REW Beta Release Confusion over combined speaker response

[Exponential]

Hi all.

I have recently sold all my passive gear in favour of an "active" set up and I would like some advice on the new measurements I have taken recently.
The speakers are a pair of Genelec 8341's and I've bought the GLM (Genelec Loudspeaker Management) software to go with them.
The GLM software took a while to get used to but I got the hang of it in the end.
It allows you to calibrate your speakers either in a symmetrical set up or individually, I chose individually as my room and speaker placement is certainly not symmetrical!
I'm running them with a minidspSHD which I will use to apply EQ and crossovers to my dual subs (running mono) and to also apply a crossover to my mains.

Before I even start trying to apply a crossover or select an adequate one for that matter, I am a little confused over the "calibrated" combined response of the 8341's.
There is a huge suck out in the crossover region which is a big concern for me.
I was under the impression that the GLM software sorts all that out for you but, thinking about it, I have measured them as individual speakers so I guess that could be the problem?

Anyway, I have attached the MDAT file if anyone is interested in attempting an answer.

Thank you.

Mike.

 

[sm52]

If you are talking about a pit at 100 Hz, then it so happened that the right and left sides at 100 Hz are in antiphase. Moreover, up to 100 Hz and after 100 Hz everything is fine. The subs are working fine.

 

[Exponential]

If you are talking about a pit at 100 Hz, then it so happened that the right and left sides at 100 Hz are in antiphase. Moreover, up to 100 Hz and after 100 Hz everything is fine. The subs are working fine.

Agreed, the two subs together are remarkably good considering there is no delay or EQ applied.
It's the mains I'm concerned about and how to fix it if it is even possible!

Once I've figured out how to fix the mains and that suck out, I then need to choose a suitable crossover.
How do I go about doing that?

 

[sm52]

It's the mains I'm concerned about and how to fix it if it is even possible!

This may remain so, unfortunately. I don't know the exact solution. Perhaps the problem is in the location of the mains between themselves and in relation to the walls and other reflective surfaces. You can try to move a little, a couple of centimeters back and forth, then measure what happened. But fixing the 100Hz pit can get the problem elsewhere. I am writing this so that you understand that this is a slippery slope and that you will have to stop at some option.

 

[Exponential]

This may remain so, unfortunately. I don't know the exact solution. Perhaps the problem is in the location of the mains between themselves and in relation to the walls and other reflective surfaces. You can try to move a little, a couple of centimeters back and forth, then measure what happened. But fixing the 100Hz pit can get the problem elsewhere. I am writing this so that you understand that this is a slippery slope and that you will have to stop at some option.

I know mate, it really is a minefield and it can cause you to lose your enjoyment of the whole experience because you're always second guessing yourself with it.

I've been fiddling about with REW for years and I still don't know how to correctly implement a good crossover.
I read an article the other day about using a FDW window to determine the correct phase but it went over my head to be honest.

 

[sm52]

The frequency response of your system is good today. There are 100 Hertz you would like to get out of the pit. So by and large you can relax and listen as it is. But, if the desire to beat 100 Hertz is strong, then you need to study the issue of using REW for

correctly implement a good crossover.

Read this https://www.avnirvana.com/threads/n...-understands-a-tittle-german.8730/#post-66869

 

[Exponential]

Goo

The frequency response of your system is good today. There are 100 Hertz you would like to get out of the pit. So by and large you can relax and listen as it is. But, if the desire to beat 100 Hertz is strong, then you need to study the issue of using REW for

Read this https://www.avnirvana.com/threads/n...-understands-a-tittle-german.8730/#post-66869

Good morning.

Thanks for the link! I appreciate it.
I definitely want to at least try to understand it all if nothing else.
It's half the fun trying to figure it out as, when you get it right, you have a sense of completion and satisfaction.

Have a great day!

Mike.

 

[DanDan]

Dips in the response are often due to non resonant reflections. BIR. Try moving your mains to almost touch the front wall.

 

[Exponential]

Dips in the response are often due to non resonant reflections. BIR. Try moving your mains to almost touch the front wall.

Hi DanDan.

Yeah, good point! As it happens, I've just been watching a video on SBIR and it struck a chord with me and my situation.

The next time I've got my measurement equipment out (which I hope is very very look soon), I'll try it out to see the result.

It'll be a great experiment to try out for a newbie.

 

[Exponential]

I just thought I would post my latest MDAT files if anyone cares to take a look.

I've settled on an 80hz crossover and have aligned my two subs so they are in phase with each other.
I then looked for a good summation between my phase aligned subs and my mains and that's where I stumble.
To be honest, I'm a little out of my depth and could do with a bit of guidance on how to proceed.
I've tried reading the various guides out there but I can't seem to be able to follow it all correctly??

Anyway, the MDAT files are attached.

 

[jtalden]

Here are different settings you may want to try:
Decrease the delay of the L SW by 3.5 ms
Decrease the delay of the R SW by 14 ms
Reverse the polarity of both SWs

If the SWs DSP delay settings will not allow these large delay reductions you can first add an equal amount of delay to all speakers (both mains and SWs) prior to making these changes.

