What to know ehen choosing/searching a Ultrasonic microphone

[Detzi]

Good Day,
i am in search for a pcb mounted ultrasonic microphone with an analog out and a bandwidth up to 100kHz, so i had a look around.

ST manufactures not really something usefull, the best match given by their support fails any of my requirements but has an analog out...

Knowles seems to offer quite a lot MEMS Microphones where one of them the SPU0410LR5H-QB seems to tick almost all boxes. It has at least specified a free field response up to 80kHz more than most others.

TDK also has a good match, the ICS-41352 sadly it has a PDM and no analog out and again it ranges only up to 80kHz.

When searching for alternatives it feels quite time consuming because there are no filters like "frequency range" or "ultrasonic" on manufacturers sites. Mouser and digi-key do have such filters but only Mouser lists a device (a single one) after selecting just the frequency range.

Has one of you guys a good way to search for these things, or can you even recommend a specific one?
Has someone experience with these MEMS-Devices and can give me some tips about their system integration?
Are there advantages/disadvantages compared to other technologies, like electret microphones?

Thanks in advance!

[coppercone2]

I am suspicious that you can take a typical tiny microphone, like one of those pinhole types you have in a phone thats SMD mount, and then mill out the hole to greatly increase frequency response. I noticed the ultrasonics in the same form factor have like, 10x the area at the inlet (big square) and its usually meshed, vs a tiny hole for regular mics.



I think the brand I have is MEMS

[SiliconWizard]

The mounting will probably be critical to get reasonable freq response at those high frequencies. I'd guess a top port would be easier to deal with than a bottom port, but what about the enclosure it's going to be in?

I remember we already had a thread about something similar (would have to find it back). There were very few MEMS microphones for frequencies that high.

PDM is not that hard to deal with, do not let that deter you. You can basically convert that to analog with a low-pass filter of reasonable order. (1-st order will probably not be enough.)

Advantages of MEMS: usually more rugged, much smaller. Obviously direct PCB mounting.

[coppercone2]

https://www.digikey.com/en/articles/techzone/2011/may/ultrasonic-mems-sensor-spm0404ud5

[SiliconWizard]

https://www.digikey.com/en/articles/techzone/2011/may/ultrasonic-mems-sensor-spm0404ud5

See the typical frequency response above ~15kHz though - it would require some serious calibration (unless the OP doesn't care much about amplitude). Hope the OP is ready to do that.

[coppercone2]

no idea how to do that easily

[Habropoda]

This app note discusses some of the options and has charts for several mics up to 100kHz:
https://www.knowles.com/docs/default-source/default-document-library/an-17-issue01.pdf?sfvrsn=2

Much depends on where you are using it:  IIRC the 0404 models have a big problem with humidity, the 0410 models are somewhat better and the FG models are the best.  Somewhere I read that the bottom port mics have better response than top port mics in the upper frequencies and that has been my experience.

I've been using the 0410 bottom port mics in bat detectors at 192kHz sample rate with decent results.

[Detzi]

At first thank you all so far. What i can see i was just searching in the wrong category...and havent been lucky enough that google would have pointed me there.

Searching in the right category i have now much more of a selection easy.

The mounting will probably be critical to get reasonable freq response at those high frequencies. I'd guess a top port would be easier to deal with than a bottom port, but what about the enclosure it's going to be in?

So i thought about that before and came to the conclusion that it is the most difficult part for me to build, because i have not much experience with "audio" waveguides. The plan is to get it working and then 3D-Print a Case with the waveguide build into it. I remember seeing i video on youtube about a  guy using a special calculator helping him to match the waveguide to a speaker.

I remember we already had a thread about something similar (would have to find it back). There were very few MEMS microphones for frequencies that high.

I actually searched before opening the thread, could not find something that matched. If you are able to find that thread, i'm glad for any bit of usefull information.

PDM is not that hard to deal with, do not let that deter you. You can basically convert that to analog with a low-pass filter of reasonable order. (1-st order will probably not be enough.)

It's not that i find PDM "bad" it would even have the advantage of being able to easylie mux a stereo mic. The ICS puts the PDM on falling or rising edge of the clock cycle depending on which channel (left, right) you set on the mic. On the opposing edge it goes into tristate. So that is really a charming option. But i would like to have filtering before the delta-sigma Converter to get rid of the stuff i don't need. But PDM might be woth a second consideration if the datasheet holds information about their ADC they use.

See the typical frequency response above ~15kHz though - it would require some serious calibration (unless the OP doesn't care much about amplitude). Hope the OP is ready to do that.

I do not care about the gain over frequency much. It should not be like +-60db but as long as i dont clip i am fine. But a bit of "calibration" in the simulation phase wont hurt anyways i guess. But for now i think it will be fine to match the filter response to the microphone respone and try to flatten them. In my mind its like: MIC-> active 3rd-order Bessel (+ maybe flattening) -> dsADC.

Much depends on where you are using it:  IIRC the 0404 models have a big problem with humidity, the 0410 models are somewhat better and the FG models are the best.  Somewhere I read that the bottom port mics have better response than top port mics in the upper frequencies and that has been my experience.

I've been using the 0410 bottom port mics in bat detectors at 192kHz sample rate with decent results.

Thank you, i will have a closer look at these. Do you know how the humidity exactly causes problems?

[Conrad Hoffman]

My guess is the Knowles curves stop because that's as high as they can measure. I know one of their mics has a bandwidth approaching 200 kHz, but I can't remember which one. I used a small piezoelectric disk to excite it and the bandwidth was far beyond what they claimed.

[SiliconWizard]

My guess is the Knowles curves stop because that's as high as they can measure. I know one of their mics has a bandwidth approaching 200 kHz, but I can't remember which one. I used a small piezoelectric disk to excite it and the bandwidth was far beyond what they claimed.

Yes. I remember the specs of some Knowles products are a bit confusing in that respect, because some state a bandwidth much wider than what the freq response charts show, so all you can guess is that they will work up to 100kHz+, but you just don't know what the response looks like.

[Conrad Hoffman]

I suspect when you get up that high the wavelengths are so short that the response is all over the place, depending on the mounting and surroundings. At 100 kHz we're only talking a few mm.

[SiliconWizard]

Absolutely. And likewise, most app notes for those microphones only mention tips for mounting them for audio purposes (so up to 20kHz or so), and do not help for anything else.

That said, I remember Knowles used to have a pretty nice tech support (hope they still do), so do not hesitate to contact them.

[mark03]

There aren't many MEMS mics fully specified in the ultrasonic range, but according to the Knowles support rep I talked to, almost any bottom-port device will have a similar frequency response as the ones advertised for ultrasonic coverage.  I've been using, e.g., ICS-40730 in a bat receiver with no problems.

As someone already pointed out, the frequency response of these mics above 20 kHz is definitely not flat.  If anyone has an idea for DIY calibration I would be very interested.  If you have another microphone with known response it would be relatively easy, but that means $$$.  I wondered if a very small spark might be a good broadband (flat) source.

PS  it sounds like you are planning to digitize the signal.  Why would you go to all of the effort of building an analog filter when you could just do it in software?

[Detzi]

My guess is the Knowles curves stop because that's as high as they can measure. I know one of their mics has a bandwidth approaching 200 kHz, but I can't remember which one. I used a small piezoelectric disk to excite it and the bandwidth was far beyond what they claimed.

The Application Note linked above let's me guess the same, they just picked three well suited ones and measured them above the regular 10kH.

While in most cases Knowles microphones are used
for receiving audible signals, several models are
also used in ultrasonic applications. Knowles
production line testing typically takes place at
several frequencies up to 10kHz. Therefore the
microphone specifications are also written with
performance data up to 10kHz but not beyond.
While this is sufficient for ensuring the quality of
the outgoing microphone, it does not mean that
the microphone is limited to 10kHz in practice.

There aren't many MEMS mics fully specified in the ultrasonic range, but according to the Knowles support rep I talked to, almost any bottom-port device will have a similar frequency response as the ones advertised for ultrasonic coverage.  I've been using, e.g., ICS-40730 in a bat receiver with no problems.

As someone already pointed out, the frequency response of these mics above 20 kHz is definitely not flat.  If anyone has an idea for DIY calibration I would be very interested.  If you have another microphone with known response it would be relatively easy, but that means $$$.  I wondered if a very small spark might be a good broadband (flat) source.

Thank you that is some very usefull piece of information, i have already contacted the knowles support to assist. I am very courious what they have to say but have no answer yet.

PS  it sounds like you are planning to digitize the signal.  Why would you go to all of the effort of building an analog filter when you could just do it in software?

Connecting the ΔΣ-ADC directly to the microphone would not be very robust neither will it give best results in terms of signal quality. Which means i will have to do some signal conditioning to get best results out of the ADC. So when i'm at it why would i pass the "noise" through the adc when i can cut it off with a few cents worth of extra components. This way i can safe money on the µC and time programming it. I honestly don't feel like building a active filter is much of an effort either so i guess it is personal preference too.