Headphone Measurements Using Brüel & Kjær 5128 HATS

[JohnYang1997]

Compare the excess phase plot posted above to the CSD phenomenon there
View attachment 82091
View attachment 82092
The region in question isn't measuring as being minimum phase, so I wouldn't expect a minimum phase EQ to perfectly correct its time domain measurement. As I've said, I have an explanation for that which is not "headphones are less minimum phase at low frequencies as an inverse function of output level" - that is, noise - and which seems to cohere to me. You see where I'm coming from?

Yep. Was going to ask for more tests but I think I am happy with this. Would be great to see csd for ER4 EQed to flat tho. (I have done many of these but long time ago)

 

[Mad_Economist]

Yep. Was going to ask for more tests but I think I am happy with this. Would be great to see csd for ER4 EQed to flat tho. (I have done many of these but long time ago)

Buy me an ER4, and I'd be happy to

 

[JohnYang1997]

Buy me an ER4, and I'd be happy to

I thought that would be something everyone has.

 

[DualTriode]

Mad_Economist,



I do believe that I see where you are coming from.



I do believe that what we are seeing is environmental noise, noise that lives on even when you unplug the headphones.



Do you have fans or other mechanical equipment running in your space? I know that in what I call Sparky’s Lab (Sparky is my cat) in northern California the fart fan running in the other end of the house puts a lot of low frequency in my FFT’s.



In my career I was called to a brand new research lab building that was a ductile moment steel frame structure built to be earthquake resistant. Steel is resilient and spring like, it shakes and moves if bumped and shakes for a long time. We put accelerometers at locations in the building. No amount of vibration isolation attenuated the building shaking when a car or truck would drive by on the street outside. The shaking was microscopic. The field of view in the scanning electron microscopes shook so much that they were useless. The Architect ended up paying errors and omissions. The building ended up being expensive medical school faculty offices and storage that looked like a lab. Noise and vibration is expensive.



You might try placing your HAT holder on an isolation base made of layers of neoprene kneeling pad, cork board and a heavy layer of particle board on top.



Thanks DT

 

[Mad_Economist]

Mad_Economist,



I do believe that I see where you are coming from.



I do believe that what we are seeing is environmental noise, noise that lives on even when you unplug the headphones.



Do you have fans or other mechanical equipment running in your space? I know that in what I call Sparky’s Lab (Sparky is my cat) in northern California the fart fan running in the other end of the house puts a lot of low frequency in my FFT’s.



In my career I was called to a brand new research lab building that was a ductile moment steel frame structure built to be earthquake resistant. Steel is resilient and spring like, it shakes and moves if bumped and shakes for a long time. We put accelerometers at locations in the building. No amount of vibration isolation attenuated the building shaking when a car or truck would drive by on the street outside. The shaking was microscopic. The field of view in the scanning electron microscopes shook so much that they were useless. The Architect ended up paying errors and omissions. The building ended up being expensive medical school faculty offices and storage that looked like a lab. Noise and vibration is expensive.



You might try placing your HAT holder on an isolation base made of layers of neoprene kneeling pad, cork board and a heavy layer of particle board on top.



Thanks DT

These tests were at night, with all of the non-vital equipment (my laptop, interface, and amplifier) off, but of course there is a noise floor to any room CSDs, due to their relative nature in ARTA's visualization as well as its short analysis windows, make this seem particularly significant - you wouldn't see much on such a measurement with a larger decay envelop and with absolute level plotted (if you like, I'll figure out CSDs in REW again and post one tonight).

I've experimented with vibration isolation before, but my general finding was that much of the noise is present in the air in my room - and, of course, there's some mains noise leakage, since the brilliant designers of HATSes decided to make the darn things internally single-ended

 

[BYRTT]

.....if you like, I'll figure out CSDs in REW again and post one tonight.....

If you get REW running down the road is it okay ask if you would share a REW mdat-file that includes your HATS calibration file and a HD650 sweep, would love have that fine reference data to comparison my own chain of HD650/miniDSP EARS ..

