Using a microphone calibrator

Discussion in 'Measurement Setups, Systems, and Standards' started by atomicbob, Dec 4, 2016.

  1. atomicbob

    atomicbob dScope Yoda

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    6,928
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    On planet
    Obtaining accurate results using a microphone calibrator requires a controlled interface between the measurement microphone tip and calibrator cavity. Professional and laboratory equipment have precise specifications for measurement microphone dimensions ensuring this interface is maintained properly. Less expensive mics do not adhere to the dimensional precision.

    Examples
    inches - outside diameter of microphone tip:
    0.520 - ACO Pacific 7046
    0.525 - Josephson C550H
    0.422 - Dayton emm6
    0.236 - typical 6 mm omni capsule from digikey

    Without a proper fit into the calibrator cavity resulting sound pressure experienced by the microphone may easily be 3 dB or more in error. From the examples above Dayton's emm6 is 0.10 inch small and a typical 6 mm mic capsule is going to need an adapter.

    ACO Pacific 7046 properly inserted into ND9 calibration cavity: IMG_1589a.jpg

    Data is accurate as expected.
    20161127-1650 ND9 N785787  as left Batt V 9_760 94dBSPL=-16dBFS 7046-28439.png


    Dayton emm6 has a loose fit to ND9 calibration cavity:
    IMG_1591a.jpg

    Measurement is 97.12 dB when it should be 93.98:
    20161127-1745 ND9 N785787 94dBSPL=-16dBFS emm6-2372 not sealed.png

    A shim will be improvised to fit the Dayton emm6 properly into the ND9. This strip of gaffer tape is approximately 6 to 7 cm in length. A post-it note could be used as an alternative with only a minor degradation in result obtained. Be very careful to keep adhesives from contaminating the microphone calibrator cavity.
    IMG_1593a.jpg IMG_1595a.jpg IMG_1597a.jpg IMG_1598a.jpg

    Note the snug fit now achieved between emm6 and ND9
    IMG_1599a.jpg

    emm6 properly sealed into ND9 cavity. A result that is within 0.02 dB of the expected measurement.
    20161127-1750 ND9 N785787 94dBSPL=-16dBFS emm6-2372 sealed properly.png
     
    Priidik, Azteca, Serious and 6 others like this.
  2. atomicbob

    atomicbob dScope Yoda

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    6,928
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    On planet
    Calibrating a typical 6 mm microphone capsule DIY measurement mic.

    Simply inserting a 6mm mic capsule into a microphone calibrator cavity will be in error similar to the improperly fitting emm6 without a shim.

    6 mm mic capsule not positioned correctly in ND9 calibration cavity:
    IMG_1600a.jpg

    Resulting calibration is 2.79 dB high (should be 93.98 dB)
    20161128-1100 ND9 N785787 94dBSPL=-16dBFS 6mm mic not sealed.png

    Similar to improvising a shim for the emm6 an adapter is improvised from two pieces of clear PVC tubing, one inside the other.
    0.50 OD and 0.375 ID PVC tubing
    0.375 OD and 0.250 ID PVC tubing
    IMG_1601a.jpg

    Now the mic capsule is positioned flush to the surface of the adapter which will be inserted into the microphone calibrator cavity:
    IMG_1604a.jpg

    6 mm mic capsule and adapter inserted into ND9 calibrator cavity:
    IMG_1605a.jpg

    Resulting calibration data is 94.02 when 93.98 was expected. Within measurement uncertainty. This is a much better result.
    20161128-1100 ND9 N785787 94dBSPL=-16dBFS 6mm mic sealed.png
     
    Last edited: Dec 6, 2016
    johnjen, spoony, Serious and 4 others like this.
  3. atomicbob

    atomicbob dScope Yoda

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    6,928
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    On planet
    part three in progress - content may change again over the next several days.

    Gain Staging for acoustic measurements
    The sensitivity, noise floor and SPL at 1% THD limit of the measurement microphone used determines the dynamic range over which measurements may be accurately obtained. Setting the mic preamp gain appropriately will aid optimization for tasks of interest. Standardizing on levels allows calibrated recordings to be made which can be shared and compared among multiple investigators. I have found two levels which have served well in acoustic labs for several decades using inexpensive measurement microphones:

    A 94 dB SPL = -16 dBFS (typical range 40 to 100 dB SPL with 110 dB SPL peak)
    B 94 dB SPL = -26 dBFS (typical range 50 to 110 dB SPL with 120 dB SPL peak)

    During gain staging it will be helpful to have a digital level meter monitoring the measurement mic audio input channel. An example is the Digital Level Meter from Darkwood Audio:
    http://www.darkwooddesigns.co.uk/pc2/meters.html#Digital

    If you have a Digital Audio Workstation then you probably already have suitable digital level meters.

    With the mic calibrator set to 94 dB SPL adjust the mic preamp until the desired standardized level is obtained. Note the gain knob position for future reference. Changing this gain setting will un-calibrate the system.
    Interface Input Gain Staging.png


    Now calibrate the measurement system following the directions provided with the system being calibrated. REW and ARTA will be used as an examples.

    REW SPL reading calibration.png
    Select "Use an eternal signal" and enter 94.0 for the level.


    REW multi-display.png
    A display such as this may be seen with REW for verification. Please note the default 0 dBFS reference setting for REW uses a square wave Crest Factor rather than a sine wave Crest Factor as reference. The two references differ by 3 dB. As such REW will read -19 dBFS instead of -16 dBFS by default. A configuration setting may be adjusted to change to the sine reference if desired. Also note that other waveforms have different Crest Factors.


