CHANGSTAR: Audiophile Headphone Reviews and Early 90s Style BBS
Main Deck => Discussion for Registered Members Only => Topic started by: Hands on April 18, 2014, 11:17:13 AM
-
Keith (bluemonkeyflyer at Head-Fi) sent me one of his measurement kits to play around with. It's a Panasonic WM61A mic hot glued in a tri-flange tip. A phantom PSU was provided. Basically, you insert the mic into your ear, the mic faces outwards, and you put headphones on like normal (other ear has plug in it).
This has some obvious upsides and downsides, and it's not likely to be as consistent or accurate as measurements from a dedicated rig with good compensations. One obvious downside is that you'd need to switch the mic to your other ear to measure the other channel, thus hindering measurement repeatability and consistency, among other issues.
For now, I've just been experimenting with trying to get results that make sense, are somewhat repeatable/consistent, and not too far off what you might see elsewhere. I have a very basic FR compensation ready, which corrects for the bass-roll off from either my line-out or mic-in (not sure which) and corrects for the slight treble increase most WM61As exhibit. I tweaked this a bit, as 20Hz was listed as -15dB down when calibrating through software, but I set my compensation instead to -10dB. Subjectively, this gives what looks to be more accurate results, but it may actually be more correct if I left the bass compensation as it was (stronger).
Lately, I've been testing as follows: Creative SB1240 line-out -> Leckerton UHA-6S MkII line-in -> Leck. HPO to headphone -> mic in ear -> phantom psu -> SB1240 mic-in
What I need to do is get something setup so I can accurately measure the SPL of a headphone, thus making it possible for me to more accurately calibrate all of my software and volume settings for my mic-in. Currently, I just have my mic-in set to a static volume setting (80/100) and adjust the HP volume accordingly on the amp so that it gives around -10dB at 1KHz during ARTA tests. I set the mic volume setting simply by trial and error based on how "off" the THD measurements looked, and 80 seemed to not be wildly off, maybe. FR didn't change much regardless. Basically, I used trial-and-error to produce results that very loosely matched Marv's measurements, not to say my methods are at all similar to what he does.
I think the only headphone I have on hand that isn't modified in some way is the UE6000. I'll measure that next. For now, I measured my Mad Dog 3.2, which I applied several front-damping tweaks to a few weeks back (damping dot in middle, single ply tissue and foam disks around that, mostly). The idea was to smooth out the treble and make them a bit less U-shaped sounding. Though not comparable on many levels, Marv has the two different MD3.0 measurements on this site, Dan/MrSpeakers has posted measurements from time to time that are scattered around at various sites, and Tyll has measurements of a supposedly faulty MD3.0, I believe, listed at some URL that doesn't appear to be posted publicly on his site (hint: open MadDog measurement listed on his site, which is old, and replace the "A" at the end of the URL with "2013").
L/R FR, HD, and CSD are attached for my personally tweaked MD3.2. Any advice on HP measurements is appreciated! I can post further details on my setup and settings as well.
One thing I've noticed in my early measurement experiments is that it seems they all show the same dip around 8KHz, but I'll have to keep testing and investigating. Often this gets at least partially filled out in CSDs. Channel matching is not too bad, given the method, and might even be within spec or what other MD measurements have shown in the past. Let's not also forget my tweaks were handmade, and not precisely at that!
Edit: Updated R CSD, did not have data open to adjust L
-
Logitech UE6000 in passive mode. Like I did for the MD3.2, I had the mic at 80/100 and tried to get 1KHz to around -10dB. Measured right channel after without changing amp volume to see if there were any major channel differences. If you see anything particularly weird, I did do these multiple times to rule out easy artifacts.
Tyll's measurements: http://www.innerfidelity.com/images/LogitechUE6000passive.pdf (http://www.innerfidelity.com/images/LogitechUE6000passive.pdf)
I haven't been able to find other measurements of this headphone, but mine do share similarities with his. I'm not sure if the bass differences between the channels is accurate, but I did pop open the left side of the headphone a couple times (for science), but not the right. The adhesive for the pads and for the driver baffle are probably not sealing as well as they should now. There are a few possible reasons for this, one potentially being due to my measurement kit and its inherent difficulties.
I hear these as closer to what my measurements suggest than Tyll's. The first thing you'll notice about them is the additional, but not terribly distorted, bass (cleans up once the pads warm up and seal better). Other than that, I found them to be fairly even sounding, but not perfect. Definitely not as laid-back and dark as Tyll's measurements suggest.
