[Overnight Sensations MT] My first attempt at measuring speakers (Help wanted!)

Discussion in 'Speakers' started by thegunner100, Sep 26, 2015.

  1. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    Earlier today, I received my miniDSP Umik-1 calibrated by Cross-Spectrum. I opened it up, made a platform out of my chair, and plopped the mic onto the platform in-between the two speakers at my listening level. I downloaded REW and read some basic tutorials from miniDSP and REW on how to set up REW properly.

    [​IMG]

    I'm guessing that measurements are generally done at 90 degrees but in this case, is it okay to do 0 degrees? I have the calibration files for 0, 45, and 90 degrees for my umik-1.

    Anyways... my measurements turned out looking like this.

    [​IMG]

    I just left it zoomed all the way out because I don't really know what's appropiate for the Y axis. Obviously for the X axis, only 20hz-20khz really matters. I only have OJneg's initial measurements and the designer's measurements to really go by, so I don't know how far off my measurements are. And of course, there's room acoustics to be taken account of as well.

    I'll probably measure the speakers again from a couple feet back, where I would be listening at a equilateral triangle position with the speakers. I also have a pair of Audioengine A2s and Polk Audio Monitor monitor 40 series ii that I could measure sometime to see how they are.

    So... some questions:

    1. How should I do the separate left and right channel measurements? Should I leave the mic where I had it and just disconnect one of the speakers? Or should I place the mic right in front of a single speaker (and disconnect the other of course).

    2. Preferred measurements at 0, 45, or 90 degrees? I don't have a mic stand or tripod (my camera one doesnt fit), so I'm not sure if I can pull off 45 or 90.

    2. I know that there isn't a standard for speaker measurement, but what should I ideally set the scale on the X and Y axis to?

    3. What volume should I be measuring at? Looking at OJ's and Ultrabike's measurements, I see that they measured a lower volume than I did.

    4. Is there much that can be improved upon with EQ? I'm not really sure how to even go about doing it with windows or foobar, since the EQ values from REW can't just be plugged into a PEQ like EasyQ on foobar.

    I'm hoping that some experienced guys (OJ, UB, Marv) can chime in with some suggestions. But I'm open to anything!

    Edit: please move to another sub-forum if this isn't the appropriate one.
     
    Last edited: Sep 26, 2015
  2. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    No one does measurements at 90 degrees.

    You can approach this two ways.
    1. Take a listening position measurement, which is what it appears you have done. You do this head-on. Sometimes you might need to take another measurement at 10 or 15 degrees if there are are comb and strange polar patterns from the two drivers and crossovers. If the setup is symmetrical, I run both speakers at the same time because I am lazy. Since you have that one wall, you might want to take separate L and R measurements.*
    2. Take a pseudo anechoic measurement. Take one speaker to a super large room where the boundaries are far apart or take it outside. Measure the speaker with the microphone at about 1 meter away with the microphone vertically level with the tweeter. You can then rotate the speaker to get 15 and 30 degree off-axis measurements.
    If you know where your room boundaries are, you should be able to see wall reflections in the impulse response. What you can do is gate the FFT window so the end of it is before the second reflection. Of course the issue with this is that the frequency response will only be accurate up to the longest frequency as determined by the length of that window. Bass sounds waves tend to be very long, on the order of tens of feet. You really can't do jack about the bass and lower mids anyways since much of that is dictated by the room. Gating the FFT window will give you a lot of insight about the driver / speaker itself.

    *EQ'ing frequency response irregularities caused by a wall is a bit tricky. The FFT spectrums are really an average of energy over time which may include some early reflections. Our ears can distinguish between the original signal and reflections. If I have both sets of measurements above, I'll start with #2 to EQ, and then use #1 to fudge things. For example, if the wall causes a -6db anomaly, I might only compensate +2 or +3 db since I know it caused by a reflection instead of the driver itself. EQ'ing is a bit of a art.

    Just keep playing around.
     
    Last edited: Sep 26, 2015
  3. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    I got a tripod from Guitar Center for really cheap ($20). Get the most entry level and that should suffice. It's pretty handy.

    I also measure at 1 meter using a large room or the largest room I have, and at 0 degrees.

    I definitively like what I'm seeing from your FR plot BTW. The speakers seem to have extension down to 40 Hz with a nice warm tilt! The depression around 2 and 3 kHz may be due to the cross over or a reflection perhaps from the table. 2kHz wavelength is about 0.5 ft I think. May want to measure using slightly different positions to get the best out of your speakers.
     
    Last edited: Sep 26, 2015
  4. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    Thanks for the quick response guys!

    Gated at 4.0ms, it looks like this:

    osmt gated.jpg

    I went for 4.0ms after inputting my measurements onto this calculator. It looks like the pre-reflections response has smoothed out the bass region tremendously but makes the dips in the treble region more apparent. There's also a wider dip at 675hz as well. The drivers crossover at 4khz, so I think the 2-3khz dip may be caused by a reflection from my table, and then filled in after more reflections in the area. Please correct me if I'm wrong in assuming this.

