-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Physical Characteristics Of Headphones Used In Psychophysical Experiments
- Thread starter Wombat
- Start date
As is conventional, they dismiss phase response as immaterial as long as it is "smooth". Has anyone tried DSP-ing headphones to flatten the phase response and get the ideal step response? (In fact, phase seems so consistent across headphones that you could do it from the average of the graphs in that paper without having to make measurements).
Last edited:
No...
Here's a "first try" measurement out of curiosity on my HD650 - just stick the UMIK-1 into the left earpiece and go for it... no "eq", 1/48 smoothing, about 20 feet of cable from the DAC.
1/6 smoothing Group Delay
Last edited:
That all looks sorta better than I might have expected, given the sub-optimal technique.
From the paper above:
For anyone interested:
From the paper above:
For anyone interested:
The distortion measurement above... Looking closer...
I can't say if the measurement is even sensible because all the harmonics REW thinks it sees lie below the noise floor.
I suppose I can say the level of distortion presented by the headphone is negligible.
I can't say if the measurement is even sensible because all the harmonics REW thinks it sees lie below the noise floor.
I suppose I can say the level of distortion presented by the headphone is negligible.
Last edited:
- Joined
- Feb 23, 2016
- Messages
- 20,701
- Likes
- 37,442
So given your above results, one would think any distortion lies at least 15 db or more below the noise floor. Good job noticing what is going on with the harmonic distortion.
It was all somewhat illuminating, I suppose.
I would have guessed "higher" for distortion and "worse" for the other items.
Nice to see reasonable curves for the Impulse and Step and Phase.
This headphone is 20 or more feet from a little LME49680 output stage, $3.27 each...
---
The low-end rolloff - because the 'phone as measured is open air and not making a little pressure chamber on the sides of my head?
I would have guessed "higher" for distortion and "worse" for the other items.
Nice to see reasonable curves for the Impulse and Step and Phase.
This headphone is 20 or more feet from a little LME49680 output stage, $3.27 each...
---
The low-end rolloff - because the 'phone as measured is open air and not making a little pressure chamber on the sides of my head?
Last edited:
I made my own test rig using a Behringer calibrated mic and a plastic cup filled with foam. Here's the frequency response plot of my HD650s before EQ. Looks sufficiently different from yours, curious if I need to make some adjustments to my test equipment 
- Joined
- Feb 23, 2016
- Messages
- 20,701
- Likes
- 37,442
I made my own test rig using a Behringer calibrated mic and a plastic cup filled with foam. Here's the frequency response plot of my HD650s before EQ. Looks sufficiently different from yours, curious if I need to make some adjustments to my test equipment
View attachment 11271
Looks like your cup is creating problems on the measurement if I were guessing. Easy to check. Do the measurement again the way Ray did. Just put the tip of the microphone into the ear piece and see what you get.
We probably both do...
I got what I was more interested in, though, transient, phase, and distortion data.
I figure the frequency response is pretty good, and they (audibly, to deaf me) match my (corrected) speakers (or maybe the speakers match the headphones) quite well.
You made a little Room for them to play into, I didn't.
Could you post a few milliseconds of Impulse and Step Response from that measurement?
Last edited:
- Joined
- Feb 23, 2016
- Messages
- 20,701
- Likes
- 37,442
You guys could fix the issues of low end measurements without including an enclosed space. Get a big sheet of plywood or particle board. Make a hole the size of the earpiece. Mount to it with a bit of pressure to seal, and measure on the other side. The plywood would be effectively an infinite baffle down to a pretty low frequency. Easy for me to imagine test rigs you guys need to make as long as I don't need to do it.
Last edited:
Looks like lots of reflections in your little room...
Zoom so only 4ms show on the right side of the Impulse...
The step is pretty good, though.
Zoom so only 4ms show on the right side of the Impulse...
The step is pretty good, though.
Yes, the cup was likely causing most of the reflections. My goal was to get a repeatable measurement for EQ. Next time I'll try to find a less reflective material than plastic
Dimitri
Senior Member
How about comparing the measured output from whatever DAC was used, to the headphone output of the Behringer UMC-xxx ?
Mmm... Ok...
Lots of variables here.
Headphone left channel is hand-held next to mic
Gold - REW sweep tone -> USB -> Signstek -> Optical -> Switch -> Behringer DEQ2496 -> miniDSP OpenDRC-DI -> Benchmark DAC2 -> HD650 -> UMIK-1 microphone -> USB -> REW (no EQ engaged)
Red - REW sweep tone -> USB -> UMC 202HD DAC -> Headphone out -> HD650 -> Behringer ECM8000 microphone -> UMC 202HD ADC -> USB -> REW
Something wrong on distortion display, maybe... it gets flaky in REW sometimes
Pretty close, though there might be a problem with me or the gear on the Red trace.
Further investigation obviously required.
Sometime.
---
Noise Floor / Distortion - didn't match SPL calibration - need to make/find a calibration file for the ECM8000
Last edited:
While looking for a calibration file:
This is new...
This is new...
Dimitri
Senior Member
It's not you, it's them.
Either they decided to do it or clever hacker decided for them
PS: Thanks for the plots btw
- Joined
- Feb 23, 2016
- Messages
- 20,701
- Likes
- 37,442
@RayDunzl
You could subtract the uncalibrated ECM8000 response from the calibrated Umik1 response in REW. I assume you know about that function. That file will cal the ECM8000 to the UMIK1 as a reference.
Then change parameters to give a good graph of the difference file on screen with the difference result. Take a screen shot. You can then import that file into Web Plot digitzier and it will spit out a CSV file that REW can use for calibration. In a perfect world once you have that cal file if you position both mics exactly in the same place they will give exactly the same result.
https://automeris.io/WebPlotDigitizer/ You just upload the file in a browser window. You then use your mouse to pick two points on the x and y axis and give values. Then use your mouse to paint across the graph line. It populates it with closely spaced dots and spits out a CSV file with proper values.
EDIT to ADD: I forgot that REW when it does A-B it isn't subtracting db from db, it is subtracting the values mathematically those represent. So the resulting A-B curve will not be the one you really are looking for on calibration.
You could subtract the uncalibrated ECM8000 response from the calibrated Umik1 response in REW. I assume you know about that function. That file will cal the ECM8000 to the UMIK1 as a reference.
Then change parameters to give a good graph of the difference file on screen with the difference result. Take a screen shot. You can then import that file into Web Plot digitzier and it will spit out a CSV file that REW can use for calibration. In a perfect world once you have that cal file if you position both mics exactly in the same place they will give exactly the same result.
https://automeris.io/WebPlotDigitizer/ You just upload the file in a browser window. You then use your mouse to pick two points on the x and y axis and give values. Then use your mouse to paint across the graph line. It populates it with closely spaced dots and spits out a CSV file with proper values.
EDIT to ADD: I forgot that REW when it does A-B it isn't subtracting db from db, it is subtracting the values mathematically those represent. So the resulting A-B curve will not be the one you really are looking for on calibration.
Last edited:
Similar threads
