REW interpretation help

[Deepinton]

I measured all my amps and speakers with the microphone 2" away from the drivers to ensure that my monoblocks, the speaker crossovers and drivers are all good (which they are). Now I am trying to understand my room issues. I had a very dead room (padded walls, acoustic tile on ceiling, berber carpet) so I added some curved slat diffusers and hardboard reflectors to the side and rear of the room. It all sounds good to me but I was curious to see how it all measures, especially RT60, and if there are some placement tweaks to improve things. My room is not symmetrical, with a 3x4' corner abutment along the right front wall and a 5'x5' entry way along the right rear. The front of my speakers are about 7' from the front wall and 14,7' from the rear wall. they are about 32" from the side wall and about 10' apart (midline of tweeter). I sit in an equilateral triangle. Speakers point straight ahead. Phantom center is dead on. Attached zip folder has just 2 measurements (L and R speaker measured individually), taken with the microphone at midpoint of my ears at the seated position and pointed straight at a point between the 2 speakers. I'm happy to take more measurements. I can see some room modes at 26 and 56 hz which I can try to address with the equalizer in my c55. I am still trying to learn about phase angles and such. I use a Umik 1 with the correct calibration file.
Apologies if I am doing something wrong, but I am a quick and eager learner. Thanks in advance.

 

[Keith_W]

Firstly, do not measure speakers 2" away. This is because of the baffle step. The baffle step is a volume loss of about 6dB that occurs because sound transitions from a hemispherical to spherical radiating pattern. Read more about it here.

1. For high frequency drivers, the mic needs to be placed about 2 baffle widths away. Or a standard 1m.
2. For low freq drivers, the mic needs to be spaced 0.11x the width of the baffle (Keele's method). Keele's 1974 paper assumed an infinite baffle, i.e. driver mounted to a wall. Calculate the baffle step, and if the bandpass of the driver is involved, the measurement needs to be taken 2 baffle widths away.

Naturally, further measurement distance = more likelihood your measurement will be contaminated by reflections. This is why people take their speakers outdoors.

Re: your questions about RT60. See the eBook in my link and go to the section on RT60.

 

[Deepinton]

Firstly, do not measure speakers 2" away. This is because of the baffle step. The baffle step is a volume loss of about 6dB that occurs because sound transitions from a hemispherical to spherical radiating pattern. Read more about it here

Thanks for the input. For clarification, the attached measurements were taken at listening position. I measured about 2-3" from mid-range baffle just to confirm that the repairs I performed to the mid range driver and the crossover on my right speaker were successful. I believe that so long as the mic was at the same location on the other speaker, my conclusion that the repairs were successful still holds as the overlays matched quite well.

 

[Keith_W]

Yes I am aware that those are MLP measurements.

You did not say anything about repairs to your speaker in your original post. It makes me wonder why you have this:

When the measurements of your B&W 803 D3 speakers on spinorama.org show the exact opposite, a treble lift:

Are you sure something did not go wrong when you repaired your speaker? You might like to confirm this. Place your mic 1m on-axis from the speaker, at tweeter height. Use a mic boom tripod stand.

The decay in your ETC is excellent, which suggests you are using some absorptive room treatment. But there are a lot of loud and early reflections.

And please feel free to ask any questions about the ETC and RT60 once you read the relevant section in the book and find out how to take that measurement and understand its limitations.

 

[Deepinton]

Well, I had to reboot Windows and now am struggling to get my SPL in rew's to read correctly. Always something or the other

 

[Deepinton]

I read all the REW instructions, am using the right calibration file, umik is configured properly in Windows.... And I tried all the AI help. But my SPL in REW is still reading 20 db higher than it should. Anyone know what I'm doing wrong?

 

[Keith_W]

How do you know it's reading 20dB higher than it should? Do you have an SPL meter?

Anyway, if you have a proper SPL reference and you suspect the UMIK-1 is wrong, follow the REW SPL calibration process. It doesn't really matter for the type of measurements you are doing anyway. It matters for formal testing.

 

[Deepinton]

How do you know it's reading 20dB higher than it should? Do you have an SPL meter?

Anyway, if you have a proper SPL reference and you suspect the UMIK-1 is wrong, follow the REW SPL calibration process. It doesn't really matter for the type of measurements you are doing anyway. It matters for formal testing.

