B&O Balance // Measured significant drop in highs

[5omeone]

Dear ASR community,

I would like to optimize the sound of my B&O Beosound Balance speaker pair by using REW and a UMIK-1 for the measurement. The microphone calibration file is loaded, and I followed the REW tutorials for setting up the measurement. The microphone is pointing directly at the two speakers.

I have noticed that with neutral sound settings, there is a significant drop in frequency from 6 kHz upwards. Is this normal (for B&O)? Or have I measured incorrectly? I assume that the speakers are working properly. High frequencies can be heard, although not very distinctly.

What is your suggestion or remark? Appreciate your feedback.

Best regards!


PS: Room correction (RC) is activated; without RC I measure the nearly exact same response.

 

[staticV3]

Very strange treble response...

In REW->Settings->Soundcard, make sure that you set the driver type to Java and then select the EXCL versions of your mic input and soundcard output. That bypasses any APOs that may mess with the sound.

For in-room frequency response measurements, I highly recommend you use the Moving Microphone Method:

The moving microphone method (MMM) for Dummies using REW

After getting a couple of weird results on my first MMM measurements, I went searching around the forums & YouTube for what went wrong. To save others from gathering various bits and pieces of info from a dozen sources, I decided to publish this quick write-up here. You can use it as a rough...

www.audiosciencereview.com

There are tutorials on YouTube as well.

How are you sending the audio to your speakers? Something in that chain may be affecting the measurement.

Also, when displaying in-room frequency response measurements, it is convention that the highest displayed dB number is about 50dB above the lowest. This prevents weirdly stretched or compressed graphs.

You can set that using the Limits button.

If some part of the frequency response does not fit within this 50dB window, then something's definitely wrong.

Also, have a look at the top left corner of REW's response graph window. There's a handy camera icon which exports clean graphs automatically.

 

[Sokel]

Very strange treble response...

In REW->Settings->Soundcard, make sure that you set the driver type to Java and then select the EXCL versions of your mic input and soundcard output. That bypasses any APOs that may mess with the sound.

For in-room frequency response measurements, I highly recommend you use the Moving Microphone Method:

The moving microphone method (MMM) for Dummies using REW

After getting a couple of weird results on my first MMM measurements, I went searching around the forums & YouTube for what went wrong. To save others from gathering various bits and pieces of info from a dozen sources, I decided to publish this quick write-up here. You can use it as a rough...

www.audiosciencereview.com

There are tutorials on YouTube as well.

How are you sending the audio to your speakers? Something in that chain may be affecting the measurement.

Also, when displaying in-room frequency response measurements, it is convention that the highest displayed dB number is about 50dB above the lowest. This prevents weirdly stretched or compressed graphs.

You can set that using the Limits button.

If some part of the frequency response does not fit within this 50dB window, then something's definitely wrong.

Also, have a look at the top left corner of REW's response graph window. There's a handy camera icon which exports clean graphs automatically.

I'm afraid that what we see is dBr units there, chart must be changed to SPL.

 

[staticV3]

I'm afraid that what we see is dBr units there, chart must be changed to SPL.

dBr is fine.

 

[Sokel]

dBr is fine.

I know it's fine, no visual difference, but just to align with the rest 50dB SLP range charts around here.

 

[respice finem]

I've once had a similar "symptom" with my B&W CDM7, "dried out" ferrofluid in the tweeters.
But, those B&W were 20 years old, so IDK what it may be with the as-good-as-new B&O...

 

[staticV3]

I know it's fine, no visual difference, but just to align with the rest 50dB SLP range charts around here.

-5 to -55 is still a 50dB range

 

[Juhazi]

Both speakers playing at same time- interferences? Measure only one speaker at time, move the mic a little and take at least three measurements. Then the other speaker 3-4 sweeps. Overlay all measurements and you can let REW make an average. Try different gating and smoothing settings. Is REW preferences/Soundcard setting using UMIK as input?

