Burst Response! HD800, SR-207, HD650

Discussion in 'Headphone Measurements' started by purr1n, Jan 8, 2017.

  1. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    35,553
    Dislikes Received:
    54
    Trophy Points:
    113
    Location:
    Calabasas, CA
    Inspired from @Serious's work here: http://www.superbestaudiofriends.or...sr-202-srm-212-measurements.3608/#post-107440

    I'm glad he did it. I had been thinking of doing something similar for a long time now, but he got me off my ass. This is all super preliminary, but interesting. The following are actual 10kHz bursts, 100 cycles, for each headphone measured on the coupler. These burst response "measurements" differ from impulse responses in that they are actually the signals recorded from the microphone. Impulse responses from mathematically derived and recreated from noise or sweeps. Magni 2U and Modi AKM 4496 stack was used. Recording was into UMM-6 microphone and ADC into PC. STAX 2070 system amp was used for the SR-207.

    10kHz Burst WAV 100 Cycles. HD800, SR-207, HD650
    10k burst.png
     
    Last edited: Jan 8, 2017
  2. SSL

    SSL Friend

    Friend
    Joined:
    Nov 12, 2015
    Likes Received:
    1,014
    Dislikes Received:
    6
    Trophy Points:
    93
    HD 650 is "slow". Mm hm.
     
    Besnia, westermac, Ash1412 and 6 others like this.
  3. Kattefjaes

    Kattefjaes Mostly Harmless

    Friend
    Joined:
    Sep 5, 2016
    Likes Received:
    4,774
    Dislikes Received:
    41
    Trophy Points:
    113
    Location:
    London, UK
    Can you make predictions about how these will look at different frequencies from looking at the CSDs too, or is that too simplistic?
     
  4. KurtSvensson

    KurtSvensson Friend

    Friend
    Joined:
    Sep 30, 2015
    Likes Received:
    191
    Dislikes Received:
    0
    Trophy Points:
    43
    Can we get similar measurements at different frequencies? Very interesting how well the 650's are doing here. I've always considered speed to be the biggest downfall of the headphones, maybe that doesn't apply to the treble.
     
  5. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    35,553
    Dislikes Received:
    54
    Trophy Points:
    113
    Location:
    Calabasas, CA
    Too hard to say for anything for certain at this time, as more work needs to be done to validate the measurement methodology. My results are slightly different from @Serious; on the other hand, there are a lot similarities between our burst response results. As an aside I privately held on to my FR plots for over a year before I felt comfortable releasing them, but I wanted to throw this out to the community for ideas, advice, etc.

    The slowness of the HD650 probably has some to do with its highish mid-bass distortion. Looking at these, I wonder if the faster transient response of the STAXen can be attributed to the greater overshoot followed by a sudden decrease in energy. It almost seems after an initial shock, the electrostatic diaphragm goes into absorb mode before shooting off a slightly smaller load extended in time before stabilizing. Also, could the more gradual energy release of the STAXen be responsible for their smoother less grainy sound?

    That's a good question, but I doubt it. CSDs are actually averages of energy derived from the impulse responses, which are in turn derived from sweeps or noise. There is a lot of time domain information actually lost in CSDs.

    Yes. I will try 100Hz.
     
    ultrabike likes this.
  6. Serious

    Serious Inquisitive Frequency Response Plot

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    1,158
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Yours look pretty different from mine, so maybe there was some amp interaction for my SR-202 measurement. Is this on the creatology foam coupler? For high frequencies I'd just use the regular "open" V1 type coupler.

    In a general sense, yes, but these plots should give you a more precise look in the time domain, but limited to one frequency (and the distortion and IR products). You'd probably be better off doing it the other way around.

    BTW: I was inspired by these plots from way back in the day, before there were even CSDs. I forgot where I saw them.
     
    Last edited: Jan 8, 2017
  7. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    35,553
    Dislikes Received:
    54
    Trophy Points:
    113
    Location:
    Calabasas, CA
    I used the foam coupler. Results tend to be ringier, especially for planars on the open. Planars seem to require the internal pressure from going crazy.
     
    Serious likes this.
  8. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    35,553
    Dislikes Received:
    54
    Trophy Points:
    113
    Location:
    Calabasas, CA
    Give me 15 minutes for the 100Hz results.
     
  9. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    35,553
    Dislikes Received:
    54
    Trophy Points:
    113
    Location:
    Calabasas, CA
    OMG. The 100Hz results! I need to build some suspense. During the CS days, I would say stuff like this and then take a long nap before posting the results.
     
    Bagged Milk, lm4der, OJneg and 3 others like this.
  10. Serious

    Serious Inquisitive Frequency Response Plot

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    1,158
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Can you see if you can confirm my weird 15kHz results for the SR-207? I always wanted a measurement that could show us the "chaotic" movements of the diaphragm for stats (with multiple arrivals) and thought that this was a confirmation. The CSDs seemed to have a far too long timescale to show the effect, I think.
    I'll also measure the HD600 after I check that I can still get consistent results.

    I didn't think of measuring super low frequencies, but from 1-5kHz the results looked almost perfect in my case. At 100Hz we might see the effect of the port tuning, maybe we can even see the worse HD650 bass quality :p
     
  11. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    35,553
    Dislikes Received:
    54
    Trophy Points:
    113
    Location:
    Calabasas, CA
    We won't see the worse bass quality of the HD650 because distortion is very difficult to make out in waveforms, unless it is super high, or goes on forever (successive odd orders, which will tend to square up the wave). It's the just the nature of the beast. We need the right measurements to show the right things. I wish the Hydrogen Audio crowd would understand this. They are like level 3 Jedi younglings (but with bad attitude) thinking they know everything.
     
    lm4der likes this.
  12. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    35,553
    Dislikes Received:
    54
    Trophy Points:
    113
    Location:
    Calabasas, CA
    Here we go. The results are interesting. Could this be why some people say STAXen have "limp dick" bass? Too early to say. We need more data.

    100Hz Burst WAV 10 Cycles. HD800, SR-207, HD650
    burst 100Hz.png
     
  13. Serious

    Serious Inquisitive Frequency Response Plot

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    1,158
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    It's interesting because I found the SR202 to really sound like that. Bass extension was good, but it had absolutely no impact. The very definition of limp-dick. Because of the low frequency the timescale is even longer now. The effect might even be audible in the recording of it. Also bass information in music does tend to have a lot of transient information.

    I meant the resonance from the port tuning/driver resonance. If you can see distortion on a sine-wave with the naked eye, then something must be very wrong indeed.

    I thought that maybe from the actual shape of the waveform (once it has settled) we could maybe see some why different magnets sound different. I think Besnia mentioned Barkhausen noise. The effect would probably be so small that even with an FFT it would be far below the noise floor, though. I also wonder if there are hints to the different nature of estats compared to dynamics in this. The diaphragm would have to stretch for very high excursions and it would have to wrinkle for it to be perfectly centered between the stators, as it's slightly larger than a perfect plane.

    EDIT: Is it okay if I post my own HD600 results here, too? Will take a few minutes...
     
    Last edited: Jan 8, 2017
  14. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    35,553
    Dislikes Received:
    54
    Trophy Points:
    113
    Location:
    Calabasas, CA
    Post away.
     
  15. Serious

    Serious Inquisitive Frequency Response Plot

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    1,158
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Here are my results for the 10kHz signal that I used in my post, for the HD600 and the HD800 (modded). I also tried to see if the HD600 rear foam had an effect on the results, because I find it to sound snappier without the rear foam, by placing a thicker piece of felt behind the magnet vent. This felt eliminates the mid-bass-hump completely and also manages reduce the 4kHz resonance.

    10kHz, 100 samples at 44.1kHz sample rate, recorded at 192kHz sample rate:
    10kHz (2).jpg
    What's interesting is that the HD600 looks less uniform. That was consistent over 5 tries. Maybe there's a correlation to subjective grain? In terms of rise-time the HD600 might even be slightly better, but it also has more overshoot. Maybe this has something to do with the treble (and ultrasonic response), with the HD600 being peakier and the HD800 extending higher.

    100Hz, 0.1s, recorded at 192kHz SR:
    100Hz.jpg
    Personally I don't think there's much you can see here. The last bit looks weird for the HD800 plot, but I don't think this was consistent across multiple runs. Should look similar to the HD600 (felt) plot.
     
    Last edited: Jan 8, 2017
    Drakkard, Priidik, kn19h7 and 2 others like this.
  16. Armaegis

    Armaegis Friend

    Friend Contributor
    Joined:
    Sep 27, 2015
    Likes Received:
    2,900
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Winnipeg
    As I had mentioned in the other threads, I suspect this is mostly amp interactions and power supply wonkiness. With the dynamic drivers, I bet if you had a way to quantify the over/undershoot/wobble/whatever, you would be able to draw a pretty good correlation with the headphone impedance plot. I imagine the usual culprits like the Senn HE650, Beyer DT880, AKG K7## would show similarities. Actually, the HD598 might be a better test since the midbass impedance swing is much more extreme than the HD650. It would be also interesting to see how shoving a resistor in series with the output would affect the burst response.

    With the Stax, I'm not completely certain if changing amps would do it but again I feel like this is a power supply capacitance voodoo hoodoo.

    If nothing else, this might not be a tool for headphone measurements per se, but might be useful for amp manufacturers to improve amp responses. Maybe. Or I'm just piping outta my piehole and will have to eat crow after this.
     
    Serious likes this.
  17. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    35,553
    Dislikes Received:
    54
    Trophy Points:
    113
    Location:
    Calabasas, CA
    SR-007 results here. Much better.

    100Hz Burst. SR-007 and HD800
    Much better performance from the SR-207, as it should be. The first crest of the 100Hz burst from the SR-007 is a bit lower than the HD800. SR-007 decay is maybe less in amplitude, but still lengthier.
    100 burst SR-007.png

    10kHz Burst. SR-007 and HD800.
    A lot of overshoot from the SR-007. The dip after the overshoot isn't as dramatic as on the SR-207. Decay is better than SR-207, much faster and less in amplitude, but still not quite as good as HD800.
    15k burst SR-007.png
     
    Last edited: Jan 8, 2017
    dark_energy, Vorlon, kn19h7 and 5 others like this.
  18. Ash1412

    Ash1412 Friend

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    283
    Dislikes Received:
    2
    Trophy Points:
    63
    Can you add a planar magnetic into the mix? Would be interesting to see 100hz burst from a planar magnetic just to see if planar magnetics' "magic bass" can be measured. Can also give us clues as to whether these measurements are amp related or not.
     
  19. Armaegis

    Armaegis Friend

    Friend Contributor
    Joined:
    Sep 27, 2015
    Likes Received:
    2,900
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Winnipeg
    The 10k burst on the Stax show an amplitude ripple... can we quantify the size and frequency of the ripple?

    Ditto for the dynamics? though It's harder to tell from the pictures.

    edit: also curious how the ripple parameters scale with output.
     
  20. Armaegis

    Armaegis Friend

    Friend Contributor
    Joined:
    Sep 27, 2015
    Likes Received:
    2,900
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Winnipeg
    It would also be very interesting to run a multitone and see if the warbles are summative.
     

Share This Page