How to use electret condenser microphones with MOTU M2 and REW

[Mr. Haelscheir]

I recently acquired a MOTU M2 to facilitate sample-synced averaged headphone measurements with my in-ear microphones (see https://www.farnell.com/datasheets/1653468.pdf and https://www.head-fi.org/threads/mez...eadphone-official-thread.959445/post-17743502 (post #5,152)).

I was originally using the Axagon ADA-17 to capture the signal from the two unbalanced microphones (3.5 mm TRS), my separately sending the sine sweeps to my FiiO K9 Pro ESS. When resting my hand on a grounded surface, the noise floor seen on REW's input monitor can get as low as -70 dBFS, the microphone being sensitive to the faintest breaths, reaching 0 dBFS for a moderate to hard blow headphone playback on the order of 105 dBA. This was when setting the input volume control to 0.25 with the Java driver, which was the default which I was previously unaware of. See below for what the signal from a few gentle to hard blows on the microphone look like:

After changing the volume control to 1.00, the noise floor with my hand on a grounded surface rises to -45 dBFS, and the same quiet breaths occupy a larger span of the dBFS scale. That volume control setting does have me concerned for the validity of my distortion measurements and EQ judgments in https://www.head-fi.org/threads/mez...eadphone-official-thread.959445/post-17743506 (post #5,153).

For connection to the MOTU M2, I used a female 3.5 mm TRS to dual 1/4" TS adapter, particularly https://www.amazon.ca/Gold-Plated-Audiowave-Female-Splitter-Shelled/dp/B07PRV735V/, my plugging the TS ends into the respective mic/line/guitar XLR/TRS combo jacks. I am noticing here that on max preamp gain (this appears to scale linearly with the noise floor plotted on REW's input monitor) and the input volume control set to 1.00 with the Java driver, while holding the microphone in one hand, placing my other on top of the MOTU M2's chassis drops the noise floor from -40 dBFS to -60 dBFS, whereby when blowing as hard as possible on the microphone such that the MOTU M2's input meter is maxed (it so happens to not be detecting clipping), this only brings the signal up to -10 dBFS. See below for an example of the limited dynamic range and the preceding drop in noise floor from contacting a grounded surface:

I am trying to understand how the MOTU M2 could seem so insensitive to these microphones compared to the consumer-grade Axagon ADA-17. The electret condenser microphones per the linked documentation have an output impedance of 2.2 kΩ, which shouldn't be an issue as the MOTU M2 for TS and TRS inputs for both line and instrument inputs automatically sets its input impedance to 1 MΩ. I was wondering if the female 3.5 mm to dual 1/4" adapter might be defective (3.5 mm socket possibly not gripping enough), but I figured that what could be plausible is that these two devices' ADCs have different reference voltages, if that is even possible. As such, the Axagon ADA-17 for the same input voltage range is reporting a much larger range in the dBFS scale than is the MOTU M2.

Thus, is there any way to adjust the MOTU M2's reference voltage or configure the input or order to attain a sensitivity comparable to the Axagon ADA-17 for these microphones? Else, if dynamic resolution is already comparable, is there a way for REW to "zoom into" the signal more than the default volume control (or when using the MOTU M2 ASIO driver where there doesn't seem to be any volume control) while having measurements at least accurately represent the relative SPL relations for distortion measurements and EQ? If I cannot obtain a comparable dynamic range useful for headphone distortion measurements, then this MOTU M2 as of yet would be of no use to me, whereby I will have to stick to singular 4M-length measurements taken with the Axagon ADA-17.

 

[staticV3]

Electret microphones usually require 3-5V of Phantom power to power the built-in FETs for proper levels.

I'm guessing that you're currently using the mics without phantom power, which may explain the low signal levels.

The Motu M2 does not support 3-5V of Phantom power. Afaik, you can only choose between 48V of Phantom Power and none at all.

That's where adapters like the RODE VXLR+ and VXLR Pro come in as they convert 48V to Plug-in power for use with electret mics.

As for the volume control thing, try using REW's Java EXCL input method.
That uses the Wasapi Exclusive API, which bypasses the Windows audio pipeline including it's volume slider and instead takes the audio directly from your Motu.