@Mad_Economist, @DualTriode, @JohnYang1997

About precision to measure excess phase i get from some paper that the more bandwidth the better and therefor down the road i think of mod my miniDSP EARS to analog if its possible so that the ADI Pro unit i saved up for last year can run head phone measurements up a 384 or 192kHz rate.

Below is how REW look at a synthetic impulse for a 20Hz-20kHz bandpass with 2nd order BW stopbands created in Rephase for 44,1/48/88,2/96/176,4/192/352,8/384kHz rates using relative huge FFT lenght that doubles with rates and a rectangular window, dotted black is the minimum phase textbook reference for that passband ..

Animation below includes amplitude, else its same as above inluding the colour scheme..

Left one below is the above one that use a rectangular window plus relative huge FFT lenght that doubles with rates, right one use HANN window and locked FFT lenght of 262144 samples, at low end and probably especial below REW's lowest reach of 2Hz curves amplitude gets out of tune to textbook minimum phase ones..

Another technical thing that mayby can represent some of the relative small number of excess phase in measurements, blue trace is not directly important to graph and represent phase of my HD650 impedance measured with a DATS unit and should be familiar to most, the interesting phase distortion or excess phase is that tranducer itself is a current device and phase for current in transducer device is the green graph (reverse of impedance), phase for voltage in transducer device is the red graph and pure flat and not the same as for current or vice versa..

 

[Mad_Economist]

If you get REW running down the road is it okay ask if you would share a REW mdat-file that includes your HATS calibration file and a HD650 sweep, would love have that fine reference data to comparison my own chain of HD650/miniDSP EARS ..

@Mad_Economist, @DualTriode, @JohnYang1997

About precision to measure excess phase i get from some paper that the more bandwidth the better and therefor down the road i think of mod my miniDSP EARS to analog if its possible so that the ADI Pro unit i saved up for last year can run head phone measurements up a 384 or 192kHz rate.

Below is how REW look at a synthetic impulse for a 20Hz-20kHz bandpass with 2nd order BW stopbands created in Rephase for 44,1/48/88,2/96/176,4/192/352,8/384kHz rates using relative huge FFT lenght that doubles with rates and a rectangular window, dotted black is the minimum phase textbook reference for that passband ..
View attachment 82323

Animation below includes amplitude, else its same as above inluding the colour scheme..
View attachment 82327

Left one below is the above one that use a rectangular window plus relative huge FFT lenght that doubles with rates, right one use HANN window and locked FFT lenght of 262144 samples, at low end and probably especial below REW's lowest reach of 2Hz curves amplitude gets out of tune to textbook minimum phase ones..
View attachment 82328

Another technical thing that mayby can represent some of the relative small number of excess phase in measurements, blue trace is not directly important to graph and represent phase of my HD650 impedance measured with a DATS unit and should be familiar to most, the interesting phase distortion or excess phase is that tranducer itself is a current device and phase for current in transducer device is the green graph (reverse of impedance), phase for voltage in transducer device is the red graph and pure flat and not the same as for current or vice versa..
View attachment 82330

Well, I'm not sure about HD650 mdat - because I have an HD600 - but I can share the average difference between my 4128 and EARS over a set of headphones I tested, if you like.

If you wanted to set out on a DIY measurement project, I would probably recommend interaural microphones before modding the EARS - its ADCs are a little over sensitive, but they're definitely not its primary problem.

Thanks for the excellent post, by the way!

 

[Kyle / MrHeeHo]

I would be interested in finding out what the different filters that come with the Ether CX actually change, just to see how different materials impact the sound of headphones

 

[DualTriode]

I don't have the details in front of me but it has the anthropomorphic pinnae and high-resolution coupler. I will be starting a thread on it soon and post measurements and contrast with BK 5128.

Hello,

I have been researching these headphone test fixtures.

Audio Precision and GRAS answer the same telephone number in Oregon. If you purchase a headphone fixture from AP these days it will be GRAS.

AP also sells the APx1701 Transducer Interface that includes Balanced microphone +48V Phantom power and microphone inputs. Also on board is CCP power for unbalanced microphones. There is an APx1701 on my shelf.

The GRAS 45CA-10 has a pre-polarized condenser capsule. CCP is required as a power supply. If you are using the APx1701 no additional external power supply is required.

