see the first image I posted on 7th July).

I've now had a look into calibrated MIC's.

From what I can tell, there is no way of calibrating myself. Whilst my soundcard is reasonalbe flat, my speakers and room are not.

Therefore I could pay to get my $75 MIC calibrated or buy a new one. I expect calibration is not cheap, am I better off buying another one that's aleady calibrated? If so another ECM8000 or something else?

Bassmann
 

·
Super Moderator
Joined
·
9,273 Posts

What kind of signal should I use for step 4 e.g. pink noise?
That’s correct. I’d recommend parametric EQ over GEQ.


Therefore I could pay to get my $75 MIC calibrated or buy a new one. I expect calibration is not cheap, am I better off buying another one that's aleady calibrated? If so another ECM8000 or something else?
I think you can get a calibrated mic from Cross Spectrum for $85 or so (get the EMM6). Then you can sell your ECM on eBay. :T

Regards,
Wayne

 

·
Premium Member
Joined
·
6,711 Posts
When taking the measure, I used the 1M sweeps 8 times. During this I moved the mic to the left and right of the main listening position to get an average of this location.

Is this a reasonable way to measure?
Multiple 1M sweeps is much more tedious than necessary, single 256k sweeps are perfectly good for response measurement.
 

·
Registered
Joined
·
16 Posts
Discussion Starter #13
Hi Wayne,

I've ordered a calibrated EMM6.
Can also confirm I've determined GlissEQ b/w is in octaves so your conversion chart does the job well.

Now i've got the basics of the job at hand, I'll wait for my new mic and report results in due course.

Thanks for the help everyone has provided thus far.

Bassmann
 

·
Registered
Joined
·
16 Posts
Discussion Starter #14
It took some time, but I finally got my calibrated EMM6, created the MIC cal file and done some testing.

On my front left main speaker did the following;
1) Measured the response
2) Determined the filters in REW
3) Loaded the filters into my DEQ2496
4) Remeasured with filters to confirm results.

These were however, not what i expected. With filters applied a spike at around 85Hz is a little tamer but still presists. I also noticed the sound was a bit bright, this can be also seen in the spike at around 5kHz, there was a broad filter in this range but its driven more than expected.

- Is the bass gain at 85Hz likely to be reflected gain from the room rather than direct from the source which could explain why the filters didnt work in this range? What should I do?

- How does the rest of the response look. Between 150Hz and 5kHz the range is around 6dB which I expect is quite significant. Would a range of 3dB be a reasonable target to have in mind?

Pictures
- The first shows the measured and REW filters
- The second shows the overlay of the remeasured response with filters in yellow.
 

Attachments

·
Registered
Joined
·
16 Posts
Discussion Starter #16
The first picture shows the measured response in green. The calculated filters in blue

The second picture shows the overlay of the measured response in green and the remeasured response in yellow with filters applied
 

·
Super Moderator
Joined
·
9,273 Posts

I also noticed the sound was a bit bright, this can be also seen in the spike at around 5kHz, there was a broad filter in this range but its driven more than expected.
Don’t know how you ended up with the 5 kHz boost when that area really didn’t need any attention – or at least not that much. Whatever you did to cause that, undo it. :T

Looks like you severely overdid the 1 kHz area. Only needed ~4 dB cut, but you pulled it down twice that much. :(


- Is the bass gain at 85Hz likely to be reflected gain from the room rather than direct from the source which could explain why the filters didnt work in this range? What should I do?
Hard to say without knowing what filters / parameters you used. Looks like you used a fairly broad filter to pull the entire 40-200 Hz range, but you need another, tighter one to specifically address 90 Hz.

- How does the rest of the response look. Between 150Hz and 5kHz the range is around 6dB which I expect is quite significant. Would a range of 3dB be a reasonable target to have in mind?
Might be unrealistic, but you should make that decision with 1/3-octave smoothing, which is more representative of what you actually hear. That said, typically you get the best (read audible) results by just addressing the worst of the broad peaks and troughs.

Regards,
Wayne
 

·
Registered
Joined
·
239 Posts
I concur with Wayne on peak at 85-90Hz. With DEQ2496 add PEQ in that range with 1/3 octave BW or even less and gain of -7dB and see what that does to it. If that moves the peak you can use that as starting point for tweaking that PEQ.

BTW, red/green colorblind. Looking real close I see, but thanks for clarifying for me.

Andrew
 

·
Registered
Joined
·
16 Posts
Discussion Starter #19
Before I draw any conclusions, i've taken step back to re-calibrate the levels as they might be affecting results.

I've turned up the amp to normal listening levels for calibration (i assume they needn't be ear piercing).

With the EMM6
At this i'm getting SPL 100dB(C) but a low input signal of -50dB FS with the preamp turned up to the max.
If I increase the amp volume i get SPL 120dB with an input signal of -30dB FS

I also tried my ECM8000 and get the following;
SPL 100dB(C) and -40dB FS

Are these typical figures, if not what is?
Should the ECM8000 be more sensitive than the EMM6 by 10dB? Is this a problem?

I don't really believe I should need so much preamp gain, maybe I have a problem with either my cable or preamp.

Any thoughts?
 

·
Registered
Joined
·
16 Posts
Discussion Starter #20
Just did some more SPL meter checks with the EMM6 as follows (n.b. the amp was at a med/low level for the test);

1. Preamp gain halfway (12o'clock)
Ambient - SPL 40dB(C), Input -80dB FS
With Signal (i.e. calibrate) - SPL 60dB(C), Input -60dB FS

2. Preamp gain full
Ambient - SPL 100dB(C), Input -40dB FS
With Signal (i.e. calibrate) - SPL 120dB(C), Input -20dB FS

3. Preamp full gain with -20dB pad
Ambient - SPL 60dB(C), Input -60dB FS
With Signal - SPL 80dB(C), Input -40dB FS

My thoughts on this are;
- Aren't the ambient SPL's quite high for 2 and 3 (isnt a typical house SPL 40-50dB) which makes be think I'm adding unnessessary gain which may add distortion to results.
- The SPL of the signal seems to be 20dB (i.e. the difference between ambient and signal for each test). Is this too low for a reasonable test?
- The Input signals are way too low, ideally I need around around -20dB FS right?

On this basis I think option 3 is probably the best balance of SPL and Input but I'd need to drive my amp another 20dB which will be ear piecing at higher frequencies during the sweep and well above normal listening levels in the room.

Interested to hear views from others on this...Does my gear look ok or could I have a preamp or cable issue (hard to test because I don't have spares handy).
 
1 - 20 of 21 Posts
Top