-
UE6000 in NC mode. Started with the R side this time instead of L, but the practice was the same.
Tyll's measurements: http://www.innerfidelity.com/images/LogitechUE6000noisecanceling.pdf (http://www.innerfidelity.com/images/LogitechUE6000noisecanceling.pdf)
-
Hi Hans,
Very cool!
Can you tell me more about the measurement kit? How much did it cost? I couldn't find anything on Head-Fi about it.
I am interested...
Thanks,
-Darin
-
Great work hans.
My own measurement rig (literally a piece of wood and an omni mic) gives me consistent measurements like yours. No compensation needed for relative differences. I might look into using the WM61A (I have a few on hand) and embarking down that path to see where it takes me.
Also, the 8k dip appears in my plots as well. Likely some sort of interaction as the wavelength is significant to the size of the coupler at that point.
Also, a while back Mr. Purrin promised to make a thread regarding the specifics of his measurement rig. He's busy with life crap ATM, but I'm expecting to learn much more when El Capitan decides to share his secret booty walk the plank2
-
Hey Darin
Keith is a really great guy. His approach can be found here: http://www.head-fi.org/t/618659/fostex-t50rp-incremental-mods-and-measurements#user_DIYMeasurementKit (http://www.head-fi.org/t/618659/fostex-t50rp-incremental-mods-and-measurements#user_DIYMeasurementKit). The principal behind this is to help people who like to mod headpones, not to generate more measurements for the interwebs.
-
Hey Darin
Keith is a really great guy. His approach can be found here: http://www.head-fi.org/t/618659/fostex-t50rp-incremental-mods-and-measurements#user_DIYMeasurementKit (http://www.head-fi.org/t/618659/fostex-t50rp-incremental-mods-and-measurements#user_DIYMeasurementKit). The principal behind this is to help people who like to mod headpones, not to generate more measurements for the interwebs.
Thanks for the info. I will read up on that.
No, I am not looking to add to the measurements being posted online. I am actually looking to do measurements using in-ear microphones for my Out Of Your Head software. I also want to add headphone EQ compensation into my software.
-
That post was not meant to warn you off :) , I meant it more as a methodology for self learning .
I would love to see some of your experiments with DSP
..dB
-
That post was not meant to warn you off :) , I meant it more as a methodology for self learning .
I would love to see some of your experiments with DSP
..dB
The result of my experimentation is available as a commercial software product, Out Of Your Head (https://fongaudio.com/out-of-your-head-software/) . There is a Windows free trial version available on my website.
If you don't have a Windows machine, you can hear some pre-recorded demo files here:
http://fongaudio.com/demo (http://fongaudio.com/demo)
But I am always looking to improve my methods and my software.
-Darin
-
Could you get the results to be fairly repeatable? You could mod the WM-61 into a wand mic pretty easily too and with the materials, probably build the same rigs as Ultra has done sometime.
-
Can you tell me more about the measurement kit? How much did it cost? I couldn't find anything on Head-Fi about it.
Hey Darin, hopefully dBel got you pointed where you needed to go. Keith actually sent me my kit as a loaner, but the BOM was relatively low when I asked if I could purchase it (I am still in testing phase, actually). He may or may not be willing to make another kit, but you'd have to contact him to find out.
If you have any other specific questions, let me know!
Also, the 8k dip appears in my plots as well. Likely some sort of interaction as the wavelength is significant to the size of the coupler at that point.
I'm actually beginning to believe they're not entirely artifacts. I think they can be to some extent, but when I ran a sine sweep on the MD later with my earballs, there was definitely a dip there. In fact, it actually seemed to match the measurements fairly closely, but I could be biased there. I'm also not super great in the 20Hz-1KHz band or the 10KHz+ band, but what I hear from 1-10KHz on most headphones when running a manual sine sweep app does seem to match the measurements decently, though perhaps only in characteristics and not absolutely accurate amplitude/compensation.
TBH, the only headphone I have on hand that has been measured and documented elsewhere is the HE-500 with jerg pads. But, what I have doesn't measure like how Marv's did. On the other hand, my pads are not modded identically, and I have applied a couple other tweaks that I don't believe Marv had for his measurements. And while, again, I could be biased, playing around with SineGen seems to more confirm my measurements than not. I am planning on playing around with the HE-500 and will come back with findings, if I find anything.