The GRAS 45CA-9 requires an external power supply to polarize the condenser microphone capsules.

The 45CA-9 and 45CA-10 are much the same in function. The industry seems to think that the 45CA-9 is perhaps a little more stable in measurement and over time. Both Harmon and Head-Fi appear to be using the 45CA-9 with external power supplies. This is the best that I can decipher. A side note Harmon has used the 45CA for the bulk of their headphone testing and preference curve development. Head-FI has the 45CA in their lab as well.

https://www.grasacoustics.com/products/test-fixtures/product/ss_export/pdf2?product_id=734

Thanks DT

The 45CA-9 is my pick.

 

[outerspace]

I've experimented with vibration isolation before, but my general finding was that much of the noise is present in the air in my room

Several layers of towels can effectively reduce the noise. You can try to cover the rig with it before measurement. But it may not help with lower frequencies.

(sometimes I can hear the rumble from fan of my notebook in earphones through the thin silicone cable laid down on desktop surface... so vibration isolation can be very difficult thing)

 

[DualTriode]

Just FYI the primary frequency from a fan is: RPM times the number of blades divided by 60 (the number of seconds in a minute).

 

[Degru]

The response hump is a phenomenon of planar (whether electrostatic or magnetic) drivers with open rear volumes and sealed front volumes - with an ideal coupling to the ear, the response is essentially linear, but when that coupling is lost, you see a rise around the resonance of the driver, followed by a drop below.

The distortion measurement of the SR007 is somewhat surprising - typically the Omegas have extremely low distortion.

The distortion is likely also a result of poor seal. I've measured 300% bass distortion out of a Lambda before without seal. I was getting 20% just letting it sit loosely and jumped up to an amusing 300% when I attempted carefully pressing it down to achieve seal (didn't work, obviously). So many Stax measurements out there are flawed in this way, because they are very seal dependent and things like Lambda pads for example are contoured for a human head, not a measurement rig.

 

[Blujackaal]

0.3% where it matters isn't bad. It's about the same as Marv measured (SBAF). His measurements show its mostly 3rd harmonic though.

Speakerphone also measured it in that ballpark and also mainly 3rd harmonic.

So it pretty much kills the claim that it being single driver limited is bunk, It bass is 0.25% on most units. Seen some EQ there ER4SR's to mimic the ER4XR because of the lower THD, Seen 1.7% ER4XR's chart sheets.

Weird how the certain few have moved to using the ER4XR instead despite it using a different driver or pretend that unit varation is not a thing as a gotcha argument. The ER4S could be 0.6 ~ 1% on some & <0.4% on others. But dosen't change that with how popular the ER4XR/ER3XR are just shows musical enjoyment & it clarity it not impacted since I never had any issue with my ER4XR being <0.85% even with EQ on top.

 

[JohnYang1997]

I know exactly what drivers are in er4sr and er4xr. If you want, you can search on the internet where I shared a couple of times.

 

[Blujackaal]

I know exactly what drivers are in er4sr and er4xr. If you want, you can search on the internet where I shared a couple of times.

Is it the RAB series?, Since the one in the ER4SR 1mm smaller than the old ED29689 which was 6mm.

ER4S - ED29689

ER4SR - RAB 32063

 

[JohnYang1997]

Is it the RAB series?, Since the one in the ER4SR 1mm smaller than the old ED29689 which was 6mm.

ER4S - ED29689

ER4SR - RAB 32063

no
If you type in er4sr driver + johnyang1997 on Google you'll find your answer.

 

[TLEDDY]

I cannot pretend to understand the technical portion of the discussion! So, my question is simple:

How is the $41,000.00 investment recovered??

Tillman

 

[solderdude]

It isn't, that's why Amir didn't buy it and bought a much less expensive one.

 

[TLEDDY]

It isn't, that's why Amir didn't buy it and bought a much less expensive one.

Thanks! I missed that!

What did he purchase?

Tillman

 

[NTK]

Thanks! I missed that!

What did he purchase?

Tillman

Please see this post.
https://www.audiosciencereview.com/.../asr-getting-into-measuring-headphones.18086/