    I'll try measuring at different positions in my room next time, and seeing how those turn out. I'll try measuring a single speaker at 1 meter away in my living room as well, even though it's not a very large room. Unfortunately living in NYC, my backyard isn't that big either. Marv when you say 10 and 15 degrees, which way do you mean to turn the mic?

    My virtue one.3 is small enough to bring around for measurements, but It'll be a pain to carry the Yggdrasil around. Does using a different dac matter for speaker measurements? I was thinking that it may be easier just using the GO 450 when I go to other rooms.

    I think I had some subjective measurements of these speakers from CS 1.0. I'll look for those sometime later, and update my impressions with the Yggdrasil and actual speaker stands in the equation.
     
    Last edited: Sep 27, 2015
  5. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    1. Might be the table since 4ms would be too long too to cut the table reflection out. The early table reflection, isolated from the wall and ceiling boundaries might be emphasizing the dips. You are probably right in your assumption. Keep in mind the bass measurement below certain frequencies should be considered totally inaccurate once you gate the FFT window. Some software will highlight that region or cut off the bass measurement to denote so.
    2. As far as the off-axis measurements, rotate the speaker, not the microphone.
    3. Just use GO450. Errors from speakers are several magnitudes greater than errors from amps or DACs.
     
  6. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    New sets of measurements! I'm using a different scale and different volume for these, so they can't be compared to the previous ones. I forgot to mention this, but my speakers are about 35" away from each other (going by middle of one woofer to the other)

    OSMT Avg.png
    Average response from 7 different listening positions: 28" Center, 38" C, 38" L, 38" R, 56" C, 56" L, 56" R.

    I believe this is probably the most accurate response that I can get from my room. The peak at 130hz exists in every listening position but it gives the OSMT a very nice bass punch. In addition, the bass extension is fairly good for such small monitors, extending to 49hz before sharply dropping off in my room.The midrange region is quite flat until it starts to dip around 1.5khz. I believe the dip was intended to reduce any possible listening fatigue. While the dip seems drastic in measurements, I don't hear it that way. These speakers aren't as dark as the graphs may lead you to believe. Ignoring the dips in the treble region (reflections from the table and keyboard tray), the treble is relatively flat as well. I don't notice any emphasis in the treble region while listening. I believe that dips in the FR are much harder to hear than peaks.

    When I first asked OJneg about these speakers, he told me that they were quite similar to the hd6x0s. In a way, the measurements between the OSMT and the senns aren't too far off. Subjectively speaking, these speakers are fairly neutral with a bit of a boost in the bass and slightly reduced treble emphasis like the modded hd650s. More on subjective impressions and quasi anechoic measurements another time.

    So it's been a very educational day of playing around with measurements, interpreting them, and relating them to some of my subjective experiences! Any thoughts or comments?

    OSMT Left Distortion.png
    Left FR & Distortion (28" away listening position)
    OSMT Right Distortion.png
    Right FR & Distortion (28" away listening position)
    OSMT L+R.png
    LR overlay and combined (individually from L+R)

    Slight differences between L and R channels can be seen here.

    OSMT 28.png
    Listening position 28"

    This is usually where I'm sitting in front of my computer. Overall pretty good! Though mids and treble could be a bit forward sounding due to the close proximity to the drivers. Subjectively, this position has the least bass in comparison to the others.

    OSMT 38.png
    Listening position 38"

    Not a huge difference in distance, but it helps to even out the mids and treble relative to the bass. Soundstaging and imaging is improved.

    OSMT 56.png
    Listening position 56"

    Probably my favorite listening position. Best bass extension, soundstaging, and imaging.

    OSMT 3 listening position overlay.png
    28", 38", 56" overlay and offset.

    Looking at the overlay between the three positions, we can see that the 56" has the least drastic dips in the treble and a less drastic peaks in the bass. The region between 49-74hz is flatter and may be contributing to the bass extension that I'm hearing at that position. Plus, there is less of an emphasis at 130hz.

    Far KB vs NKB.png
    56" away KB vs no KB
    Middle KB vs NKB.png
    38" away, not as drastic

    The dips and peaks in the treble region are partially caused by early reflections on my keyboard tray, and can be seen in the differences in measurements between the keyboard tray out, and the tray tucked inside.
     
  7. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    Just wanted to reference OJ's measurements his own pair of Overnight Sensations, which were done in various areas. There are also some helpful posts about measuring speakers as well.

    http://www.changstar.com/index.php/topic,833.msg27988.html#msg27988

    OJ's measurements do seem to also show some treble shelving, but not quite as drastic as my measurements do.

    And also, measured using the Real Trap 38% option 1 positioning from noaudiophile. My small room is littered with furniture and what not, so the measurements aren't much better at all compared to the avg of my normal listening position. More prominent dips on the treble, but the bass is more filled around 80-100hz.

    RT v Near.png

    Or for a more condensed view:
    RT v Near 2.png
     
    Last edited: Sep 28, 2015
  8. OJneg

    OJneg The Most Insufferable

    Pyrate BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    3,923
    Trophy Points:
    113
    Location:
    Grand Rapids, MI
    20dB shelf is kinda crazy. It doesn't sound like that I hope.
     