Could you get the results to be fairly repeatable? You could mod the WM-61 into a wand mic pretty easily too and with the materials, probably build the same rigs as Ultra has done sometime.
I've taken a lot of test measurements, and even with removing and re-seating headphones, or making slight adjustments, you'd be surprised how much consistency and repeatability I can get. Perhaps I'll try to post results of that.
WM61A could definitely be modded for purposes like that, which is an option I've left on the table for later.
-
OK, I've got a couple things for this post and the next. First, I've revised my compensation curve to account for a hump in the ~2-4KHz area that measurements were producing. This was most obvious on the HE-500 and modded HD598 I measured, which I won't be posting just yet. I spent quite a bit of time with subjective listening and careful sine sweep examinations with SineGen to help me get the compensation curve closest to what I think I hear. When compensating for the hardware bass roll-off, the mic's slight sensitivity in the upper treble (using calibration settings found online), and the stronger-than-what-I-actually-hear hump around 2-4KHz, I'd argue these measurements are not too far off from what I'm actually hearing (i.e. they might eventually hold enough value for others to use as general measurements, given many more tests and validation with many more headphone samples and once I get my methods down). I do still need to dial in on SPL calibrations and such to produce the most accurate THD measurements possible.
Also, I did a basic test for measurement repeatability/consistency, but only on one HP for now. Attached are 4 measurements of the tweaked Mad Dog 3.2 left channel. In between each test, I removed the headphones completely, wiggled and pushed on the mic in my ear a bit, and put the headphones back on for best comfort and assumed seal/fit for sound. Slight differences show, but at a smaller level of variation than I had initially expected (keeping in mind these measurements are zoomed to a fairly large scale). I also attached the CSD of the 4th measurement. I believe the FR/HD scale is different than last time around for better readability.
Treble response through SineGen is on-par with what measurements indicate, including that dip (though do check to see if dips get smoothed out any in CSDs).
-
Attached the FR/HD/CSD from the UE6000's left channel using the new compensation curve. I was able to get a better bass seal this time, it seems, but you can compare with the earlier measurements to get an idea of how repeatable/consistent tests are (same for the MD measurements from earlier and in the previous post).
Also attached the L/R, FR/HD/CSD measurements of my modded Creative Aurvana Live! (CAL). Again, matches what I hear pretty closely, down to the channel imbalances (these were not precisely modded). Subjectively, these are fairly neutral with a warm/dark tilt. Bass gets pretty muddy when you turn it up, hence the high harmonic distortion. Good detail, but slightly shelved mids. Does really, really well at low volumes. The bass and overall sound cleans up, the details still shine through...very nice. Even at the bare minimum output volume from most devices, these still do really well.
Tyll's measurements of the un-modded CAL: http://www.innerfidelity.com/images/CreativeAurvana.pdf (http://www.innerfidelity.com/images/CreativeAurvana.pdf)
Golden Ears: http://en.goldenears.net/4446 (http://en.goldenears.net/4446)
1. UE6000 L
2. UE6000 L CSD
3. CAL L
4. CAL R
5. CAL L CSD
6. CAL R CSD
-
I don't mean to imply that they are artifacts, but rather that the microstructure above a certain point is caused by that sort of interaction. You'd expect those same peaks and dips to be audible if the coupler is indeed your own head/ear. :D
-
Absolutely, but it doesn't hurt to double check, especially when there is junk in your ear that could cause otherwise abnormal interference. :) I was glad to see it usually matched subjectively.
Edit: Assuming some level of accuracy with these measurements, I think my measurements are a bit telling of my personal preferences based on how the two tweaked-by-myself-in-some-manner headphones measure. Fairly linear/neutral-ish, dark-ish tilt. :)) I'd have to try some wildly different headphones on my setup first, though.
BTW, I don't think the MD has as much bass as you'd expect based on these measurements, but that often seems to be true of the T50RP.
Edit 2: Russian measurements of what I believe is the Mad Dog 3.2: http://personalaudio.ru/raa/otchety/naushniki/mrspeakers-mad-dog/ (http://personalaudio.ru/raa/otchety/naushniki/mrspeakers-mad-dog/) (compare against my personally tweaked version)
-
Between what Keith has written on this kit, my friend's successful attempt and yours hans, I'm sure I'd be able to get mine right. Some day... When I have the time......