 

[Mr. Haelscheir]

I see. So the ADA-17 is probably using that 3-5V "consumer plug-in-power" (https://en.wikipedia.org/wiki/Phantom_power considers "phantom power" in this case to be an improper term). Would the "operating voltage" specification within https://www.farnell.com/datasheets/1653468.pdf be indicative of the phantom power it is expecting? I also realize from https://www.reddit.com/r/audio/comments/10fgulg that the TRS inputs probably weren't even going through the XLR mic preamps.

Anyways, thanks for the adapter suggestion. Costly, but versatile.

 

[staticV3]

Would the "operating voltage" specification within https://www.farnell.com/datasheets/1653468.pdf be indicative of the phantom power it is expecting?

Yes. Seems like anything between 2V and 10V plug-in power will be fine.

 

[Mr. Haelscheir]

I am happy to report that the RØDE VXLR Pro did the trick. I am now able to get averaged measurements with a pretty low noise floor and substantially more consistent phase response measurements.

Figure 1: female 3.5 mm TRS to dual male mono 1/4" TS adapter (I couldn't find a long enough equivalent with 3.5 mm terminations) with additional adapters feeding them into the VXLR Pros, the MOTU M2 then outputting the test signals into a dual male 1/4" TRS to male 4.4 mm Pentaconn adapter plugged into the FiiO K9 Pro ESS's line in.

I had chosen the VXLR Pro over the VXLR+ since its balanced output and likely excellent noise performance (probably useless when directly plugged in like thus) seemed attractive (and the amazon.ca price was more favourable at the time), but I had forgotten from the Topping EHA-5 measurements that transformers tend to struggle with lower frequencies. As such, I am concerned that compared to my original measurements taken with the Axagon ADA-17, the bass now has elevated third-order harmonic (orange trace) levels, and a different second-order harmonic (red trace) profile in the midrange and on:

Figure 2: Meze Elite left driver in-ear distortion measurement through MOTU M2 ADC with harmonic distortion plotted at their harmonic frequencies. 4M measurement length averaged across 8 repetitions (the distortion profile is generally the same when without averaging). 100 dBA measured with an SPL meter with foam wind shield positioning the mic capsule around 2 cm from the driver's inner grille.

Figure 3: Meze Elite left driver in-ear distortion measurement through Axagon ADA-17 ADC. 4M measurement length without averaging. 100 dBA. Note that this was from a different measurement session than the one where I had the time to use averaging, but the distortion profile through the MOTU M2 from this session was the same as in the previous session.

I tried mixing and matching use of the MOTU M2 and Axagon ADA-17 ADCs with driving the FiiO K9 Pro ESS's THX 788+ with its internal DAC versus the line in from the MOTU M2, whereby the only changes in distortion profiles came from the ADCs. As for the MOTU M2, the distortion profile for the same headphone playback volume including the elevated third-order harmonics stayed largely the same, only receiving vertical translations and some truncation from the noise floor between having the preamp set to its lowest gain versus the highest gain before clipping, so I doubt that this distortion was introduced by the MOTU M2's preamp or ADC, leaving the VXLR Pro with its transformer as the main culprit. These third-order harmonics do lower in level with lower headphone playback levels, its being a question of whether or not these are real headphone distortions that the Axagon ADA-17 somehow failed to pick up, or distortions of the VXLR Pro or MOTU M2.

I loathe having to order the VXLR+ to see if this remedies the issue and then figure out what best to do with the VXLR Pros.

 

[Mr. Haelscheir]

Good news. The VXLR+ as can be seen below did indeed completely do away with the odd-order harmonic distortion issue introduced by the transformer the VXLR Pro uses to convert the unbalanced input to a balanced signal.

With VXLR Pro:

With VXLR+:

The second-order harmonic distortion profile is fairly similar between both adapters at least above 200 Hz. I will take the VXLR+ as yielding the more accurate distortion figures for the transducer, but interestingly, if not expectedly, the VXLR Pro which converts the input to be balanced does in fact have upwards of 10-dB lower noise floor around 900 Hz compared to the VXLR+, so it could still have a use for taking measurements there, though third-order harmonic measurements won't be useful until past around 1 kHz, and higher-order harmonics past 600 Hz.

Here is the result with the VXLR+ after averaging eight 4M-length measurements:

The VXLR Pro as seen in my previous post would have cut around 10 dB off of the noise floor around 900 Hz, but again, with inaccurate odd-order harmonic distortion below that.

So far, most of the headphones I've measured have their distortion measurements being limited by the noise floor when playing the sine sweep at 85 dB SPL which I would consider moderate to loud for orchestral music.