    ARTA multi-display.png
    This is the display that may be seen with ARTA after calibration for verification.

    Determining microphone sensitivity
    Please be careful connecting, disconnecting and using loopback for P48 microphones. Always turn Phantom Power off before making any connection changes. DO NOT power on P48 when running loopback connections. P48 should remain OFF for loopback.

    This step assumes a DAW such as Audacity and general knowledge of how to use it.
    Generate 60 seconds of 440 Hz sine at -10 dBFS.
    Turn the audio interface hardware output volume control all the way down.
    Connect an appropriate loopback cable from channel one out to channel one in (assuming channel one in was the channel used for measurement mic calibration).
    Monitor the input channel digital level.
    Play the 440 Hz -10 dBFS sine in a loop.
    Adjust the hardware output volume knob until the calibration level appears on the input digital level meter (eg: -16 dBFS)
    Now attach a multimeter set for mVrms to the channel one output - for a balanced connection this will be across pin two (hot) and pin three. Read and document the value observed. It should be in the range of 11 to 28 mVrms with a nominal of 17.8 mVrms for a wm61. For a Dayton emm-6 it will be 10 mV nom with a range of 7 to 14 mVrms. This is the calibrated sensitivity for your specific microphone and may be used to check or restore the system calibration if the mic preamp gain knob is ever changed for other use.

    Using a signal generator to maintain calibration
    If you use your audio I/O device for other purposes than acoustic measurements then recalibrating the system is reasonably easy once the specific measurement microphone sensitivity is known and documented as performed in the previous step.

    Using the same 60 second 440 Hz -10 dBFS calibration sinusoid from the previous section and a multimeter connected to the channel one output, adjust the audio interface hardware gain knob until the meter reads the same value in mVrms as the mic sensitivity determined and documented in the previous section.
    Now, with P48 off, connect the channel one output to channel one input (loopback.) Monitoring channel one input with a digital level meter, adjust the preamp gain knob until the standardized level used previously for calibration is achieved. eg: -16 dBFS. The system calibration has now been restored.

    Measurement Microphone Calibration Curve
    So far all calibration activity has been oriented to achieving calibrated sensitivity at 1000 Hz. Specific compensation for microphone Frequency Response is beyond the scope of this particular post. It requires ability to measure the particular measurement microphone FR, document and enter the appropriate compensation into the acoustic measurement system.

    However it should be noted that many measurement mics utilize 6 mm mic capsules that are substantially similar in their respective FR. Looking at the data sheet for Panasonic wm61 replacement from PUI one can see the FR charts are nearly alike, all with an approximate 5 or 6 dB peak at 10 KHz. Over the decades I have obtained very similar results in examining measurement microphone performance. It is a fair bet that using a generalized compensation will achieve a reasonable improvement, though not as accurate as a calibration for the specific measurement microphone. Here are both the table of data and FR for such a generalized calibration compensation. Use them at your own risk.

    panasonic wm61 mic calibration example.png

    panasonic wm61 mic calibration example - data table.png
     
    Last edited: Jun 2, 2017
  4. sorrodje

    sorrodje Carla Bruni's other lover - Friend

    Staff Member Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    2,698
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Dijon / Burgundy / France / EU
    Great. I'll redo my calibration once the calibrator will be back home :)
     
  5. spwath

    spwath Collegiate hijinks master

    Friend
    Joined:
    Dec 13, 2015
    Likes Received:
    2,778
    Dislikes Received:
    18
    Trophy Points:
    113
    Location:
    Hartford, CT
    kind of a nooby question, how do you actually make the calibration file?
    I use rew.
     
  6. atomicbob

    atomicbob dScope Yoda

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    6,928
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    On planet
    reserved for part 4, coming soon.
     
  7. JohnM

    JohnM Author of REW - Rando

    Joined:
    Dec 13, 2015
    Likes Received:
    86
    Dislikes Received:
    0
    Trophy Points:
    13
    The reading of rms level for a full scale sine wave can be changed in the REW View preferences (interface section). By default REW would show -3 dBFS as the rms level for a full scale sine wave, it would show 0 dBFS for a full scale square wave. I happen to think that's right and proper, but the AES disagree. AES17-1998 assigns 0 dBFS to a full scale sine wave. By that definition a full scale square wave has an rms level of +3 dBFS. The notion that a digital signal can have an rms level that exceeds full scale seems an abomination to me, but no doubt some folk at the AES thought long and hard before inflicting that mathematical travesty on the world so the option is there for those that have been drawn to the dark side, or simply want readings that tie up with their other audio test equipment. :)
     
    Priidik, atomicbob and Serious like this.
  8. atomicbob

    atomicbob dScope Yoda

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    6,928
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    On planet
    Hi @JohnM , I didn't word that very well. I'll change it in the near future. I only meant to avoid having novices become concerned about the 3 dB difference, not imply any sort of deficiency. In fact I use REW to demonstrate using ETC to find reflections in rooms. Gave a presentation to local AES and several audio groups with REW as the featured method for acquiring and displaying the curves. David Johnston fought the same battle for Cool Edit Pro about which crest factor, square or sine, should be the reference nearly two decades ago and ended up with the same solution; have a user configuration allowing choice of reference.
     
    JohnM likes this.

Share This Page