Are settings right for a stereo pair? https://support.bang-olufsen.com/hc/en-us/sections/360008022911-Beosound-Balance
It is possible that tweeters are dead or not connected, but I suppose quality contol is good.... Play noise and listen in nearfield.
Take measurement of some other hifi loudspeaker



Finally - it is very likely that the speaker really does not produce anything above 7kHz. B&O knows psychoacoustis well!

 

[staticV3]

To measure the combined L+R response without encountering interference issues, use the Beta version of REW and in the Generator, enable Uncorrelated.

Or you can measure them separately, of course.

 

[5omeone]

Wow, many thanks for the great feedback! Appreciate your support!

In REW->Settings->Soundcard, make sure that you set the driver type to Java and then select the EXCL versions of your mic input and soundcard output. That bypasses any APOs that may mess with the sound.

Hmm, I looked for it but couldn't find it. Is there another menu?

How are you sending the audio to your speakers? Something in that chain may be affecting the measurement.

I use AirPlay—that's actually the only way to send signals to the speakers at the moment.

Also, when displaying in-room frequency response measurements, it is convention that the highest displayed dB number is about 50dB above the lowest. This prevents weirdly stretched or compressed graphs.

You can set that using the Limits button.

If some part of the frequency response does not fit within this 50dB window, then something's definitely wrong.

Also, have a look at the top left corner of REW's response graph window. There's a handy camera icon which exports clean graphs automatically.

Thanks for the tip! I played around with it a bit and can now provide you with a better export, including a legend.

I'm afraid that what we see is dBr units there, chart must be changed to SPL.

You're right, I changed it to SPL and created an improved export. I used dBr in past times.

I've once had a similar "symptom" with my B&W CDM7, "dried out" ferrofluid in the tweeters.
But, those B&W were 20 years old, so IDK what it may be with the as-good-as-new B&O...

Good point, but since I can't detect any imbalance between the speakers, I assume that both are working properly.

Both speakers playing at same time- interferences? Measure only one speaker at time, move the mic a little and take at least three measurements. Then the other speaker 3-4 sweeps. Overlay all measurements and you can let REW make an average. Try different gating and smoothing settings. Is REW preferences/Soundcard setting using UMIK as input?

Are settings right for a stereo pair? https://support.bang-olufsen.com/hc/en-us/sections/360008022911-Beosound-Balance
It is possible that tweeters are dead or not connected, but I suppose quality contol is good.... Play noise and listen in nearfield.
Take measurement of some other hifi loudspeaker

View attachment 495887

I will remeasure asap. But I need to get the family out of the house, as the measurements are challenging with my toddler. I'll keep you posted.

For in-room frequency response measurements, I highly recommend you use the Moving Microphone Method:

The moving microphone method (MMM) for Dummies using REW

After getting a couple of weird results on my first MMM measurements, I went searching around the forums & YouTube for what went wrong. To save others from gathering various bits and pieces of info from a dozen sources, I decided to publish this quick write-up here. You can use it as a rough...

www.audiosciencereview.com

Wow! I think I need to take a little more time for this. I just wanted a simple answer to perform a simple optimization without perfection. Well, now here we are.

Finally - it is very likely that the speaker really does not produce anything above 7kHz. B&O knows psychoacoustis well!

Interesting. I wouldn't be surprised.

To measure the combined L+R response without encountering interference issues, use the Beta version of REW and in the Generator, enable Uncorrelated.

Or you can measure them separately, of course.

I'll try out as soon as I can and cannot achieve a better performance with the other approaches. Many thanks!

Please find below my setup and the export.

 

[staticV3]

Hmm, I looked for it but couldn't find it. Is there another menu?

The EXCL thing only applies to 64-bit versions of Windows running a somewhat modern version of REW.

On macOS, 32-bit Windows, or old REW versions, EXCL does not exist.

Please find below my setup and the export.

The graph is still quite compressed at 120+ dB between bottom and top.

Furthermore, the Measure dialog shows that L+R is being measured, so interference may be a factor.

Try measuring just Left or just Right and if the treble response changes significantly, then you can roll with the individual response graphs, or, if you're curious what the combined bass response looks like since bass is Mono in 90% of recordings and doesn't simply sum between L and R, then you can measure the combined L+R bass response plus clean treble response w/o interference all in one go, using the Moving Microphone Method combined with Generator->Uncorrelated->On.

