Building a Headphone Measurement Lab

[Tyll Hertsens]

Quote:

Originally Posted by jp11801 /img/forum/go_quote.gif Possibly more important than the raw data is the analysis of the data. Typically raw data is misinterpreted by under-informed but well meaning hobbyists. So just not measuring but providing insight to what the data could mean is where the value lies in my eyes.

Just addressed that while you were posting, but it's worth repeating, because while I'll be able to get some decent measurements pretty quick, it'll take quite a while to learn how to massage the data so that it's meaningful to a fairly uneducated eye.

Quote:

Originally Posted by JPnum-nums Amp measurements would be very cool as well.

Ooops! Forgot to address that. I absolutely will be measuring amps. There's a lot of stuff for me, myself, and I to do, so things will have to happen a bit sequentially, but I'm going there.

Actually, that brings up another point, I can acquire data all day long, but it's got to get displayed somehow. HeadRooms graph tool will continue to be a good resource, but it's likely to contain the data that HeadRoom needs for it's own purposes. So I've got to put up my own website and graphing tool.

I think it's important for me to focus on the content creation side and not spend weeks and months getting self educated on building dBase web widgets. I guess what I'm saying here is: If you want this data and lots of it with any type of speed, I'm going to need volunteer help, particularly on the web back end side. If you are adept at working with Drupal, PhP, MySQL, Apache, Linux, and the like, and would like to help, please PM me.

I also want to commit to you all that this venture is NOT for my benefit. I will be starting a non-profit 501(c)(3) company which will be dedicated to this effort. That company will belong to the hobby, not me, and it will be dedicated to providing great information and listening experiences for hobbyists. More to come on this in future as I'm just beginning to gather capable members for the Board of Directors and Advisory Board, but I wanted to let all of you know that your volunteer efforts will not be going to lining my pockets (though this does need to be a job for me, and the board will control that aspect). You're volunteer efforts will be going towards building a solid resource for yourselves as hobbyists, and for the general public as the distilled wisdom of the members here.


Argh, yabber, yabber, yabber, out for that walk.

 

[3602]

Tyll,
I would very much like you to do something about headphone burn-in. See whether the data changes significantly after n hours of burn-in with various popular ways (pink noise, frequency sweep, etc).

 

[StevieDvd]

Not sure if it's been mentioned before but could you use the same equipment to rate the isolation factor of a set of headphones too? I'm thinking of something like a sound source in the box and recording the level heard by the dummy head.

Or in terms of the questions that comes up 'How do headphone xyz isolate in my office/home/bus/train?' etc.

PS What have you done with the real Tyll, we all know he'd never wear such a plain t-shirt?

 

[kboe]

I'm not one for the science end of this hobby but this all sounds fantastic and quite useful. I'm so happy that even though you've left HeadRoom, you found something that your still highly passionate about. Cheers mate!

 

[Tyll Hertsens]

Quote:

Originally Posted by 3602 /img/forum/go_quote.gif Tyll, I would very much like you to do something about headphone burn-in. See whether the data changes significantly after n hours of burn-in with various popular ways (pink noise, frequency sweep, etc).

I would like to do this, and am planning now to do some measurements in July. I've got vacation plans that I've had for a yaer then and won't be around the house much. I'm planning to do a measurement at that time re burn-in where I can let the experiment run for a number of weeks without disturbing the headphones. Again, the big issue here is that simply removing the headphones and replacing them will disturb the measurements likely far more than the effect of burn-in itself, so it is important that the experiment run for it's duration without disturbance.

I'm thinking AKG K701s are the can to use since it's so commonly held that they need to break in so much. I tend to hold that burn-in really doesn't have a large effect, but the K701 makes a believer out of me that there is some effect. Please let me know if you think there's a better headphone for that purpose.

Quote:

Originally Posted by StevieDvd /img/forum/go_quote.gif Not sure if it's been mentioned before but could you use the same equipment to rate the isolation factor of a set of headphones too? I'm thinking of something like a sound source in the box and recording the level heard by the dummy head.

You'll see exactly that in just a few more posts.

Quote:

Originally Posted by StevieDvd PS What have you done with the real Tyll, we all know he'd never wear such a plain t-shirt?

