Measurement Setups - Post Your Rig!

Discussion in 'Measurement Systems and Standards' started by Luckbad, Oct 5, 2015.

  1. johnjen

    johnjen Friend

    Friend
    Joined:
    Sep 29, 2015
    Likes Received:
    360
    Trophy Points:
    63
    Location:
    Well Grounded
    I didn't know this thread was here until Maxx134 pointed it out, thanks man.

    I recently obtained a 3Dio binaural mic setup that works rather well for dialing in the mods I'm applying to my 800's.
    Yeah it ain't cheap but it's WAY less expensive than just about anything else I could find.
    And I figured one of those $10 styrofoam heads would not make an ideal acoustic platform to take measurements from.

    I'm using the 3Dio mic and a Steinberg UR22 mkII mic preamp to feed Audacity, REW, and Cubase s/w.
    https://3diosound.com/
    https://www.steinberg.net/en/products/audio_interfaces/ur_series/models/ur22mkii.html

    I'd post a pic or 2 but they are on my Mac and not from a URL.
    Oh well.

    JJ
     

  2. Biodegraded

    Biodegraded Acquaintance

    Joined:
    May 28, 2017
    Likes Received:
    43
    Trophy Points:
    18
    Location:
    Calgary AB, Canada
    Here's an inexpensive one built around a Dayton iMM-6 mic (I'm new here so in case I've done the usual thing and screwed up embedding the images, there are direct links too):

    [​IMG] [​IMG] [​IMG]

    http://i.imgur.com/j7nSVoj.jpg http://i.imgur.com/kSasdwZ.jpg http://i.imgur.com/eCRDYdc.jpg

    Construction & calibration

    Components (all prices in $CAD) are:

    Dayton iMM-6, $33.06 (on sale)
    TRRS to TRS adaptor, this one overpriced at $19.99; whether necessary depends on your input jack
    USB audio adapter, $13.99; whether necessary depends on your audio setup
    Headphone extension cable, $8.99
    Upholstery foam from a wedge-shaped car-seat cushion, $9.99
    2 6" squares of EVA (Creatology) foam
    2 DVDs (1 under each EVA square)
    2 small pieces of 1/16" thick felt glued together (an 'ear', which after trial measurements I prefer not to use)
    Rainbow sponges (proudly made in Canada by Royal Sponge Mfg Co.), $1.50 for an 8-pack at the Dollar Store
    Elmer’s or similar spray-on glue.​

    Total about $CAD 75.

    Wedge shape and size is similar to my own head (see picture at left). A hole and slot are made in the foam to accommodate and support the mic and cable. The EVA & DVD are glued to the one side of each ‘half-head’ and 4 of the sponges are glued to the other side of one of the halves. The EVA ‘skins’ are marked with 1 cm gradations on the horizontal and vertical axes to aid consistent positioning. The hole in one of the skins is bigger than the mic, so the mic is supported entirely by the soft cushion foam; in the other, the mic is a push-fit (effects on measurements described below). The felt ‘ear’ is a push-fit onto the mic and is left free to rotate to the appropriate angle for left or right. Signal is supplied by a cheap USB DAC/amp (Topping NX2). Software is REW.

    Mic, sound ‘card’ and SPL all require calibration. The Dayton mic calibration file is loaded directly via the ‘Mic/meter’ tab of the REW preferences. A sound card calibration file can be made in the ‘Soundcard’ tab using a signal looped from the amp output to the mic-in. SPL calibration was performed using the Bosch iNVH smartphone app, which purportedly has internal calibrations for the internal mics of various phones including mine (Galaxy S4). How good this calibration is given phone variation I don’t know, I didn’t test it vs a proper SPL meter; but any error from this is probably much less severe than the error arising from how the measurement was done…

    [​IMG]

    http://i.imgur.com/x93A9DD.jpg

    Yeah, would’ve been better to connect the phone to the Dayton (after calibrating them against each other), but I don’t have the right cable. I note @atomicbob’s demonstration that an omnidirectional mic (same capsule, I think) that’s an imperfect fit into a proper calibrator unit gives a reading >3dB out. I wouldn’t be surprised if SPL I get from this technique is inaccurate by more than 5 dB.

    A 1k sine wave from REWs signal generator was fed to the phones and adjusted to 90 dB using the volume pot on the amp. To get there, sweep level (soundcard preferences) was set to -3dBFS, output volume 0.6, input volume 1.0 (adjust input volume down if necessary to prevent clipping with more sensitive headphones), and recording & playback device levels in the computer to 100%. Then, the signal generator (same 1k sine wave) was used to set REW’s internal SPL meter to 90 dB (activate the signal generator first, then choose ‘use an external signal’ under the SPL meter’s ‘calibrate’ button). It’s only necessary to perform this calibration step once; subsequently setting the same SPL for different headphones is done using the signal generator and adjusting the volume control on your output (amp) until the SPL meter display reads 90 or whatever reference you’ve chosen (hit the red ‘record’ button while playing the signal).

