Ultrasonic Recording

I want to make a DIY ultrasonic bat recorder using the Knowles FG 23329 microphone. I have been using commercial “bat detectors” to record ultrasonic bat and insect calls. I have almost zero knowledge of what is involved in my project, but I would like to use my computer to record at 384k sample rate, using Audacity and MATLAB, and to use a TASCAM DR 100 MKIII to record at 192k sample rate.

I think what I need to do is find a suitable housing for the microphone, attach a suitable interface (ie XLR for Tascam and Microphone Jack for the computer), and provide phantom power for the mic? Is this sufficient, or is there more involved here in processing the signal before feeding it to the computer or digital recorder? Is the phantom power from the Tascam suitable for this?

Knowles FG 23329

The [u]specs[/u] I found say it goes to 10kHz (100Hz to 10kHz). It also runs from about 1.5V (1.3 to 3V). Standard phantom power for studio condenser mics is 48V. Electret condenser “computer microphones” run from 5V supplied from the soundcard in series with the signal (not phantom). Usually, these things are called “microphone elements” since they need power and are not ready to plug-in and use by a consumer. (It might actually work directly into a sound card with 5V, but you might also kill it and your soundcard/laptop probably doesn’t go into the ultrasonic range.)

48V can be regulated-down or you could use a battery. Do you know any electronics?

but I would like to use my computer to record at 384k sample rate, using Audacity and MATLAB, and to use a TASCAM DR 100 MKIII to record at 192k sample rate.

Upsampling doesn’t gain anything and upsampling in real-time means more processing and more data and a higher chance of glitches/dropouts.

A sample rate of 192kHz guarantees that you can’t record signals above 96kHz, but since these are audio devices there is no guarantee that they can record much above the audio range of about 20kHz? Have you confirmed the frequency range of the TASCAM?

I trust TASCAM if they say it works at sample rates up to 192kHz but you can’t always trust what the manufacturers say because the drivers can re-sample whenever necessary and you never know what the hardware & drivers are doing. The manufacturer can claim recording capability up to 384kHz even though the actual analog-to-digital conversion is happening at 44.1 or 48 kHz…

There’s some relevant information about “Ultrasonic Applications for Knowles Electret and MEMS Microphones” here:

I will not actually be “upsampling”, I will be recording at the data rate that I will use for analyzing the signal. The Knowles FG series microphones are used in ultrasonic applications and have a published frequency response up to 100k and even higher, so I am not worried about the frequency response/frequency range of the microphone. https://www.digikey.com/en/pdf/k/knowles/ultrasonic-electret-mems-microphones

I do not have any electronics experience, so I will need to get commercially available equipment to power the mic. I’m sure there is something out there, but I have no idea what at this point. I am trying to minimize my learning electronics and concentrate on signal analysis, but at the same time I don’t want to spend $600 or $1000 on a microphone. Any ideas on how I can get a decent inexpensive interface that will power the mic and pass the audio through to the computer or recorder?

You may be able to get help here: https://forum.pjrc.com/threads/38988-Bat-detector/

I will not actually be “upsampling”, I will be recording at the data rate that I will use for analyzing the signal.

If the hardware is running at 192kHz but you are “recording at 384kHz”, that’s upsampling. :wink: If you want to upsample it’s better to do it after recording because then speed/timing are not a factor. You can change the Audacity Project Rate before exporting or MATLAB can probably do it.

I do not have any electronics experience, so I will need to get commercially available equipment to power the mic.

I’d recommend a battery unless battery life is a problem. A battery makes a good low-noise power source and it’s cheap (if you don’t have to replace it too often). There a few electret “stage” microphones that run from a battery.

Your datasheet shows shows 10V operation with some added resistors and a capacitor. That should also work with a standard 9V battery but I’d try a 1.5V battery without that other circuitry first.

The interface has a balanced connection but your mic has an unbalanced output. The mic ground has go to one of the signal-inputs on the XLR connector. You might want to look-up how to adapt balanced & unbalanced connections.

You might need some additional amplification beyond whatever gain the interface has. The microphone is less sensitivity at high frequencies and it’s probably at a greater distance than a typical audio studio-recording situation. But I don’t know… Maybe the acoustic signals are stronger than I think…

You’re designing a microphone.

