Low frequency roll-off? - MiniDSP EARS on Windows 10

[tomai]

Hi!

I'm measuring headphone frequency response using REW v5.2 and miniDSP EARS on Windows 10.

I'm observing a 6dB/octave roll off below 80Hz when comparing new measurements to previous measurements of the same components. Previous measurements were done on an older laptop some versions of REW ago. Is there something in Windows 10 settings or REW preferences that could be causing this? I've disabled Windows audio enhancements for the output and input already.

Appreciate any advice here!

Thanks.

 

[skid00]

-How- are you measuring the HP's? I built a flat surface with a hole in the middle, and poked a UMIK thru it, then measured various positions. Each position varied wildly in response curve.

 

[John Mulcahy]

Probably easiest to post an example mdat to check, but there aren't any REW settings that roll off the low end.

 

[tomai]

Hi - thanks for getting back and apologies for the slow follow up, but I'm still having this problem intermittently. Attached is an mdat file with left/right frequency responses for 4 different sets of headphones over 2 different sessions.

Session 1: 'Headphone 10', Audio-Technica M50xBT2, DT770 Pro - all with 10dB/oct roll off below 100Hz
Session 2: 'Session 2 Headphone 10' (a different sample to the first session but should have an equivalent response) - response recorded as expected

There was varying background noise during both sessions, but this doesn't appear to correlate with the low end roll-off.

I suspected Windows 10 audio driver settings but I have disabled all audio enhancements as per the attached images.

Any help with this would be really appreciated! Thanks!

 

[tomai]

-How- are you measuring the HP's? I built a flat surface with a hole in the middle, and poked a UMIK thru it, then measured various positions. Each position varied wildly in response curve.

I'm using miniDSP EARS - responses do vary a lot, particularly below 500Hz, depending on position and earpad sealing. I either use a fixed position for all measurements (if possible) or average a range of positions.

 

[sm52]

I have a Beyerdynamic DT 770 Pro 32 Ohm headphones. And external sound card Audient iD14. These are measurements of one and two channels.

 

[sam_adams]

@tomai, you may want to take a look at the headphone measurements that have been made over at ASR. Test jigs can vary quite widely in their results. Your measurements show a distinct peaking at around 4.5KHz:

That might be something with the measurement apparatus or something else entirely.

I did notice that you had the cal files reversed for the first two measurements.

The Harman Kardon OE trace is the Harman Headphone Target curve for over-ear headphones superimposed on your measurements. If you have not already done so, you might want to research that.

A Statistical Model that Predicts Listeners’ Preference Ratings of In-Ear Headphones: Part 1—Listening Test Results and Acoustic Measurements

A Statistical Model that Predicts Listeners’ Preference Ratings of In-Ear Headphones: Part 2—Development and Validation of the Model

A Statistical Model that Predicts Listeners’ Preference Ratings of Around-Ear and On-Ear Headphones

You can download a copy of the Harman curves from this AV NIRVANA post and import them into your measurements for comparison.

One other thing, given the nature of the measurement apparatus, I don't know if using the opposite channel as the acoustic reference is of any benefit since that channel won't be 'heard' by the channel being measured since there are two separate mic capsules with separate cal files. @John Mulcahy can comment as to whether this matters or not.

 

[tomai]

@sam_adams - thanks, this is really useful.