  9. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    Yeah, it doesn't sound like that. I'm not sure what could be making it measure like that. Maybe I need to raise the mic up a bit, cause the mic is more between the woofer and tweeter than it is in front of the tweeter.

    I'll have to get myself a mic stand one of these days too.
     
  10. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
  11. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    Hmm... I tried the 3 positions again with the mic a few inches higher, at tweeter level. They kind of just shifted the reflections in the treble region, most evident in the measurements taken at 38".

    Higher Mic.png

    I'll just have to wait for a day when my family is gone, to take measurements in my living room instead. Definitely something weird going on in my room, but it certainly doesn't sound that way.
     
  12. OJneg

    OJneg The Most Insufferable

    Pyrate BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    3,923
    Trophy Points:
    113
    Location:
    Grand Rapids, MI
    I'd hate to think the L-pad came loose at some point. Did you post response 1m away from tweeter head-on?
     
  13. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    I didn't even know what a L-Pad is and had to look it up. Would that be responsible for causing the dips like that?

    Mic leveled with tweeter
    Tweeter level.jpg

    In position, 1m away. There's a ward robe to the left of the speaker (seen in the first pic) and a desk to the right. Best I could do in this situation.
    In position.jpg

    Each speaker measured individually, 1m away and tweeter head-on. Everything before 3khz is just about matched.
    LR 1m Nearfield in room.png
     
  14. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    New measurements, done properly. The reason why I was getting the weird huge dips in the treble was because I was running 8 sweeps for 1 measurement. I changed it to 1 sweep and now everything looks proper.

    WP_20151004_15_24_38_Pro 2.jpg
    1m away, mic on-axis and tweeter level

    LR individual.png
    No more weird treble dips!!!

    Excellent driver matching, except for the tweeter. The tweeter response is somewhat off. I'm not sure if that's cause if my mic was off a bit or if the tweeter responses were just different. The right tweeter has been louder with this set of measurements and in the previous ones from my last post. Maybe the L-pad might have been moved a bit during transport? I'll have to check this out more next time.

    There's a weird dip around 120hz though. Not sure what's causing that, but it's probably benign. These measurements are a much more accurate representation of the Overnight Sensations than my measurements from before. Overall pretty neutral sounding with a "slight droop" in the mids, which was intentional tuned to reduce any listening fatigue. The woofer extends to 60hz before room interactions, just like in the specs.

    I have more measurements to show, but I'll save that for a future post when I have more time. The measured response from the living room looks much better than in my bedroom, with the new proper measurements.

    Off-axis Response.jpg
    Excellent off-axis response as well! Only slight differences in the treble and around 400hz.

    LR gated.png
    4ms gated response
     
  15. OJneg

    OJneg The Most Insufferable

    Pyrate BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    3,923
    Trophy Points:
    113
    Location:
    Grand Rapids, MI
    120Hz dip is probably floor bounce. Measure reflection distance from woofer to floor to mic and find wavelength.

    I don't like that tweeter imbalance. Almost 5dB. Did you ever feel it was audible? Try some panned pink noise and see if the timbre changes as the signal sweeps from left to right. Would also be curious if you could measure the tweeter a few inches away to be sure.

    Other than that, those measurements look pretty much identical to the paper I originally sent you :)
     
  16. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    There were times where I've felt channel imbalance with vocals but I always attributed it to my speakers being at the corner of my room. I'll try some panned pink noise and measuring up close soon, and post my findings.

    And yep! Pretty close! Save for the treble differences.

    Edit: My boom mic stand is coming in Friday.
     

    Attached Files:

    Last edited: Oct 8, 2015
  17. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    Took a really really quick measurement with my mic stand, about 1" away from the tweeter. Can't be sure if it's too accurate, but I'm getting about a 2-3dB from 5khz onwards.
     

    Attached Files:

  18. ulikmadik

    ulikmadik New

    Joined:
    Oct 5, 2015
    Likes Received:
    5
    Trophy Points:
    3
  19. thegunner100

    thegunner100 Hentai Master Chief

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    2,461
    Trophy Points:
    113
    Location:
    NYC
    Sorry that I haven't posted in a while! I did confirm a few weeks back that the drivers themselves did not have any channel imbalance with the help of the boom mic and more consistent positioning. It's hard to measure the OSMT sometimes because the tweeters are off-center.

    The top two are taken about 1" away from the tweeter and the second two are from 1m away.
    Channel balance.png

    Yep! I gave it a try and it works pretty well with REW. However... there's really not much to EQ with the OSMT like OJ suggested. I tried some EQ based off of REW's filters and it just sounded weird. So, no EQ for me with my speakers.
     
  20. OJneg

    OJneg The Most Insufferable

    Pyrate BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    3,923
    Trophy Points:
    113
    Location:
    Grand Rapids, MI
    Glad they're still matched. That cheap Dayton tweeter has no right to sound/measure as good as it does. $15 tweeter has smoother and more natural treble than 98% of all headphones.
     

Share This Page