Knowles FG 23329 microphone

I don’t know that there is a pre-baked interface for the three bare wires and ground braid this microphone has.

Screen Shot 2021-04-05 at 9.59.47 AM.png
You don’t plug it into anything. You’re expected to solder those wires to a circuit board or electronics module, keeping in mind that the signal is super quiet at that point. The electronics inside the FG are there to shove the sound signal down the cable, they don’t amplify anything.

I don’t doubt the FG will work at very high frequencies. Most microphones will run outside of their published specifications. But they may not work reliably with good stability and quality. Some microphones push their distortion and noise up there secure in the knowledge that nobody would ever try to record ultrasonic signals.

There is a down-to-earth practical consideration with microphone design. How will you know when you get it working? Do you have a calibrated test bat?


I am trying to design a microphone, true.

I am not looking for a pre-baked interface for the existing 3 wires + ground, although that would be ideal. I am hoping that the line-in is sufficient, w/o a pre-amp, for recording and for computer analysis. That is one part of my question. If I put an XLR connector or Headphone jack directly on the microphone audio line, will this be sufficient input for the computer sound card (monitoring and recording in Audacity and/or MATLAB) or the TASCAM recorder? I am not listening to the output, obviously, because it is ultrasound, but will put it through MATLAB as digital data vectors for analysis.

I am not concerned with whether this microphone will work as an ultrasound microphone, I already have the answer to that, from the manufacturer and from several commercial ultrasound mics that use the Knowles FG electret microphone, so just not an issue. I have a means for testing/calibration, I can worry about that after I have a microphone. But even without testing, I can tell from the data. I do this already with commercially available ultrasound mics from Wildlife Acoustics and Pettersson Elektronik.

Thank you for your input.

I guess I should qualify my lack of electronics knowledge. I have a Masters in Electrical Engineering, with a concentration in Digital Communications and Digital Signal Processing. I have very little hardware background/experience, but I have a pretty good understanding of the signal processing end of things. My computer sound card is capable of running at 384kHz (according to the manufacturer), so the recording will not be upsampled. The Tascam samples at 192kHz, and the recording from this will also not be upsampled, it will be analyzed at the 192k rate.

The 1.5V battery solution sounds promising. Thanks. I think I can figure out how to do that.

The “unbalanced out” to “balanced in” is something I will definitely have to research. Thank you for pointing that out.

As for amplification, maybe it is best just to try it first and see. Then if/when it doesn’t work I will go back to the drawing board.

Thank you for your input. Very helpful!


Interesting thread. Thanks for the link.

I am hoping that the line-in is sufficient, w/o a pre-amp, for recording and for computer analysis. That is one part of my question. If I put an XLR connector or Headphone jack directly on the microphone audio line, will this be sufficient…

Microphones require a preamp and your TASCAM recorder has a built-in preamp.

Microphone signals are in the ballpark of 10mV depending on the loudness of the sound and the sensitivity of the microphone. Of course it varies a LOT so interfaces & recorders always have a recording-volume control. If you have an idea of the SPL level you can estimate the voltage from the microphone specs. But I assume SPL is unknown? 1Pa = 94dB SPL and 0dBV is 1 Volt. …If you can handle MATLAB I assume you can calculate dB.

Line level signals (RCA outputs from a CD/DVD player or TV) are about 1V, again depending on “loudness”, etc. These can usually go directly into an analog-to-digital without a preamp. Headphone signals are in the same ballpark and compatible with line-level signals. The difference is headphone outputs are capable of driving lower impedance loads.

Well, bat calls can exceed 140dB SPL at 0.1 meter, but the bats I’m recording are free-flying, they are facing random directions at varying distances, and rarely less than a few meters. And ultrasound intensity drops off much faster than audible frequencies, the higher the frequency the more rapid the attenuation as a general rule. ie about 0.7dB/meter at 30kHz up to 3dB/meter at 100kHz. I’m typically looking at about 60kHz calls, but some go as high as 120kHz or so.

I’m not sure I understand how to calculate the voltage from the spec. The sensitivity says “-53dB ±3dB @ 74dB SPL”. So, I think the sensitivity is at 74dB rather than the standard 94dB? And this is presumably at 1kHz, there is a drop in sensitivity at ultrasonic freqs, according to Knowles. Anyway, I think there are too many unknown variables here for the calcs to be useful.

So, I guess the answer here is that it may work without a preamp, but I might want a preamp to adjust the level for more distant targets, for targets not facing my direction, etc. I sent an e-mail to Knowles about going without a pre-amp, hopefully I’ll get a response. At any rate I plan on trying it and seeing how it goes.

Thanks again for the input!

Perhaps this will help: https://www.analog.com/en/analog-dialogue/articles/understanding-microphone-sensitivity.html#

There’s another fuzzy rule here. The noise in a system is determined by the first electronics that the microphone signal hits. It will never get better than that without extraordinary gymnastics.

A line-in connection might have a noise floor of about -60dB with a show volume of about -6dB. That’s perfectly ordinary. If you apply your microphone signal of -53dB to that, your bats and the Line-In noise level are about the same volume.

So somewhere in the SHSHSHSHSHSHSH spring rain in the trees sound is your bat chat.

And that’s if everything goes well. Did we cover how you’re going to supply battery to the amplifier in the microphone? There is a tiny transistor up there and it needs juice from somewhere.

The XLR connection is not magical, although it seems that way. Look closely at the connector. Pin 2 has the performance electrically right-side up. Pin 3 has the identical performance upside down. Pin 1 is the protection braid/shield wound around everything.

The XLR receiver/preamp only looks for a difference between 2 and 3. Hum, interference, buzz, and other electrical trash appear the same on both 2 and 3 and within reason are completely ignored. That gives you an XLR rock band microphone playing into the audience sound mixer 125 feet away with perfect fidelity and no noise. I’m not making that up.

It is possible to connect pin 3 and 1 together and send a signal down pin 2. That’s a simple unbalanced to balanced converter. You give up much of the buzz and trash management and long distance cables, but that does work.

The question is how you’re going to do that without a soldering iron and the skills to use it. That’s the same skillset needed to connect your microphone to anything.


DVDdoug suggested using a 1.5 volt battery to power. I think that is worth a try.

I expect that to be the case, that the call is below the noise floor, but I think I need to try first to see.

Why is anyone still soldering in the 21st century? Oh, that’s a rhetorical question, no need to reply. I expect I can crimp with Neutrik connectors, at least I hope I can.

Because microphones use such low level signals, I would recommend soldering crimp connectors rather than relying on the crimping. Crimp + solder is much more reliable than just crimp (and very easy to solder a crimped connector).

More commonly Neutrik plugs have “bucket” connectors (an indentation in the end of each pin). These are not fun because you need a lot of heat so that the solder melts in the bucket, but not too much or the plastic holding the pin will melt.

An easy to use alternative to soldering Neutrik plugs is to get leads that already have the plugs, cut the leads, and solder the leads instead of soldering the plugs. While experimenting with hardware designs, you can “twist and tape” the wires instead of soldering, but don’t rely on that for long term use.

Sounds good. I will probably ask one of the techs at work to do the final soldering for me. That’s what they do for a living.
As soon as I figure out what to use as a housing I think I’m ready to start experimenting with designs.
Thanks for your help!

Why is anyone still soldering in the 21st century?

Why is someone trying to build a microphone from parts in the 21st Century.

I think I left my skycar double parked.

Screen Shot 2021-04-06 at 10.52.51 AM.png


Exactly, why are we still soldering and not driving sky cars in the 21st Century. And you’d think we would have built everything by now and not have to build anymore. I mean, what, 6000 years of civilization and we haven’t built everything yet?

But just look at the mark-up on these when put in a housing with a cable and connector. Like $50 in materials is going for $600 - $1500. And I am learning quite a bit in the process.

Anyway, Knowles got back to me. Here is what they said:

“You can use a mic level input to a computer sound card. If you want to power the mic from the sound card, add 2.2k resistor from output to negative. Then wire to 3.5mm stereo audio plug, positive = tip, negative = sleeve.”