Well, rew has consistently read my room at 19 db C S. for the past year. Which aligns with the coarse meter on my phone. So some handshake is breaking between REW and windows. I don't have a calibrated SPL reference

 

[Deepinton]

Yes I am aware that those are MLP measurements.

You did not say anything about repairs to your speaker in your original post. It makes me wonder why you have this:

View attachment 536942

When the measurements of your B&W 803 D3 speakers on spinorama.org show the exact opposite, a treble lift:

View attachment 536944

Are you sure something did not go wrong when you repaired your speaker? You might like to confirm this. Place your mic 1m on-axis from the speaker, at tweeter height. Use a mic boom tripod

I have the 800D3. Only the right speaker had blown driver and crossover. I did a quick test to compare the repaired speaker to the good one and the overlay was quite good, but not perfect.

 

[Keith_W]

I have the 800D3.

My mistake. This is the 800D3, taken from this review.

There is no 5dB step at 1.5kHz. I still think you need to investigate that.

 

[Deepinton]

My mistake. This is the 800D3, taken from this review.

View attachment 537063

There is no 5dB step at 1.5kHz. I still think you need to investigate that.

Since I used similar measurement technique for left and right and since the left speaker never had any problems, could it just be the measurement technique showing that 5db step

 

[Keith_W]

Since I used similar measurement technique for left and right and since the left speaker never had any problems, could it just be the measurement technique showing that 5db step

Absolutely. The possibilities are: (1) that published measurement is incorrect, which is unlikely - given that it was taken by professionals; (2) B&W manufactured your speakers incorrectly. Also unlikely, given that L and R are the same; or (3) there is a problem with your measurement technique.

I am guessing that you have not toed-in your speaker and you are measuring the off-axis response, but that is a pretty big drop in SPL from lack of toe-in alone. Just get a proper 1m on-axis measurement for peace of mind. Then consider toe-in to get rid of that 5dB step.

The other possibility is incorrect calibration file, but it is rare that mics are so nonlinear in the 1.5kHz range. The nonlinearities are usually higher up, from 15kHz or so.

 

[LSPhil]

Thanks for sharing the repair details. Out of curiosity, do you know what caused the midrange driver and crossover to fail? Was it a component fault, accidental damage, amplifier clipping, or something else?

Also, I'm a bit surprised by the measurements you've posted. They look so severe that it's hard for me to imagine these speakers delivering reasonably accurate sound in their current state.

Would it be possible to do another set of measurements with the bass drivers disconnected, measuring only the midrange and tweeter section?

It would be very helpful to see both:
a measurement at 0.5 m distance, and
a measurement from the main listening position.

That would make it much easier to understand what's going on and whether the remaining issues are related to the bass section, crossover integration, room effects, or something else.

 

[Sokel]

Absolutely. The possibilities are: (1) that published measurement is incorrect, which is unlikely - given that it was taken by professionals; (2) B&W manufactured your speakers incorrectly. Also unlikely, given that L and R are the same; or (3) there is a problem with your measurement technique.

I am guessing that you have not toed-in your speaker and you are measuring the off-axis response, but that is a pretty big drop in SPL from lack of toe-in alone. Just get a proper 1m on-axis measurement for peace of mind. Then consider toe-in to get rid of that 5dB step.

The other possibility is incorrect calibration file, but it is rare that mics are so nonlinear in the 1.5kHz range. The nonlinearities are usually higher up, from 15kHz or so.

On top of no on-axis I would also suspect measuring mic's height.
Mic put on the seat of MLP with a small tripod for example.

 

[Deepinton]

Absolutely. The possibilities are: (1) that published measurement is incorrect, which is unlikely - given that it was taken by professionals; (2) B&W manufactured your speakers incorrectly. Also unlikely, given that L and R are the same; or (3) there is a problem with your measurement technique.

I am guessing that you have not toed-in your speaker and you are measuring the off-axis response, but that is a pretty big drop in SPL from lack of toe-in alone. Just get a proper 1m on-axis measurement for peace of mind. Then consider toe-in to get rid of that 5dB step.

The other possibility is incorrect calibration file, but it is rare that mics are so nonlinear in the 1.5kHz range. The nonlinearities are usually higher up, from 15kHz or so.