    Physical setup, measurements & comparisons

    The setup I prefer is using both ‘half-heads’ free-standing (first photo above). One half can also be used with a thick book or on a desk (photos below); I initially thought a reflective surface would boost the bass response but it doesn’t, at least not consistently (it seems to become more position-sensitive). The mic is mounted in the side with the big hole in the EVA. Using the side in which the EVA holds the mic tight boosts the bass but leads to ringing around 2 kHz. Presence or absence of the felt ‘ear’ changes the response in a more complicated way (see Conclusions).

    [​IMG] [​IMG]

    http://i.imgur.com/MKyIrZL.jpg http://i.imgur.com/kKKts1K.jpg

    I don’t have access to HD650s so it’s hard to compare my results to those from other rigs in this thread. I’ll borrow some other models that have been measured by @Marvey, @ultrabike and others, but for now I’ll compare AKG K702s to Changstar/SBAF results, and a couple of closed-back pairs with interesting FR profiles to compensated plots from Innerfidelity and HeadRoom.

    AKG K702: compare to ultrabike’s results here. Bass rolls off much more, probably due to less compression across my narrow head. The lumpiness in the lower midrange could be partly due to inaccuracies in my mic’s calibration file (solid black curve in the FR plot), it’s smoother when this is removed. Upper midrange is not too dissimilar, gentle climb up to about 2.5k, and then the plunge to 3.5 k is mirrored (although I think some of this in my measurements is an artifact; see later). There’s a contrast in the upper treble where I have a peak between 5.5k and 7.5 k, then a rapid drop whereas the ultrabike measurements show this peak starting and finishing higher. I think this difference, particularly the steep drop from ~7 to 10, is a problem in my measurements, it’s mirrored with these and other phones vs Innerfidelity measurements (see here and below). Also, I do perceive these as bright in that region whereas the ultrabike-measured pair were thought not too bad. There might be a systematic manufacturing difference too: mine are from the new factory in China whereas I presume the other pair (measured late 2013) were from the Austrian factory. So all in all, probably not the greatest comparison. CSD plots are broadly similar and I’ve included distortion for completeness despite the SPL uncertainty and the compensation necessary to correct the soundcard FR (almost 15 dB down at 20 Hz!).

    [​IMG] [​IMG] [​IMG]

    http://i.imgur.com/U7oILV5.jpg http://i.imgur.com/SotMznq.jpg http://i.imgur.com/HbkXicB.jpg

    Actually my measurements look closer to ultrabike's ones of the AKG K7XX, but still low from 8-10 kHz and beyond and in the bass as referred to above. @Marvey and @Hands also have measurements of the AKG7XX here for comparison. Above about 2 k mine seem pretty similar to Marv’s, but these are (supposedly) different models of headphone so this could just be coincidence.

    I’ve been playing with mods to these to tame the treble, and will post results and measurements in the dedicated thread.

    Sennheiser HD280Pro: (Innerfidelity measurements here). Boost in the mid-bass, drop in the upper bass, flat through the midrange before dropping into the lower treble then rising again. My measurements show somewhat lower bass overall, a bigger hole in the upper bass, a steeper dip to 3-3.5 k and then a gentle climb out to a lower mid-treble peak; apparently similar behaviour to my AKG measurements, too deep at 3.5 k, too high around 5 k and too low thereafter.

    [​IMG]

    http://i.imgur.com/Cx5Tthb.jpg

    Sennheiser Momentum On-Ear: (Innerfidelity measurements here). Again somewhat lower bass. The dip into the lower midrange is mirrored, but on the other side my measurement climbs about 5 dB higher to ~2 k. The drop to 4.5 k is mirrored – so steep and deep it’s hard to see if there’s much difference around 3.5 k – then mine climbs to a peak at ~5.5 k before dropping again whereas Tyll’s climbs more gently to a plateau at 7-10 k. I’ve included the distortion measurement below because it’s interesting; apparently most of the high bass distortion (mentioned in the Innerfidelity review) is 3rd harmonic, 5th also being high.

    [​IMG] [​IMG]

    http://i.imgur.com/YgzBZXO.jpg http://i.imgur.com/aGkxttI.jpg

    Conclusions and pros & cons

    Pitfalls in comparing different measurement systems notwithstanding, I think my measurement setup is systematically low around 3.5 k, high around 5.5 k, and low out to >10 k.

    The above comments are using the half-head with the larger hole in the EVA, i.e., with the mic supported by the soft upholstery foam. Using the half in which the EVA supports the mic gives the K702s up to ~5 extra dB at 7-12 kHz range (max at 10k) and again at a narrow peak at 15 kHz, but significantly lengthens the CSD ridge at 2k. I don’t perceive excessive shrillness/shoutiness in this region, so I use the soft-supported side for measurement. Using the EVA-supported side for the HD280s and MOEs gives ~5 dB more around 8-10k (good) but also 5-10 dB more in the lower midrange, >10 dB more in the bass (too much!) and no profound differences in CSD; again I feel better about using the soft-supported side for these closed-back phones. Using the felt ‘ear’ gave all 3 headphones the HD280s and MOEs a few dB less in the 7-12 kHz range and a few dB more in the bass. To me this is counterintuitive: I’d expect the treble to be enhanced by the closer mic (which was aligned flush with the coupler or felt in each case) and the bass to be ‘soaked up’ by the felt, and in the case of the over-ear Momentums to leak through the imperfect seal, but the opposite seems to be happening. Whatever’s going on, because of the apparently systematically low treble of this rig, the measurements I prefer and which are presented here are without the ‘ear’ and because of the huge bass boost it imparts to (my) closed-back phones, without the direct mic coupling to the EVA ‘skin’.