• This new delay setting is the smoothest found and most favorable for SPL. It will allow for relative moderate PEQ through the XO range.
• The 2 SW are currently timed 10.5 ms apart from each other so that is why their delay changes are by different amounts. They now will be timed together and are providing nearly the same SPL as with the current setting.
• This setting has the SWs timing leading the mains by about 8 ms. so it is more favorable in reducing group delay (phase rotation). Phase tracking is still very good and better than other options for SPL. The more conventional timing resulted in more SPL issues.

 

[Exponential]

Here are different settings you may want to try:
Decrease the delay of the L SW by 3.5 ms
Decrease the delay of the R SW by 14 ms
Reverse the polarity of both SWs

If the SWs DSP delay settings will not allow these large delay reductions you can first add an equal amount of delay to all speakers (both mains and SWs) prior to making these changes.

• This new delay setting is the smoothest found and most favorable for SPL. It will allow for relative moderate PEQ through the XO range.
• The 2 SW are currently timed 10.5 ms apart from each other so that is why their delay changes are by different amounts. They now will be timed together and are providing nearly the same SPL as with the current setting.
• This setting has the SWs timing leading the mains by about 8 ms. so it is more favorable in reducing group delay (phase rotation). Phase tracking is still very good and better than other options for SPL. The more conventional timing resulted in more SPL issues.

View attachment 43349

View attachment 43350

View attachment 43351

Hi @jtalden

Thank you for taking the time to have a look at my rubbish measurements!

It is quite clear that my room is the limiting factor here and there is not much I can do about it at all.
@DanDan had a great suggestion of moving the main speakers closer to the front wall to see what the result would be as far as limiting SBIR but I haven't had the time to do that yet.

In order to allow me to have a play around myself without having to continually post here asking for help, would you mind telling me the process of how you got to where you did because it baffles my mind so I can learn!

I can get myself so far in doing measurements but then I fall down and get frustrated with myself for not knowing what to do!

If I had a guide, I could refer back to it each time I get chance to do a measurement (family and children permitting) and actually have half a chance of getting it right!

I hope I'm not asking too much here!

 

[jtalden]

Yes, moving the main speakers is excellent advice. The room is indeed creating a phase difference between L and R speakers centered about 100 Hz. This easy to see this by looking at the combined L+R measurements. The SPL chart below shows the problem clearly. All I did is align the impulses of the L and R measurements and then use REW alignment tool (or trace arithmetic) to sum them together. I then applied 1/3 smoothing to highlight the the problem area.

So try different mains positions until this situation is improved.

The timing recommendation I provided above only applies to the current setup and indeed there would be a big difference between panned left and right channels in this area. PEQ is not very effective in correcting this type of situation.

 

[Exponential]

Thank you for looking into this for me. I will not settle for this though and will experiment more to try to find a happy middle ground.
I will do a combination of moving the mains further back and/or altering the crossover point to see if that improves things at all.

Can I ask, when aligning your subs to be "as one", which is more favourable, phase alignment or SPL across the measured frequency response?
If I measure my subs together without any EQ or delay applied, I get a really nice, linear SPL from 14hz right through to around 70hz.
If I phase align them, the result is what you saw in the previous measurements.

 

[jtalden]

Can I ask, when aligning your subs to be "as one", which is more favourable, phase alignment or SPL across the measured frequency response?

I think good SPL support and smoothness is more important - so long as the resulting delay timing stays close to the 'preferred' timing...

I first align the impulses (or step responses) and then adjust timing a little as needed to get the closest direct sound phase tracking. If needed, we can use an FDW of maybe 2-6 cycles to better define the direct sound phase trace. The FDW is then removed to see the actual SPL response with this 'preferred' timing. I save this SPL trace and then adjust the timing a little to see if there is a better delay timing that significantly improves the SPL. I try not to change timing more than ±1/2 wavelength of the XO frequency. If a better setting for SPL is found in that range I recommend it.

Occasionally we can find a better SPL result with the 2 SW's sound arriving at the mic 1 or 2 wavelengths out of time. I don't recommend this timing based on my understanding of the tradeoffs, but if it sounds better to the user, that's fine with me.

If I measure my subs together without any EQ or delay applied, I get a really nice, linear SPL from 14hz right through to around 70hz.
If I phase align them, the result is what you saw in the previous measurements.

I don't understand this statement about the original delay setting of the SWs being smoother SPL than the aligned timing I suggested. Below is the data I was looking at.

The blue trace is an ideal 80 Hz LR-4 filter response. Green is my recommended SW timing. Red is your the original timing from Post-10. The original (red) trace does not have the 80 Hz XO applied so the comparison above 50 Hz is not as clear. Looking below 50 Hz I felt the 12 dB reduction of the peak at 40Hz was a significant advantage over the original timing. The 60 Hz peak may require a little more reduction, but I judged this to be a favorable tradeoff. The new timing should require less aggressive PEQ.

 

[Exponential]

Hi. @jtalden. Again, thank you for taking the time to explain your thought process and how you apply your knowledge to adjust the data as you see fit.

