Question about REW L+R vs L vs R speaker measurements

[nadrealista]

So recently I got hold of Polk LSim 705 speakers and finally got around to getting REW and minidsp calibration mic. My set up is Pioneer Elite SC-95 AVR as pre amp to Outalaw 5000x amp running front stage in 5.2.4 set up.

Here are the 2 ch plots I got and my question is why does the combined L+R measurement has a FR dip in the 7k to 14k region when individually measured speakers do not?

Is there some kind of cancelation happening there?

Any other plots or measurements I should provide to figure this one out?

RT60

waterfall:

but them I erased all AVR calibration and got this:

 

[staticV3]

Is there some kind of cancelation happening there?

Likely.

For in-room frequency response measurements, I'd highly recommend you use the Moving Microphone Method with no additional smoothing, and if you want to measure combined L+R response, use the beta version of REW and in the Generator->random PN-> turn Uncorrelated On.

 

[Keith_W]

Is there some kind of cancelation happening there?

Any other plots or measurements I should provide to figure this one out?

Yes, it's likely cancellation. You will need to look at the phase graph under "Overlays". The fact that the dip in the response between 7-14kHz only appears WITH DSP suggests that you are doing something inappropriate in that region. Likely PEQ in one channel but not the other.

You could read the REW eBook. Unfortunately I haven't finished writing the part on "Phase" yet. But you can view the work in progress here.

Would you be able to upload your MDAT? You need to zip it first.

 

[nadrealista]

Sure here are two measurements with phase info, calibrated one has what looks like phase inversion right in the middle of the that dip. Pioneer calibration has phase matching feature but Polk LSim has inverted phase by design (Polk LSiM series speakers are intentionally wired with an inverted phase on one of the drivers, typically the tweeter, due to a second-order (12 dB/octave) crossover. This is a deliberate design choice to improve driver integration and performance, and it is not a wiring error) which is probably throwing calibration phase alignment off?

Should I be seeing these 3 phase shifts in the bass region , subs are crossed at 80Hz?

mdat file attached as well.

I will try to use use the Moving Microphone Method once I figure out how

calibrated:

no calibration

here are the speaker specs and crossover points/orders which apparently creatate phase shifts as well
6dB/Oct = 90 degrees
12dB/Oct = 180 degrees
18dB/Oct = 270 degrees
24dB/Oct = 0 degrees

 

[Keith_W]

I am sorry but I do not know what your labels mean.

I would have thought that "L+R M6" is "L M6" and "R M6" measured together. But when I sum "L M6" and "R M6", the result looks the same as "L+R NO Calibration".

Can you please label your graphs properly. To analyse why you have a dip between 7kHz - 14kHz, we need a set of graphs like this:

- L+R with DSP
- L with DSP
- R with DSP

- L+R NO DSP
- L NO DSP
- R NO DSP

 

[DVDdoug]

Yeah, normal cancelation... 10kHz has a wavelength of 1.35 inches. A distance difference of 3/4 of an inch makes the waves 180 degrees out of phase. Plus there are also lots of reflected waves combining with random timing/phase.

So, mic or ear position make a big difference. And of course your left & right ears are not in the same spot.

Try playing a constant 5 or 10kHz tone and you'll be surprised how much variation you get with slight head movements. But don't play constant-loud test-tones because you could blow your tweeters. Tweeters can't handle as much power as woofers and our ears are not as sensitive at high frequencies so it may not sound that loud.

here are the speaker specs and crossover points/orders which apparently creatate phase shifts as well

In a properly designed speaker that will (mostly) be handled. For example, the low-pass for the woofer might have a -90 degree phase-lag and the crossover frequency and the high-pass for the tweeter (or midrange) a +90 degree shift. That puts them 180 degrees out-of-phase, causing cancelation and a dip at the crossover frequency where both drivers are operating. But they can be put back in-phase by swapping the +/- connections to the tweeter (or midrange). In a 3-way, the tweeter connections would be swapped again (making the tweeter the same as the woofer) putting the midrange and tweeter back in-phase at their crossover frequency.

 

[nadrealista]

View attachment 489614

I am sorry but I do not know what your labels mean.

View attachment 489617

I would have thought that "L+R M6" is "L M6" and "R M6" measured together. But when I sum "L M6" and "R M6", the result looks the same as "L+R NO Calibration".

Can you please label your graphs properly. To analyse why you have a dip between 7kHz - 14kHz, we need a set of graphs like this:

- L+R with DSP
- L with DSP
- R with DSP

- L+R NO DSP
- L NO DSP
- R NO DSP

Sorry for the confusion. M6 is MCACC calibration memory slot so all M6 measurements are with AVR calibration.