 

[kidprofessor]

I use AirPlay—that's actually the only way to send signals to the speakers at the moment.

I had the same problem using AirPlay previously, I ended up using a direct digital connection to fix it, then the high end showed normally. It's highly doubtful it is an actual physical effect, the drop off wouldn't be that smooth even with variable smoothing applied.
I used AirPlay on a different device than the computer that is measuring recently and there has not been the high end drop off problem

 

[staticV3]

The Beosound Balance does have a Mini Toslink input, so you could test a lossless digital connection to the speaker against lossy Airplay for cheap, using a Cubilux SPDIF Adapter, Hifime UT23, or similar.

It's also possible that Airplay itself is fine, but its's specifically the Beosound's Aiplay receiver that's causing issues.

In that case, a no-brainer would be to buy a WiiM Mini and plug it into the Beosound via Toslink.

Not only would you get a flawless AirPlay implementation through the WiiM, but you'd also add serious room correction capability to the system.

 

[5omeone]

Hi everyone,

just to let you know: I’ve run several tests in the meantime and was able to quickly identify AirPlay as the cause of the drop in highs. The problem was resolved immediately with a direct USB connection.

I’ve now connected a Wiim Pro Plus in front of the Beosound Balance to manage a few room modes. I’m using an optical connection between the streamer and the Balance. That works (mostly) without any issues as well.

P.S. and Background: I’d like to measure more frequently in REW, especially with AirPlay, since that’s the realistic listening scenario. Unfortunately, all the cabling involved is a bit of a challenge for me. I can’t get AirPlay to work reliably as a sound card in REW, though... It’s a shame, really.

 

[staticV3]

P.S. and Background: I’d like to measure more frequently in REW, especially with AirPlay, since that’s the realistic listening scenario. Unfortunately, all the cabling involved is a bit of a challenge for me. I can’t get AirPlay to work reliably as a sound card in REW, though... It’s a shame, really.

There's a simple solution as direct playback from within REW is not necessary:

In REW->Generator-> export your test tones (Pink noise, Measurement Sweep, etc) to .wav using the "Save to file" function and load them into WiiM Home, or any other playback software that supports AirPlay.

For the Measurement Sweep, make sure that you enable Add Timing Ref in the Generator. The two other options "Play dither" and "Fill silence" are also recommended to turn on.

In case you use Pink noise for the Moving Microphone Method (this is the preferred method for cutting room modes using Parametric EQ), open the RTA, configure it for MMM, then play the Pink Noise.wav on repeat and click the Start button in REW (red circle).

In case you use the Measurement Sweep, open the Measure dialog and set Method->Sweep, and Playback->From file. Load the Measurement sweep.wav file into the dialog, then click Start.
REW will measure the Ambient noise floor, then wait for you to start external playback of your .wav file. It will then detect the sweep and do the measurement like normal.

 

[5omeone]

There's a simple solution as direct playback from within REW is not necessary:

In REW->Generator-> export your test tones (Pink noise, Measurement Sweep, etc) to .wav using the "Save to file" function and load them into WiiM Home, or any other playback software that supports AirPlay.

For the Measurement Sweep, make sure that you enable Add Timing Ref in the Generator. The two other options "Play dither" and "Fill silence" are also recommended to turn on.

In case you use Pink noise for the Moving Microphone Method (this is the preferred method for cutting room modes using Parametric EQ), open the RTA, configure it for MMM, then play the Pink Noise.wav on repeat and click the Start button in REW (red circle).

In case you use the Measurement Sweep, open the Measure dialog and set Method->Sweep, and Playback->From file. Load the Measurement sweep.wav file into the dialog, then click Start.
REW will measure the Ambient noise floor, then wait for you to start external playback of your .wav file. It will then detect the sweep and do the measurement like normal.

Thank you so much for the really helpful information! I'll give MMM a try as soon as possible and I'm curious to see how much further I can fine-tune the system. I’ve already noticed that the traditional "fixed mic" measurement method is significantly more time-consuming if you want to consider different listening positions.