I've had a job cancelled today so I have a free day, I'm going to take the opportunity to try moving my speakers around in various positions to see the effect.
I will post the following MDAT data later on:

L Main on its own full range
R -----------------------------------
L Sub on its own full range
R ----------------------------------

All of the above will have a timing reference from the L Main.

Once we have a favourable L & R Main summation (hopefully, without the nasty phase issue), we can then look into selecting an appropriate crossover between 80-100hz.
That's assuming you don't mind helping me out.

By the way, the measurement of the two subs you saw in the above MDAT was a result of my trying to time align them but, I think getting a more linear SPL summation is probably the right way to go.

Do you feel a multipoint measurement could maybe help me out as it would give me an average of a small area as opposed to a single point in the room?
It is only me that cares about this so I guess a single point will do just fine.

 

[Exponential]

@jtalden

Hello again. After a number of hours of trial and error, I have settled on what I think is a suitable compromise.
I noticed that, whenever I moved the mains closer to the wall, it had the effect of moving the suck-out at 100hz further up the frequency range to a point where I would not longer be able to hide it with using the subs to fill in the gaps.
There was one particular measurement (not supplied) where the suck-out at 100hz combined with the dip at around 200hz and caused a huge null.
It was a very interesting but ultimately fruitless exercise as I am not in a better position at all.

I have come to terms with the fact that this is my room and it will not get any better other than applying room treatment which the wife will NOT allow.

The measurements attached are with no EQ applied to either the subs or the mains. Just the raw response.

I will post again once I have used the GLM software to calibrate my mains to see how it deals with it.

 

[Exponential]

@jtalden

Ok, I have run the Genelec GLM software on my mains and it has had a positive effect although, I do still have suck-outs here and there, I believe, psychoacoustically, I won't really hear them is the overall listening experience and I'm ok with that.

What is my next step? Do I EQ the subs with a nice house curve and choose a suitable crossover or am I ahead of myself here?

I think, due to my mains being so close to the walls, I will have to apply a 100hz LR 48db per octave to completely cancel out the room effects to the low end on my mains so they don't interfere with the subwoofers and effect their response when all speakers are playing.

I'm using a Minidsp SHD so I have a lot of control over crossovers and PEQ filters.

Just as a trial, I have applied a 12 db per octave LR filter to my subs and a 48 db per octave filter to my mains. The MDAT is attached.
Now, I am well over my limit of knowledge here so please do correct me on this but I feel in my mind that this could be a positive approach in order to achieve a good summation across the frequency range.

I have no idea though so please guide me in the right direction if possible!

Thank you.

 

[Exponential]

If I boost the subs by 6db and align phase at the 100hz crossover frequency, I get very favourable summation.

Do you agree?

 

[jtalden]

Do you feel a multipoint measurement could maybe help me out as it would give me an average of a small area as opposed to a single point in the room?

I prefer averaging several points around the LP for the purpose of PEQ.

 

[jtalden]

@jtalden
I have come to terms with the fact that this is my room and it will not get any better other than applying room treatment which the wife will NOT allow.

Many of us in that situation.

The measurements attached are with no EQ applied to either the subs or the mains. Just the raw response.

I will post again once I have used the GLM software to calibrate my mains to see how it deals with it.

This data is does not add additional information.

 

[jtalden]

@jtalden

Ok, I have run the Genelec GLM software on my mains and it has had a positive effect although, I do still have suck-outs here and there, I believe, psychoacoustically, I won't really hear them is the overall listening experience and I'm ok with that.

What is my next step? Do I EQ the subs with a nice house curve and choose a suitable crossover or am I ahead of myself here?

Some EQ of the SWs prior to delay timing is fine and can be helpful. It is usually also okay to wait until the end when the XO and delays are finalized.
[The SWs should have been measured full range, but this is good enough for this analysis.]

I think, due to my mains being so close to the walls, I will have to apply a 100hz LR 48db per octave to completely cancel out the room effects to the low end on my mains so they don't interfere with the subwoofers and effect their response when all speakers are playing.

The 100 Hz XO will help the SPL sag issue a little, but there will still be a difference between the L and R channel SPL responses in that area.

Just as a trial, I have applied a 12 db per octave LR filter to my subs and a 48 db per octave filter to my mains. The MDAT is attached.

I analyzed that this 100Hz data file for delay timing. I didn't see clear evidence that a different XO setting would be more favorable. There are numerous other XO options, but I am not willing to analyze more of them.

The only change needed is to this setup is to reverse the polarity of the 2 Subs. The delay timing you have currently set is already very favorable with that change.

Recommended next steps for you are to:

1. Choose a house curve
2. Adjust the levels of the R, L and SWs channels as needed so they are favorable positioned relative to the house curve
3. Apply additional PEQ as needed

I will follow with a post of couple of charts and my mdat file.

 

[jtalden]

Charts and data:

 

[sm52]

jtalden
hello. What does negative group delay mean?

 

[Exponential]

Thank you very much for taking the time to take a look at my current set up.
It must be a bit of a curse knowing so much about REW and how it works as you have people like me pestering you all the time!

Anyway, I will take a look at what you've suggested and apply it to my system.

I'll post back when I've done it for a bit of feedback.

Thanks again!

Mike.