-
Great job Hans! I might have read wrong, but it seems you are using your own head/ears as coupling between the cans and the mic. This could make things somewhat closer to Tyll's setup, though according to his discussion here (http://www.innerfidelity.com/content/ear-opening-experience-chesky-ultimate-headphone-demonstration-disc), he might be using a smaller mic that goes dipper into the ear, therefore requiring different compensation curves.
I believe something similar to your approach was discussed here (http://www.head-fi.org/t/641860/hrtf-and-binaural-measurements-of-sennheiser-hd650-hd700-akg-k550). Hopefully the links can be of assistance.
-
That's correct, UB. Same basic method as Keith/BlueMonkeyFlier on Head-Fi...I did get the kit from him, after all! But I did deviate from him in my choice of software and by building my own, fairly basic compensation curve.
My current goal is to test as many headphones as I can, especially already-measured cans on various setups (trying to avoid brands or headphones with known QC and variance issues). Once I have the gear necessary to calibrate everything, as I still have no real way for me to guarantee I'm running a headphone at, say, 90dB, test headphones out, and tweak my compensation curve, I'll evaluate from there whether or not I can produce measurements this way with enough accuracy and repeatability/consistency to provide general measurements of any headphone to the public. If it's not quite up to par, and I decide to not build a more dedicated rig, it's still very useful for my modding endeavors!
I have to say, I'm surprised I'm not seeing more seal-related issues. I figured my long-ish hair, for a male, would cause problems (I keep it pushed back/held behind ears for tests), but that seems to be much less of an issue than I anticipated. (I wonder if it's at all due to the fact I have a lot of thick, straight, smooth hair...I do use argan oil in it daily, which may help? LOL)
-
Some updates on my measurement stuff:
1. Slightly tweaked my compensation curve again. While most WM61As show a treble sensitivity that peaks around 15KHz (measurements vary from unit to unit, but characteristic is similar), I wasn't entirely sure the measured HP results quite matched what I was hearing subjectively (i.e. pushed down the around around 10KHz a touch too much). I also tweaked the area around 2-5KHz for a touch more attenuation after compensation.
2. Invested a whopping, oh, $15-20 on a digital SPL meter. The idea was that I could use that to, at least somewhat, more accurately say, "This headphone is outputting 90dB at 1KHz." Thus, I would be able to more accurately set my mic-in volume setting so that my computer would also show 90dB when playing a 90dB signal. I haven't quite taken it that far, yet, but I have used the meter to get a rough idea of what the headphone is putting out.
I found it more difficult that anticipated to properly measure the headphone's output levels. Where exactly do you place and angle the meter relative to the driver? Is some sort of sealed environment necessary or not? And so on. I experimented with open measurements and sealed-environment measurements and played around a lot with position and angling of the meter.
The sealed environment method made it more difficult to position and angle the meter in a fashion similar to the placement and angling of my ears, but the results were, as expected, more "stable." Another downside was that, subjectively, getting a 90dB reading in the sealed environment was difficult and sounded much louder than I expected it to.
The open environment gave me a lot of room to play around and position the meter similar to how my ears would be, roughly. With this method, I found I did not have to drive the headphones as hard to get the meter to read ~90dB, and I was actually able to listen to music at that volume setting (albeit a bit uncomfortably due to the MDR-1R's sound). That was more in line with what I expected, so I decided to use this method to estimate the output of a headphone. I used that to verify a rough baseline of an 80/100 mic-in volume setting, and I'm still measuring headphones by matching 1KHz to -10dB in ARTA. I like Marv's tests and am roughly trying to emulate how his look and turn out (also for other's familiarity), though it remains to be seen what I'll end up using in the end.
When accounting for slightly different positioning and angling, and doing a sort of mental averaging of these values in my head, the open environment metering method indicates the headphone is outputting somewhere around 87-91dB at 1KHz. Even if not, I can simply classify it as being fairly loud but not ear shattering during music playback.
MDR-1R Thought and Measurements
Looks good. Presentation, down to the fabric-lined packaging, is excellent. Feels lighter than expected, but it also didn't look like it was made from as much plastic as it is. Fairly comfortable, though I haven't worn it for an extended period. Ears sometimes touch the baffle and driver, which is not the best in terms of comfort.