You know, I was just thinking that yesterday. The truth is that the real Tyll wears t-shirts and sweatpants almost exclusively. The Hawaiian shirt thing is me caving to the need to be promotional in my past life. I'll still wear them, of course, at shows and stuff, but I reckon you'll see me as I really am from now on. I'll try to remember to shave and get a haircut regularly.

Quote:

Originally Posted by kboe /img/forum/go_quote.gif Cheers mate!

Cheers, Kev.


Roit, starting the box post now! See ya in a bit.

 

[Bubu1]

Wow, this is all great news! For those of us who are scientifically inclined (I'm a molecular biologist) but not experts electronics, this will be a fantastic resource. Great to hear the passion in your writing again!

Will you be able to or consider measuring some of these gizmos like the M2tech Hiface, Bel Canto USB lnk etc. to see how they perform?

Not to derail the thread but did you happen to "make off" with any amps/dacs from headroom? You're the reason I recently purchased the ultra desktop stack and although you don't work for Headroom now, I know that you developed it and I have to say you did a great job with this piece of kit!!
Best of luck!!
Tom

 

[xnor]

Cool, that sounds like a very nice idea.

I have some suggestions / ideas for improvements over the headroom measurements:

Frequency response:

The smoothing used should be specified, just like the equalization (FF, DF ...).
And take a look at ARTA.

CSD can be done, as seen here with the AH-D1001. The question is if it's just something that just looks nice but has no real use.

Impedance:

The y-axis should be dynamic. Just look at the T1 headroom graph, lol.


And some SRH840 examples:

Distortion:

Impulse response:

All headphone measurements should be done at a reference level, with the same equipment. The test setup should be documented (including things like output impedance of the amp...).

RMAA can be used to test DACs and amps. When testing amps with RMAA there should be at least a dummy load connected (e.g. 16, 32, 320 ... ohms) or even better: headphones.

When testing such equipment by looking at the resulting headphone measurements, you ideally just switch the headphone plug from setup A to B without touching the headphone itself.


It would be very cool to be able to download some raw data (e.g. the impulse response file, impedance data ...) which you can do some funny things with, like plotting the FR with different output impedances.

I'd love to help you to code some of the backend stuff but the next ~3 months I'll be very, very busy (job + exams + writing thesis + ...). So I don't actually have time to code in my free time, but maybe I can help you with one or the other coding related problem/question. Feel free to PM me.

 

[Tyll Hertsens]

Quote:

Originally Posted by xnor /img/forum/go_quote.gif Cool, that sounds like a very nice idea. I have some suggestions / ideas for improvements over the headroom measurements:

Wow, great post mang. All good ideas.

As we get things rolling I certainly would love for you to lend a hand.

 

[Tyll Hertsens]

Building the Isolation Box


Allllrightythen …


There are a number of reasons why we need a box to measure headphones in:

1. You want to measure in a quiet environment to keep the noise in measurements low.
2. We need to expose the head to a known broadband noise environment to get isolation measurements.
3. You need an repeatable and damped acoustic environment for measurement repeatability. Not only in the isolation measurement, but open back headphones allow sound to leak out, reflect of walls/objects, and potentially affect measurements.
4. Now, I could modify a closet or room in my house to do this, but I kinda don't have the room, and I'd like to be able to bring the measurement system to CanJam, so it has to be a box to be portable in this case; sized somewhat between a washing machine and refrigerator at most.

Isolation


To make the box quiet, you have to build its walls that don't pass sound. There is two ways a wall stops sound transmission: by being immovable (massive and stiff) and reflecting it; or by being absorptive (fiberglass and foam disturbing the passage of air) and attenuating it. The very best killer isolation rooms are a “room within a room” where the room's walls are concrete/cinderblock, and the space between the rooms is filled with absorbing material. Here's an interesting Wiki page.


The room I built at HeadRoom, was a normal room that I put multiple layers of drywall on, and then built an air space and then a wall of fiberglass and foam absorption material. It was more of an anechoic chamber than an isolation chamber. Also it was hard to make a door that sealed well out of a standard door. Bottom line, it didn't isolate very well, and we had to do all the measuring at night or early in the morning so that it was quiet in the building when we did the measurements.