Here is the 1m on axis to tweeter for L and R speakers, albeit with an SPL anomaly. Much appreciated.

 

[Deepinton]

On top of no on-axis I would also suspect measuring mic's height.
Mic put on the seat of MLP with a small tripod for example.

the mic is on a boom stand.

 

[Deepinton]

Thanks for sharing the repair details. Out of curiosity, do you know what caused the midrange driver and crossover to fail? Was it a component fault, accidental damage, amplifier clipping, or something else?

Also, I'm a bit surprised by the measurements you've posted. They look so severe that it's hard for me to imagine these speakers delivering reasonably accurate sound in their current state.

Would it be possible to do another set of measurements with the bass drivers disconnected, measuring only the midrange and tweeter section?

It would be very helpful to see both:
a measurement at 0.5 m distance, and
a measurement from the main listening position.

That would make it much easier to understand what's going on and whether the remaining issues are related to the bass section, crossover integration, room effects, or something else.

The right speaker probably had a hot plug event that fried the resistors and one of the capacitors and took out the voice coil of the mid-range. I put in a new crossover, new resistors and a new mid-range driver. The speakers sound better than the 802D's and the t+a criterion I have laying around, and they matched the unaffected speaker, so I didn't give the repairs much thought

 

[Sokel]

the mic is on a boom stand.

your measurements:

They are both close, not as close as I would like to see in a treated room (RT is on the extremely dry side below 200 ms) but highs are sure strange.

Can't imagine both speakers got the same damage so must be something backwards. Are you sure your electronics are in check and no DSP or something are in works?
I must admit I have never seen such a deep null at 200Hz-300Hz.Are you sure the x-overs are ok? Do you by any chance bi-amp or something in some strange way?


I would measure at 0.5 meters and play with height (somewhere between mid and tweeter) .


And with some gating to get some room out of the way:

 

[Keith_W]

Here is your latest measurement with the published measurement I posted earlier on top. Both graphs have the same vertical and horizontal scale of 60dB. Above 1kHz or so, your measurement should look very close to the published measurement of the speaker. As you can see, with your speakers, there is a volume loss between 3kHz - 15kHz compared to published measurements. I don't know why your speakers are doing this.

The 1.5kHz step observed in your previous measurement is now absent. I think we can safely say that the 5dB step in the measurement that you first posted is due to lack of toe-in. It would be nice to confirm this with published off-axis measurements of your speaker, but unfortunately there aren't any.

Anyway, from these measurements it looks as if you would benefit from some toe-in. Got to love the free improvements

Since we now have two sets of measurements of your speaker from three different locations, I took the opportunity to see how much the RT60 varied between locations. Answer: not much above 1.5kHz. As Sokel noted, it's a bit on the dry side.

 

[Deepinton]

View attachment 537102

Here is your latest measurement with the published measurement I posted earlier on top. Both graphs have the same vertical and horizontal scale of 60dB. Above 1kHz or so, your measurement should look very close to the published measurement of the speaker. As you can see, with your speakers, there is a volume loss between 3kHz - 15kHz compared to published measurements. I don't know why your speakers are doing this.

The 1.5kHz step observed in your previous measurement is now absent. I think we can safely say that the 5dB step in the measurement that you first posted is due to lack of toe-in. It would be nice to confirm this with published off-axis measurements of your speaker, but unfortunately there aren't any.

Anyway, from these measurements it looks as if you would benefit from some toe-in. Got to love the free improvements

View attachment 537108

Since we now have two sets of measurements of your speaker from three different locations, I took the opportunity to see how much the RT60 varied between locations. Answer: not much above 1.5kHz. As Sokel noted, it's a bit on the dry side.

Thanks for the analysis all.
"As you can see, with your speakers, there is a volume loss between 3kHz - 15kHz compared to published measurements. I don't know why your speakers are doing this."
I am using a c55 preamp straight to my mc501 monoblocks. The equalization on the preamp is disabled. Could the room cause this? I added a bunch of curves slat diffusers, and a bunch of hardboard over my padded walls to get the rt60 up. I believe this problem has nothing to do with the space behind the speakers, only forward of the speakers? What other tests could I do to narrow down the solution space? Thanks again