"I would have thought that "L+R M6" is "L M6" and "R M6" measured together. But when I sum "L M6" and "R M6", the result looks the same as "L+R NO Calibration"."
L+R M6 is L and R measured together and "L M6" and "R M6" are individual speaker measurements so either AVR DSP is messing stuff up or there is some kind of interference between the two when they are playing together irrespective of DSP but this goes away when run them together without AVR calibration which is L+R NO Calibration measurement.

relabeled measurements attached

 

[OCA]

So recently I got hold of Polk LSim 705 speakers and finally got around to getting REW and minidsp calibration mic. My set up is Pioneer Elite SC-95 AVR as pre amp to Outalaw 5000x amp running front stage in 5.2.4 set up.

Here are the 2 ch plots I got and my question is why does the combined L+R measurement has a FR dip in the 7k to 14k region when individually measured speakers do not?

Is there some kind of cancelation happening there?

Any other plots or measurements I should provide to figure this one out?

View attachment 489517

RT60
View attachment 489518

waterfall:
View attachment 489519

but them I erased all AVR calibration and got this:
View attachment 489520

Your DSP has applied a stupid allpass filter to one of the speakers at around 10k. Deaf graph chasers who boldly go beyond the room transient deserve these kind of filters

 

[nadrealista]

Yeah, normal cancelation... 10kHz has a wavelength of 1.35 inches. A distance difference of 3/4 of an inch makes the waves 180 degrees out of phase. Plus there are also lots of reflected waves combining with random timing/phase.

So, mic or ear position make a big difference. And of course your left & right ears are not in the same spot.

Try playing a constant 5 or 10kHz tone and you'll be surprised how much variation you get with slight head movements. But don't play constant-loud test-tones because you could blow your tweeters. Tweeters can't handle as much power as woofers and our ears are not as sensitive at high frequencies so it may not sound that loud.


In a properly designed speaker that will (mostly) be handled. For example, the low-pass for the woofer might have a -90 degree phase-lag and the crossover frequency and the high-pass for the tweeter (or midrange) a +90 degree shift. That puts them 180 degrees out-of-phase, causing cancelation and a dip at the crossover frequency where both drivers are operating. But they can be put back in-phase by swapping the +/- connections to the tweeter (or midrange). In a 3-way, the tweeter connections would be swapped again (making the tweeter the same as the woofer) putting the midrange and tweeter back in-phase at their crossover frequency.

interesting, but if distance to the mic is the problem then NO DSP measurement would show same cancelation and it does not?

I did check L/R distance to mic is identical, however when I looked in DSP calibration settings there is 1/2" distance difference between L and R speaker but dsp should take that into account and adjust signal delay accordingly.

 

[nadrealista]

Your DSP has applied a stupid allpass filter to one of the speakers at around 10k. Deaf graph chasers who boldly go beyond the room transient deserve these kind of filters

that does not seem to be the case, here is L vs R with DSP

 

[marin.weigel]

1) The L+R DSP vs. No DSP measurements show a 64 us time delay between both channels.
2) You're looking at wrapped phase response plots. There are no sudden phase shifts. If you unwrap phase display, you'll see the real picture.
See attached pics.

 

[Oddball]

Your DSP has applied a stupid allpass filter to one of the speakers at around 10k. Deaf graph chasers who boldly go beyond the room transient deserve these kind of filters

I would actually flip this. I think we would finally deserve a solution that works above the room transient frequency. For god's sake we have been to the moon and back so many times and still struggling with this?

There would be few better positioned than you do let us know - do we need crowd-funding on something like next gen of mics to get there? What would be needed to actually get some solid results from full range room EQ?

 

[marin.weigel]

As the best practice I'd recommend to always measure individual channels separately with the timing reference selected for your measurement signal.
If desired, you can properly vector sum individual measurements later in REW.
Understand, that it's almost impossible to properly set up the mic to be equal distance to all respective speaker chassis for the first wavefront, so you will get artifacts in REW that don't show what you hear (single mic capsule vs. two ears at a distance with auditory cortex involved).

 

[nadrealista]

As the best practice I'd recommend to always measure individual channels separately with the timing reference selected for your measurement signal.
If desired, you can properly vector sum individual measurements later in REW.
Understand, that it's almost impossible to properly set up the mic to be equal distance to all respective speaker chassis for the first wavefront, so you will get artifacts in REW that don't show what you hear (single mic capsule vs. two ears at a distance with auditory cortex involved).

thank you for pulling the plots out, I just started using REW yesterday

so based on your plots DSP is introducing delay between the channels which creates phase shift/cancelation in the 7k to 14k region, did I get that right?