Unfortunately, the “room within a room” method makes for thick walls and very difficult problems making doors that seal. To make the box small enough to fit through a door and still have thick enough walls, I'd end up with not enough space on the inside for the head. So, I've opted for a box with stiff and massive walls using 3/4” plywood and a layer of drywall heavily glued on the inside. I would have gone with two layers of sheetrock, but 200 lbs of box seemed more portable than 300 lbs of box.


Broadband Noise Environment


To get the broadband noise environment we simply mount a speaker in the box and play pink noise. Now, there's a bit of a problem that you need to solve which is that you want the noise to be coming at the head from as many directions as possible. When we had a speaker in the HeadRoom anechoic chamber the sound came from one direction only, and if you look at our isolation mesurements they look a bit lumpy. I'm guessing that may be in part from the point sound source exciting a small number of discrete resonances in the acoustic system being measured. So in the case of this box, I want the sound to be able to bounce around in the box some so that it provides more of a diffuse field of sound and hopefully will provide a more representative measure of the headphone's attenuation.




Repeatable and Damped Acoustic Environment
Lastly, since open back cans radiate noise outward, it's possible for that sound to reflect off the walls of the box and get back into the headphones to interfere with the measurement. So, while you want some “liveliness” (reverberation) on the inside of the box for the isolation measurement, you also want to damp it and prevent direct reflections to keep the measurement of open cans clean; and since resonances and reflections are positionally sensitive, you want to be able to position the head in the same place in the box for every measurement.


The bane of any technical development is the need to compromise. We're simply not going to get 30dB of isolation and make it portable. I was hoping for 20dB; looks like I'm currently getting about 15dB … I think I might be able to improve that just a bit with some tweaking, but I think it will be adequate since my house is pretty quiet.


Let's look at some pix:

Spring in a Montana driveway ....

Measure twice, cut once.

Putting the sides together.

Starting to make the box. (Man, I need a tan.)

Mad man at work.

Getting there.

Time to add some drywall.

Gluin' and screwin'.

Looks like a good avatar!

Got to make it look pretty.

The door has to close air tight, so I put noeprene on the door as a gasket ....

... and beefy latches.

Have to get the cables in there somehow; a hole filled with clay works well.

I'm working on thinking up another entry at the door for temporary cables at the show that doesn't use clay.


I've used partially inflated wheelbarrow inner tubes underneath as vibration isolation.

Inside is the head on it's stand, and it's a bit difficult to see but there is some 1x1 screwed into the bottom of the box that the stand indexes against for positional repeatability. You'll also note the absorbing foam to either side of the head and that the head is on an angle; that reduces reflected sound from the headphones getting back into the cans.

Here's a clearer view of the speaker; it's pointing sideways somewhat to cause more reflected than direct signal for the isolation measurement. The 2x4 on the side walls stiffens the walls and also stirs up the sound a bit.

And here's the whole kit and kaboodle.

Next time we'll talk about the gear, eh?

 

[mrarroyo]

Darn, nice testing gear and isolation booth. Best of luck in this new career!

 

[Valens7]

Wow! I'm reading this, and you actually seem like a pretty cool guy.

I look forward to enjoying the fruits of your considerable labor.

 

[rsaavedra]

Absolutely awesome Tyll!!!!

While it is sort of a bad news to learn that you left Headroom, this is really great news!

I hope we are getting closer to that day when we can finally see some measurements of the HP2, R10, and other big leaguers that are still missing.

 

[rsaavedra]

Quote:

Originally Posted by Tyll Hertsens /img/forum/go_quote.gif Broadband Noise Environment To get the broadband noise environment we simply mount a speaker in the box and play pink noise. Now, there's a bit of a problem that you need to solve which is that you want the noise to be coming at the head from as many directions as possible. When we had a speaker in the HeadRoom anechoic chamber the sound came from one direction only, and if you look at our isolation mesurements they look a bit lumpy. I'm guessing that may be in part from the point sound source exciting a small number of discrete resonances in the acoustic system being measured. So in the case of this box, I want the sound to be able to bounce around in the box some so that it provides more of a diffuse field of sound and hopefully will provide a more representative measure of the headphone's attenuation.

I'm wondering Tyll, how about a couple of dipole speakers? Or maybe a quadpole speaker? Just a thought, maybe overkill though.

 

[3602]

Whoa, awesome! Can't wait to see the results!

 

[xnor]

Hehe that's amazing. When I saw the first pic I finally knew what your avatar was all about.