Sound is decent to good at low volumes, but not stellar relative to everything else about the headphone. At moderate volumes, sound is noticeably more flabby but not overbearing. Not particularly great. At high volumes, sound falls apart even further. Mids and highs become harsh and a bit screeching, and bass gets even tubbier and more uncontrolled.
I only bought them to see if they'd be comfortable for long sessions. I'll mod them down the road if they are. I will note that I have not yet listened to these extensively or ran them through sine sweeps in attempt to verify these measurements, but they are a bit hard to listen to at high volumes. These were purchased used from Amazon Warehouse, and I don't know if the MDR-1R is known for product variance.
For comparison, Marv's measurements: http://www.changstar.com/index.php/topic,1300.msg34693.html#msg34693 (http://www.changstar.com/index.php/topic,1300.msg34693.html#msg34693)
Tyll's: http://www.innerfidelity.com/images/SonyMDR1R.pdf (http://www.innerfidelity.com/images/SonyMDR1R.pdf)
Golden Ears: http://en.goldenears.net/15436 (http://en.goldenears.net/15436)
Personalaudio.ru: http://personalaudio.ru/raa/otchety/naushniki/so/ (http://personalaudio.ru/raa/otchety/naushniki/so/)
I think my modded CAL sounds better.
-
Golden Ears posted UE6000 measurements:
http://en.goldenears.net/index.php?mid=GR_Headphones&document_srl=33893 (http://en.goldenears.net/index.php?mid=GR_Headphones&document_srl=33893)
Given I am still slightly tweaking my compensation curve and haven't updated the UE6000 measurements in a while, how do you think they compare?
Edit: Didn't notice they had both passive and NC measurements available.
-
First, some updates. I'm still tweaking my compensation curve, but I think I'm getting close to narrowing it down to where I want it. Some of my changes are a step back, some a step forward. In particular, I lowered the bass compensation a touch, as I felt my measurements were overstating bass quantities. I did the same for the area around 2-5KHz, which is now actually closer to an earlier compensation curve. The area around 10KHz should have less attenuation than before, but there is a noticeably greater attenuation after that point (really ramps up beyond 12-13KHz).
I adjusted the size of the CSD picture and now have it set to a window of 5ms instead of 4ms. But I made some other changes to them that make them look a bit worse in other ways, IMO, so I'm still working on that.
I also took some quick, rough measurements of my HE-500 with jerg-ish pads, but only on the left channel. You'll notice some clear similarities and clear differences compared to other HE-500 measurements floating around in the tubes. Yes, the area from 2.5-5KHz does have a subjective hump on my pair and drops off in the bass around 40Hz in a sine sweeper. While this does match what I hear closely, keep in mind differences compared to other measurements could be due to the fact that my pads didn't entirely adhere to what jerg suggested and were made sloppily (though, I did use double sided tape for the extension mod). Could also be product variation or just how my head and ears affect the sound. Not sure what the bass distortion blip is, but I'd ignore those. Like I said, these were quick measurements.
Next are some measurements of my torn up/modded alpha pads on the HE-500, still only on the left channel. These have a subjectively lighter but more even response than with my jerg-ish pads. They exhibit slightly better bass extension, but the other pads are noticeably fuller and more impactful sounding. Treble is subjectively smoother and has better details with the alpha pads, and there is noticeably more air and a larger soundstage. Much more speaker-like. It kind of reminds me of my HD598, but noticeably better. Personally, I think you can make the HE-500 sound awesome with these pads. Overall, larger, airier sound (not as powerful, though), bumped up mids and slightly attenuated treble peak, smoother overall, much more comfortable...I like it.
-
I snagged an Alpha Dog that was not quite the correct colour for a discount (darker and a touch more purple/burgundy). I took some really quick measurements of the left channel and will take some better ones in the next day or two. I haven't made any other adjustments to the compensation curve.
I'm not sure if Dan has made further tweaks to the Alpha Dog that he hasn't mentioned, but the revision number listed on the top of the comfort band is higher than the model I had previously tested. This has two of the "dots" applied to the front of the driver on each channel, almost covering the entire middle third of the driver. Could just be that.
This measurement is also using one of the felt damping discs in front of the driver. I will later take measurements of the 3 possible configurations (0-2 discs).
My Mad Dog in its current state has more of a U-shaped sound signature than the Alpha Dog and most obviously has thicker mid-bass presence. The MD has more of a rough, thin, sizzling quality to the top end than the AD. The AD has more low-bass slam and impact, though I'm wondering how much that has to do with the AD's slightly higher distortion down low. Overall distortion seems a bit better and smoother on the AD. It's hard to say, but with 1 felt disc, the AD might be more laid back than my MD (would probably be a different story if the front damping config was the same on the MD). Overall I find the AD smoother and more neutral to listen to than the MD. The MD is a pretty good all-rounder, but the AD does fewer things wrong still, subjectively.
Unless anyone strongly objects based on the results they've seen thus far, I feel I am close to getting my measurement setup and compensation where I want it, wrapping this thread up, and will probably post results in the public section after that point.
-
Love this stuff, Hans, keep up the good work.
It's exactly this kind of adept enthusiast market that will keep manufacturers on their toes.
-
Awesome Hans! Looking forward for your results.
Moar karma pts!
-
Thanks, guys! I need to invest in a better SPL meter for calibration purposes. The one I got only worked for a few days (questionable quality, in the end), so while my results aren't too far off what I'd expect, that's something I'll want to double check. This doesn't change the frequency response so much, but for adjusting the level of the mic-in, it strongly affects distortion results. So, really, it's mostly down to better calibrating everything at this point.
I'm thinking about incorporating a few more tweaks and a couple other common tests with my measurements. I know HeadRoom provides harmonic distortion measurements showing the full spectrum given a single sine wave (500Hz). Those are really quick and easy to grab with ARTA, so why not throw that in with what I do? I use that for calibration purposes before I grab the swept sine FR/CSD anyway, so it wouldn't make sense to not grab it then! Some other type of measurements I'll have to figure out how to do properly, if I want to include them. I'm open to and considering anything that might provide more information about what we're hearing subjectively!
Really, though, it all came down to Keith/BMF over at Head-Fi offering a chance to test out a measurement kit (which I have since purchased at an extremely reasonable price). I was initially afraid his particular kit wouldn't offer accurate results, but I have to say I have been pleasantly surprised by how decently the results actually match what I hear with just a bit of tweaking. I was even more surprised by the relatively consistency the measurements offer, but perhaps this is because I get tactile feedback when placing the headphone on my head. That more likely ensures a good fit/seal and fairly consistent placement. All in all, where as I thought this sort of measurement setup might not be good enough compared to a "dedicated" measurement rig (i.e. for worthwhile public results), I believe it may actually have some inherent benefits/strengths and act more as a good complement to the many dedicated rigs...just as the different dedicated rigs also complement each other with their variances.
Though, I'd also like to rule out product variance and measure one or two of the exact same headphones Marv and UB have also measured, just to see how close my measurements are. Not that they have to match exactly by any means, but as a reference. Then again, some HPs I've measured show a good number of similarities on par, IMO, with variance across other measurement rigs.
-
Very nice, I wish I could double up on my efforts too on the acoustics front. Really looking forward to the reviews!
-
OK, took the time to get some better measurements of the Alpha Dog. These do indeed seem to have a bit of a channel imbalance. It seems to primarily affect the position of male vocals.
Compare to what ultrabike posted, though I'm not sure which revision he measured: http://www.changstar.com/index.php/topic,1208.msg32468.html#msg32468 (http://www.changstar.com/index.php/topic,1208.msg32468.html#msg32468)
1. Alpha Dog, left channel, no felt discs
2. R, no felt discs
3. L, 1 felt disc
4. R, 1 felt disc
-
1. L, 1 felt disc, 2nd take (consistency/repeatability check...headphones were fully removed, mic re-adjusted, before this)
2. R, 1 felt disc, 2nd take
3. L, 2 felt discs
4. R, 2 felt discs
-
I have a new SPL meter on hand, and it doesn't seem to be quite as, well, broken as the other one. I'm hoping to really narrow down on my calibration settings with this. I haven't had a good opportunity to use it yet, though.
Haven't touched my compensation curve recently. I'm feeling pretty good about it for now.
One major revision is a change in the software I use to represent the FR/HD graphs. I am still measuring with ARTA, but I use Excel to generate graphs from the exported CSV data. This gives me a lot more flexibility, though getting the graph template the way I wanted took some time. As I said, overall system calibration isn't spot on yet, but my attached example demonstrates what I'm pushing for in the end (i.e. demonstrates 1KHz at 90dB, but I may include or switch to white noise if this SPL meter isn't a POS like the other and can read it properly). The graph is also larger and cleaner overall, so it should be easier to read. I'll eventually provide a graph that shows only the left and right channels as well.