 

[marin.weigel]

thank you for pulling the plots out, I just started using REW yesterday

so based on your plots DSP is introducing delay between the channels which creates phase shift/cancelation in the 7k to 14k region, did I get that right?

Yes, sir.
EDIT:
Question being, did you measure both DSP / No DSP without touching the mic?
DSP was done how exactly?

 

[nadrealista]

Yes, sir.
EDIT:
Question being, did you measure both DSP / No DSP without touching the mic?
DSP was done how exactly?

yes all measurements were taken without moving the mic.

DSP is Pioneer MCACC Pro calibration which does try to do individual driver/speaker phase control among other things:

MCACC Pro uses a supplied, custom microphone to measure the room's acoustic characteristics and the performance of the connected speakers. The system then automatically makes a variety of adjustments:

• Full Band Phase Control: This key feature analyzes and corrects for phase differences across all frequencies and channels, including those introduced by speaker crossover networks or in the original audio material. The goal is to achieve sound coherence similar to full-range speakers while maintaining the benefits of multi-way speaker designs.
• Standing Wave Control: This feature helps to identify and reduce unwanted resonance or "standing waves" that can build up between walls, ceiling, and floor, especially in lower frequencies, which can negatively impact bass quality.
• Precision Distance: MCACC Pro enables highly precise, millimeter-level adjustments for speaker distances, which helps to ensure optimal timing and focus of the sound field at the primary listening position.
• Independent Dual Subwoofer EQ: The system allows for separate correction and optimization for two subwoofers, addressing potential cancellation issues and enabling optimal performance regardless of their placement in the room.
• Time Axis Compensation: The measurement process takes the time axis into account to apply the most effective compensation methods, enabling optimal tuning for the specific listening environment by differentiating direct sound from reverberant sound.
• Dolby Atmos Support: MCACC Pro includes a specialized Bass Management System for Dolby Atmos-enabled speakers, which redirects directional low-frequency components to the appropriate floor speaker rather than the subwoofer, ensuring a smoother, more seamless sound.

 

[OCA]

that does not seem to be the case, here is L vs R with DSPView attachment 489647

Allpass filters by defn don't change magnitude response.

 

[OCA]

I would actually flip this. I think we would finally deserve a solution that works above the room transient frequency. For god's sake we have been to the moon and back so many times and still struggling with this?

There would be few better positioned than you do let us know - do we need crowd-funding on something like next gen of mics to get there? What would be needed to actually get some solid results from full range room EQ?

A mic array, preferably spherical and lots of maths needed. Actually Trinnov is doing it the right way.

 

[Oddball]

A mic array, preferably spherical and lots of maths needed. Actually Trinnov is doing it the right way.

Right, what better can we do than Trinnov? $1K for mic, $20K for processor and $15K (or more) for the 8 wave forming subs, even if you can place them right, is a bit too much.

Trinnov is a bit lost in the low end, but overall they do know what works up there...

 

[nadrealista]

Yeah, normal cancelation... 10kHz has a wavelength of 1.35 inches. A distance difference of 3/4 of an inch makes the waves 180 degrees out of phase. Plus there are also lots of reflected waves combining with random timing/phase.

So, mic or ear position make a big difference. And of course your left & right ears are not in the same spot.

Try playing a constant 5 or 10kHz tone and you'll be surprised how much variation you get with slight head movements. But don't play constant-loud test-tones because you could blow your tweeters. Tweeters can't handle as much power as woofers and our ears are not as sensitive at high frequencies so it may not sound that loud.


In a properly designed speaker that will (mostly) be handled. For example, the low-pass for the woofer might have a -90 degree phase-lag and the crossover frequency and the high-pass for the tweeter (or midrange) a +90 degree shift. That puts them 180 degrees out-of-phase, causing cancelation and a dip at the crossover frequency where both drivers are operating. But they can be put back in-phase by swapping the +/- connections to the tweeter (or midrange). In a 3-way, the tweeter connections would be swapped again (making the tweeter the same as the woofer) putting the midrange and tweeter back in-phase at their crossover frequency.

And the top detective award goes to @DVDdoug !

I did sweep with 1/2" difference in L/R speaker distance without DSP and sure enough same dip appeared, removed 1/2" delta in speaker distance and dip disappeared from both DSP and non DSP sweeps...so mystery solved

Thanks everyone for their input!

Now what else can/should I improve based on these measurements of my response with DSP: