On distortion measurements - an experiment

Discussion in 'Measurement Techniques Discussion' started by Serious, Apr 30, 2016.

  1. Dr. S

    Dr. S Rando

    Joined:
    Aug 28, 2019
    Likes Received:
    9
    Dislikes Received:
    0
    Trophy Points:
    3
    Location:
    Germany
    The rule of thumb estimates a THD increase by three times every 10 dB.

    However, looking at some "real world" examples there is always a certain baseline where the THD bottoms out:
    https://audioxpress.com/article/fre...ks-m50-omnidirectional-measurement-microphone
    http://www.isemcon.com/datasheets/iSEMcon EMX-7150.pdf
    I'm not sure if this is an actual limitation of the microphone or just the measurement environment, speakers and signal chain behind the microphone.

    Speaking about the signal chain, the preamp is probably another severe source of distortions.
    http://www.sound-link.co.uk/pdf/AES 145 - Improving Audio Performance of Microphones Using a Novel Approach rev23.pdf
    This is an example for an insufficient impedance bridging leading to higher THD with phantom power. You can clearly see that the 6.8 kOhm standard for balanced XLR inputs cannot compete with a 2 MOhm implementation when connecting a 1 kOhm source. Now imagine something like a EM172 which comes with 1.5 kOhm output or a PUI 5024 HD with even 2,2 kOhm. Plus the components of the circuit!
    If I compare that to the input impedance of only 2 kOhm (balanced) or 1 kOhm (unbalanced) of my Babyface Pro, I will not only end up with a serious signal attenuation (up to 10 dB in the worst case!) but also degradations of the frequency response* and THD performance. No idea why RME choose such low resistors... :confused:

    See also:
    https://mynewmicrophone.com/microphone-impedance/
    http://www.sengpielaudio.com/calculator-bridging.htm

    This is why I should probably activate the pad feature to increase the input impedance to 5.2 kOhm (balanced) or 2.6 kOhm (unbalanced). It will lower the levels by 11 dB however. Not so good for low sensitivity measurement microphones.

    At the moment I consider to buy a modified Line Audio OM1 or a iSEMcon EMM-13D082 for my 60318-4 couplers. They do come with only 100 Ohms of impedance and quite solid THD specs for the price.

    *(since the mic impedance varies with frequency)

    Regards
    Dr. S
     
    Last edited: Apr 19, 2020
    Serious likes this.
  2. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,871
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    near Munich, Germany
    For the 2nd measurement it seems to me they are hitting noise limits. For the Earthworks M50 I'd hope it was the speaker, but at 2mm from the cone I'm not sure. Maybe that Markaudio driver isn't too linear at lower levels, not sure. I'd expect unweighted noise with a 20kHz bandwidth to be higher than 30dB, maybe higher. With my WM-61A the noise difference between A-weighted and unweighted (44.1kHz SR) is about 10dB. Maybe the bandwidth was higher than 20kHz, so noise would be even higher. The way it stays constant when approaching 100dB also seems wrong to me.
    Either way, 0.15% at 100dB seems pretty bad to me. Would be really weird if @purr1n and @stratocaster could pull off less than -70dB or 0.03% at an estimated 100dB in their respective HD800 measurements with relatively cheap measurement microphones when a (more or less) proper measurement mic couldn't.

    Looking back at my measurement I linked in my last post, I probably gave the UMIK-1 too much shit. Looks like I was able to get results of around 0.05% distortion from both my UMIK-1 and my cherry-picked WM-61A back then. After an accident I had to resolder my UMIK-1 capsule. Thankfully I had lots of data to compare the FR before and after and adjust the sensitivity accordingly (as well as calibrate it against the ND-9 that atomicbob measured). But I would guess that the increased distortion (and decreased sensitivity) came from soldering the capsule without cooling it.

    I will have to do more reading about mic circuits. Also thanks for reminding me about the PUI 5024HD. I had considered one before and I'm honestly not sure why I never bought one. The other 6mm and 8mm PUI capsules look interesting aswell, especially the 2703 with better SNR that also seems similar to the Primo EM258. I'm pretty concerned about SNR overall since I know how noisy recordings even with the Primo EM172 sound. Many sounds I can hear get obscured by noise even when played back at realistic volume levels. In a dummy head with its 15dB or so gain in the upper mids it may be less of a problem, but lower noise is always better.
    BTW, I'm looking forward to seeing where you end up. I had considered making a dummy head in the past, but had mostly forgotten about it. I was pretty satisfied with how my measurements looked like with the microphone in an eartip at my ear canal entrance, but a dummy head has other uses aswell.
     
  3. Dr. S

    Dr. S Rando

    Joined:
    Aug 28, 2019
    Likes Received:
    9
    Dislikes Received:
    0
    Trophy Points:
    3
    Location:
    Germany
    Sadly, there is no official measurement to verify. Earthworks themselve don't publish THD graphs and I couldn't find any other reports examining such characteristics, yet. I wouldn't expect such an expesive piece of equipment to perform that mediocre (at least on lower levels) either tbh.

    Soldering capsules is indeed a delicate matter. That's what finally persuaded me to purchase a temperature controlled soldering station by ERSA. :D

    Do you have a link for the circuit by solderdude?
    Personally I'm preferring the Simple P48 circuit by David McGriffy. Its fairly simple to build (as the name implies), comes with an impedance balance and good noise performance. THD is still not that advanced tho. That's where the three-wire or RCA circuits come into play which add roughly 20 dB of headroom. Noise remains somewhere around 15 dB. Unfortunately those are not so easy to build since you have to physically mod the capsule and access the gate contact...

    The PUI 5024 HD is considered to be a low cost alternative for the EM172. It has very similar specs and practical performance.
    Afaik the 80 dB SNR of the 5024 HD is the best PUI has to offer for small omni capsules. What is the exact model you are refering to with the 2703?

    My project is currently leaning to a dummy-less build because all the mannequin heads I got my hands on the last two years had kind of odd shapes and sizes, leading to questionable sealings and loose fits. The base will be two 60318-4 couplers with some average sized silicone ears attached. In case this attracts any interest, I could make a thread with some further details.

    Regards
    Dr. S
     
  4. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,871
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Yup, sadly 14dBA specified noise seems to be the lowest we can do with small electret capsules. I was hoping larger 14 and 25mm electret capsules would have less noise, but I haven't found one with lower noise, yet. For a dummy head or flat plate coupler using a larger microphone isn't a problem in the slightest. For my in-ear method 6mm is about the largest I can do since they need to fit in eartips, which also have to fit my ears.
    https://www.digikey.de/product-detail/en/pui-audio-inc/POM-2730L-HD-R/668-1598-ND/7898330
    That's the 6mm one I was talking about. I think I'll order some of the PUI capsules to try out.

    This is what I was referring to: https://diyah.boards.net/thread/361/?page=2
    Just phantom power. The MOTU M4 that I now use thankfully seems to have 1MOhm input impedance. I'm also going to try other circuits.

    I do have an (admittedly crappy) temperature controlled soldering station, but I think it's more a matter of heatsinking and not heating the capsule for longer than necessary, letting it cool down after soldering each connection, etc. I've soldered many WM-61As in the past when I tried matching them and only rarely did they drop in sensitivity after soldering them many times even without heatsinking. Still, it's definitely better to be safe than sorry. I still have some leftover metal that I'll drill holes into to use as a heatsink going forward.

    Yeah, I think it would be helpful for others who want to DIY a dummy head (like me :p).
     
  5. Dr. S

    Dr. S Rando

    Joined:
    Aug 28, 2019
    Likes Received:
    9
    Dislikes Received:
    0
    Trophy Points:
    3
    Location:
    Germany
    I see. You mentioned it as "2703". That's why I couldn't find it. :D
    The 2730 is indeed very similar to the Primo EM258. I have no experience with larger badges of the PUI capsules. But so far the 5024 HDs I ordered were performing neck to neck with the EM172. For some builds I actually prefer the PUIs because they have a buzz killer cap inside which works better with ground-sensitive circuits. For impedance-sensitive setups (where the output impedance should remain as low as possible) the Primos are probably better.

    Thanks for the link, will give it a look.

    Going with an high-z input you will of course get none of the nasty impedance issues I mentioned before. That's only the trade-off with XLR inputs. The M2 has about 3 kOhm afaik. The M4 will most likely be the same.

    Haven't compared so many capsules before and after resoldering. But I heard that some are really sensitive to heat. That's why I follow the recommendation of PUI and stick the capsules into a large chunk of metal (actually an old aluminum heatsink) while soldering. I recently killed one PUI by accidentally applying too much pulling force on the wires, tho. The soldering pads came completely loose from the body. No chance to repair, unfortunately.

    Now, back to THD...
    I think that the Line Audio OM1 with only 0.5 % at 133 dB is pretty much the best I can get for less than 400 EUR. A used Earthworks might be worth a try. But they are hard to get for a good price.

    The crux of the matter is that the OM1 has a tapered nozzle which does not fit the 60318-4 coupler without further modification. Mr. Jönsson from Line Audio kindly offered me to customize the housing. Nevertheless, a modified case will also alter the FR and make it harder to calibrate...

    To be continued.
    (probably in another thread ;))

    Regards
    Dr. S
     

Share This Page