But, perhaps the largest change is that I am now presenting my measurements as an average from several takes. This graphs is from 3 averaged measurements, though I hope to do 5 for each channel in the future. Each time, I remove the headphone, check the mic to make sure it is seated properly, and put the headphone back on before measuring again. I do try and let the headphones and pads warm up on my head for a few minutes first, but this particular graph was not done as such (fairly quick measurement). In the future, I'll try to include the raw measurements as well.
In general, I'll post more details about my methods once I have things more finalized.
For now, I've attached an example of the left channel on my now-modded MDR-1R. First, I wanted to note that in order for the minor gridlines to show where I wanted, I had to set the minimum on the X-axis to 10Hz instead of 20Hz. I wouldn't worry too much about measurements below 20Hz on my setup.
Though the compensation curves I used for this and my stock MDR-1R measurements aren't exactly the same, they are (quite) close enough that you can see this modded pair is improved in many ways. Some areas may not be as good as a tradeoff/for the greater good. My stock pair sounded pretty darn bad. These are actually fairly decent, though still only at low/medium volumes. (Internal damping mods, HM5 pads modded to mount with original pad ring and also taped to baffle, open tuning hole on top of cups closed by about 2/3. Yes, these did lose some sub-bass and even gained some lower distortion, but they sound much cleaner, less bloated, and less mid-bass heavy overall than they did stock from a subjective standpoint.)
Still using ARTA to generate CSDs. It does what I need with ease.
Let me know if you see anything you like, anything you think could be improved, or just have any thoughts in general!
-
Hans can you take a (close-up) picture of how the mic is seated in your ear ?
I did tests a while ago (for Keith) with mic capsules pushed inside my ear canal and mics mounted on triflanges and compared them to my rig.
-
Sure, I hope this helps! Mmmm, ear (this is the left one, BTW). Ignore the stray hairs...I'll be getting my hair cut soon. ;)
Do you happen to have a link to the comparisons you did for Keith, or have them on hand in one way or another to share? I'd be very interested in the results and details of the comparison (such as was it done with/without compensation curves?). I also need to get my hands on an exact pair of a headphone others like Marv and UB have measured, and I do have that what-seems-to-be-awesome tour Grado on the way to me.
-
http://www.mediafire.com/view/?ac2017rapds7y0c
It isn't scientifically sound research but just a quick attempt to learn more about these methods.
As the mic wires aren't screened, do you also get a lot of hum ?
Simply make a recording with them in the ear and one with them just lying around and have a listen.
the measurements were all done on the same rig with the 16kHz already filtered but no other compensations (added LF compensation later on and re-measured some headphones)
Best way to get a baseline is by using headphones that measure well and consistantly.
Examples are HD580/600 and HD650 as IMO they do not very that much unless pads are worn.
Most measurements out there (from calibrated rigs) do match closely even when they are using different headphones.
The average of these plots should give you an idea how to calibrate your rig using a similar (or a few other) 'known' headphones.
Corrections should have to be small and broadband (except the well known peak of the WM61)
If you need sharp or much compensation something else is wrong.
For my rig I only needed the 16kHz adressed (slightly different for L and R mic and even had to replace one to get a closer resemblance) and LF compensation.
The LF comp. was very similar to that of dummy heads.
With this compensation I feel the meaurement is reasonably but not absolutely correct.
It does match what I hear and is highly repeatable.
Couldn't get that repeatability with mics in my ear.
-
Thanks a lot! I'll have to refer to that as I continue to work on my methods. I'm not sure about a hum. That's something I'll have to test out.
Haven't even the HD600/650 gone through variations? :P Seems it's hard to find a HP that doesn't show some level of unit-to-unit variation. But, as mentioned, I'm glad I'm getting that Grado in soon. But I do agree that the HD600/650 seem to be one of the better headphones with which to help calibrate a measurement rig. I have an HD598...but it is modded...
So far, my compensation curve involves only mild and broad changes. Broad lift of the bass, broad reduction of upper treble, and a slight reduction of the area around 2-5KHz (shaped somewhat like a hump, basically). These are custom based on what I believe I hear subjectively vs raw measurements.