    Pros:

    + Cheap, compact and easy to build

    + No phantom power needed for the mic, or expensive USB interface (but see also Cons)

    + Ok response through midrange and lower treble(?)

    + Can use the same mic to build a simple IEM measurement system a la @La Cenric

    + Can provide a use for books you’ll never read

    + Will work either standing up or lying flat on a table; could thus be said to be more versatile than the C.U.N.T.s popular here​

    Cons:

    - Too low through mid- and into upper treble

    - Brutal ‘soundcard’ compensation means bass distortion characteristics may not be representative (soundcard compensation is not applied to the measurements by REW, and the curve for this setup means relative bass level is low enough that bass distortion will likely be less)

    - SPL calibration!

    - USB DAC/amp and mic connections are probably quite noisy – but I guess only enough to affect higher-order distortion…?

    - Probably plenty of others…​

    Comments, advice, ridicule all welcome!

    Edit: OK, ridicule then - I did screw up the thumbnails. Click the links for full size.
     
  3. Maxx134

    Maxx134 Friend

    Friend
    Joined:
    Oct 27, 2015
    Likes Received:
    435
    Trophy Points:
    63
    Location:
    NYC
    I am following your measurement protocol in your google doc link.
    I also made note of members "ultrabike", "serious", "Sorrodje" and "Marvey" settings.

    I have both the UMIK1 with calibration file into REW, and a GOV2a for similarly.
    For the coupler I have both a dummy head with partially formed ears,
    and also a SBAF type coupler will post pics soon.
    :)
     
    Last edited: Aug 11, 2017
    Serious and philipmorgan like this.
  4. Maxx134

    Maxx134 Friend

    Friend
    Joined:
    Oct 27, 2015
    Likes Received:
    435
    Trophy Points:
    63
    Location:
    NYC
    Check out my new coupler, inspired by the famous SBAF TP mods...

    Introducing The infamous TP coupler!:
    20170811_110146.jpg
    Actually its a larger size paper roll,
    Not an actual TP size lol.
    Paper surface and felt around mike different sizes I will test to get close to the SBAF standards of purity...
    20170813_055805.jpg
    :)

    So squeezeably soft and portable,
    You can take anywhere!
    20170811_105857.jpg
    Even sleep with it as a pillow..hmm. no..
    Made with 100% recycled paper!
    20170810_153944.jpg
    It is not the normal paper-towels like "bounty" or others that are too soft.

    It is the firmer paper-towels you get out of bathrooms to dry your hands...

    Its Wrapped in electrical tape, and use foam inserts to hold mike in middle..
    20170811_104843.jpg
    :punk:
    So yes great ideas can come to you when in "sittn on the can" ..lol
     
    Last edited: Aug 13, 2017
    Biodegraded, ultrabike and cskippy like this.
  5. Maxx134

    Maxx134 Friend

    Friend
    Joined:
    Oct 27, 2015
    Likes Received:
    435
    Trophy Points:
    63
    Location:
    NYC
    I am having Problem comparing the different CSD settings everyone here posted...
    Some settings are good at seeing more detail but no good at comparing to other CSD plots.

    It would be nice if we all agreed on a general CSD settings.
    I like the more straight forward settings on Marvey CSD graphs.



    Edit:
    I have noticed the creatology foam alters the upper spectrum , so I abandoned it..


    Edit2:
    I have abandoned the felt and the foam,
    as both of them altered the upper range trebles response on my setup.
    They didn't play well with my paper based surface.

    I currently am it using as is,
    which is basically a circular rolled paper surface,
    With Center foam holding mic flush.

    The foam is eva foam/velour wich Seems to work best for me.
    Look at my coupler surface,
    Not hard or reflective but firm yet soft paper-towel(Marcal) surface,
    with a foam centered mic set flush.

    I left sticking out a bit to show the materials.
    20170813_054517.jpg
    It seems to do the job best for me currently.
    Edit** I have narrowed material to a very low density foam around mic instead.
    Now finally the treble range is pretty accurate.


    Also,
    My coupler looks like a descendat or varation of Marvey's famous "C.U.N.T." coupler design shape,
    DSC_01268d0a.jpg
    http://www.changstar.com/www.changstar.com/index.php/topic,2305.0.html

    Which I find most optimal for ease of headphone centering and positioning.

    Also flat coupler seems the best way to go for uniformity of measurements and more reliable positioning.
    I hope it was ok to post that pic here.
    :)
     
    Last edited: Aug 13, 2017
  6. Maxx134

    Maxx134 Friend

    Friend
    Joined:
    Oct 27, 2015
    Likes Received:
    435
    Trophy Points:
    63
    Location:
    NYC
    Man, I been staring at wiggly lines all weekend lol
     
    cskippy likes this.

Share This Page