While my results thus far are not perfectly repeatable or consistent, I find it to be fair and, perhaps, even better than expected. No surprise, I have to take extra caution to make sure headphones that are picky about fit and seal measure consistently. It does look like the mic/tri-flange setup I use it a bit different than what you tried. As for how well my measurements match what I hear, it seems fairly close, but this is something I continue to evaluate.
-
They've (HD600/650) gone through slight mechanical changes. But I suspect minute differences in your rig would be the dominating factor.
-
I would gladly pay shipping both ways if someone has one I could borrow. The HD650 is one of those headphones I know I should have bought forever ago, but I am often dumb like that. I have to go through my move and settle some financial matters before I should be buying more stuff. Funny or just stupid as it is, I say this as I recently bought an Alpha Dog. It makes all the sense!...
-
Nice try hans! You ain't gettin' mine! :)p13
-
This is looking very good so far.
Here's an idea: Plots showing relative differences (A - B) caused by materials applied in front of drivers to attenuate treble. I know this could be of use to modders with sensitive ears for tuning HF peaks in specific spots. I would love to contribute myself but ATM I have no rig nor an interesting headphone collection to check for consistency with other measurements.
-
That's definitely something I would be interested in doing. If you have any particular headphones and/or materials in mind, I'm sure I can work something out.
As a small update, the new SPL meter I got seems to work much better. I was probably running my previous tests closer to 96dB at 1KHz, so now it should be much closer to 90dB. And now that my mic-in volume is adjusted for that, I should be able to get more accurate distortion readings.
-
Cool! I would expect the results to transfer somewhat unchanged between headphones. For starters we could have a figure for the infamous toilet paper mods, soft/hard single/multi layer creatology felts, foams... Variants blocking only the central portion of the driver grill to prevent over damping (BMD, foamie, electrical tape or micropore / transpore, ala 7506).
Personally I would benefit from seeing the effects of the creatology felts, micropore tape, and the BMD in both the T50RP and regular dynamics (CAL!?).
Thanks a lot!
-
The idea of a 'modding material effects database' is appealing.
In my experience, however, there isn't such a thing as a standard effect for a certain material.
At least I didn't find it to work this way.
I have several types and thickness's of felt, foam etc. but the results on headphone A differs from the same material used on headphone B.
It even differs from headphone A with pad B from headphone A with pad C.
I have toyed with a lot of headphones in different configurations and have not found an 'universal' effect a mod does.
Plenty of plots have been made over time and while you can see what effect something has on headphone A it isn't a guarantee the same mod will have a (very similar) effect on headphone B.
Another thing is repeatability of results, which also greatly varies from headphone to headphone.
When placing a material in front of a driver you also change the position of the headphone slightly.
This can only be done reliably when altering things on an electronic (signal) level.
-
Yeah, the treble response being so dependent on position and earpads is a real caveat, I can notice the peaks shifting in frequency as I shuffle the headphones around my ears. I was hoping that averaging many tries would diminish this effect somewhat. What about pad-less measurements then?, they would fail to show the effects of over-damping, I reckon, but maybe the treble response would in turn become more consistent.
-
Pads are just as important as the driver is and can make or break the sound.
They provide seal needed for the bass and also absorb mids/highs or reflect them.
I see no way to exclude pads and still get meaningful measurements.
A 'free air' measurement without pads will (most likely) not be telling something about the drivers properties.
Don't have any experience with this method though and is just my opinion.
-
So its my turn for a WIP too, need foam for mounting and to cut some woodboard to mount the artificial ear, very limited capability but hopefully I will be able to get some reliable results sometime with my approach. WM-61a mics in a silicone acupunture ear model(it provides some ear gain)with the mic mounted in heatshrink rubber to simulate a 2.5cm ear canal. Also did a single side measurement of the X10 with the mic approximately 13mm from the end of the rubber tip. Compensations will come in later when I'm able to mount my HD800s on the rig. Plan to get another mic and level meter as well for reference to cal the rig. Note: Source of mics are a ebay seller from japan so yeah, might be counterfeits. Mine are etched 61031 on the sides if it means anything.
Graph of the stock Yamaha HP-1 raw response.
-
Cool, keep us posted! Good idea on reusing the thread. These early results remind me of something you'd see from Tyll or Golden Ears, with the large 1-5KHz peak that is compensated for. I'm curious how that would look if you moved the mic closer and didn't try to simulate an ear canal.
Unless that headphone actually sounds like that?