Bat detector

Passive Acoustic Monitoring (PAM) on underwater glider (6-channel volumetric array)

I am imagining a boat, open water, sunshine and sun block. Must be fun...

Can I assume you are monitoring biological sources? Do you do any spatial processing as well?

What sort of hydrophone are you using?

Awesome...

at
 
Maybe people think we might still be struggling to get the device working properly. This is not the case, we are just looking to get it to work even better than it already does.

The next path was investigating how much we could gain by not using mic-in but using line-in instead, so I did some experiments.


I tied some tests with a tone ladder injecting signals on mic in and line in at 0, 10 and 20db
I did not really see any difference, just one important difference was noticed. When ik signal generator was disconnected the noise level dropped after a few miliseconds. Looking like some "mute" function on the line input.

After these tests I wanted to confirm what I found on a bat recording so I set op my dac and speaker and recorded using the SPU microphone and amplifier in the style I was used to. Amplification was set to 9db for both line-in and mic-in.

This is an image of two recordings, the upper is on the line in, the lower is mic in.

ab1.jpg

You can see lighter vertical lines at the moments where no sounds are registered, they are similar to the lower noise signal i did see after disconnecting the signal generator in earlier tests.

I took some spectrum images at different points. This one is the spectrum around the ping at about 0.07-0.08

Line in
ab2.png

Mic-in
ab3.png


This piece is the silence between two "pings"at 0.08 - 0.18

Line-in
ab4.png

Mic-in
ab5.png


These are the short pings (feeding sounds) at 0.33-0.36

Line-in
ab6.png

Mic-in
ab7.png


Somehow the silences seem to be more silent but at the pionts where we are registering some sounds we can see almost no difference in the signals and the noise.

To me it seems the mic in will get us recordings just as good as the line input.

Edwin
 
Mic Diff AOP amplifier PCB ICS40740 40730

After many weeks I try my small pcb with simple differential amplifier: Betters resuls for spurious at 27Khz.SMD small form-factor PCB: can be added to the input connectors.Design details*:
AOP: ADA4807: low cost,low noise,low power,available farnell RS and LTC simulation.
Gain: about 2 : can be adjust with 2 classical Resistor.
Some filtering : simple input HPass to reduce intelligible audio voice and other noise.
Low pass : mandatory for AOP stabilisation, simple first order.
Differential Impedance seen by mic can be adjusted (750 ohm for this test, very close to ICS 40730 specification )
I keep the modified single transistor input stage, at the output of this amplifier:I can compare many mic inputs by simple removing the front connector in live.

Corbee, PE1pwf, if you want one Pcb,ask me. I can also provide some SMD parts for you.

Design files : Be careful there is a mistake in my PCB, visible in schematic. See easy correction on PCB with through holes resistor. I remove the GBRfile from project

Next step : Define the optimal impedance and amplitude at the codec line input with correct #define value in code.
When this amplitude wil be well defined we can increase AOPdiff amplifier gain and remove first stage transistor amplifier.

I’will also be interested to try the new ICS40740 and look at this 27Khz spurious signal.
 

Attachments

  • KICKADproj_diffaopV1.zip
    49.8 KB · Views: 123
  • 3D_PCB.jpg
    3D_PCB.jpg
    86 KB · Views: 139
  • PCB_frontCAse.png.jpg
    PCB_frontCAse.png.jpg
    143.6 KB · Views: 216
  • TOP_BOT_PCB_hand.jpg
    TOP_BOT_PCB_hand.jpg
    85.3 KB · Views: 154
  • case_screen_top.jpg
    case_screen_top.jpg
    91.7 KB · Views: 129
Congrats, good to see you did not give up hope ;) I am interested to see how it works, did you allready try to record some bats ?

regards
Cor
 
Last evenings of busy bats, bad computers and struggling with Kicad.


I think this sums up most of the activity here.
My computer stared acting up telling me not to shut down but configuring stayed at 100% and did never finish. I don't quite understand why and how but somehow it is running again and I can use it I had some scary moments. Almost considering a new install of the OS.

This weekend we had some warm evening after a few evening below zero but on these warm nights the bat's came out again and man they were busy.
They sounded like pipistrellus but I did not check any recordings. This evening is not very cold and I heared one flying by the house again. I hope to be able to monitor te check out some old bridge foundations which were adapted so bats can use it during the winter. It would be nice if they actually start using it.

Remis, your boards look nice, I just wondered why you are using tru hole resistors for R10 and R11
I wanted to have a peak at the schematic, and I guess I should just have asked for it. I installed Kicad which took a while and I tried to work with it but I guess I am top much used to working in Eagle. Well I ventually got it to whork and I brushed up the schematic a little and for those that are wondering and did not install kicad yet this is what it looks like.

kicadmicamp.jpg

I buit something similar but had the feeling I was nog very more pleased with this than with the SPU microphone.
I did have the feeling that the microphone I was using had a "dip" in the sound registration around 70Khz.
I wonder how it looks to you and if you had the time to test your bat detector.

During the time I'm writing this I had seven fly-bys here in the garden.

Kind regards,

Edwin
 
Hello Edwin
20 years ago i worked with my first PC , win, and eagle. Today my choice is: PC , Linux, and first project in Kickad. Easy, enough for our project, long term support. No cost. Kickad work also under win. A lot of shared libs and tutorials. LTC simulation works fine with wine. Arduino works under linux.

I keep both SMD and trough hole resistor for gain adjustment You can choose. Throught hole technology is more easy to buy, found, and change. You can adjust the gain with same value for R10 and R11. With 5600 ohms the gain is ‘’low’’ and the output is connected to the transistor first stage input, without any saturation effect.

I don’t see the 27Khz spurious. I hear some bats , but I don’t have enough around me, in this city.I don’t performed enough test to talk about this deep a 70Khz. Spectrum seem to be flat. And previous measurement, without mic, with generator confirm simulation. The High pass filter reduce the noise coming from classical sounds.

Major interest of AOP in diff mode is the doubling voltage seen at the mic output : both signal are in opposite side. Thus, 6dB improvement. I can also adjust the differential impedance seen by the microphone for investigation. You can have few PCBs (without correction ) if you want.
Remi
 
This is my first time post, but I have been following this thread for a long time. I'm so impressed with the work so far! I'm interested in recording and monitoring bat sounds in an art and activism context, to share my fascination with the species that live around us (above us!) by making their communication audible. So far I have built a very simple bat detector using a Teensy 3.2 + Audio board + ECM mic at 192kHz, running the granular synth object in pitch-shift mode (modded the lib to reduce minimum pitch multiplier to .10).

I use this as my realtime monitor while making recordings on a professional sound recorder, which I later slow down for recording and analysis. For an upcoming project I'm trying to develop an online livestream so people can tune into the ultrasonic world of bats. In this case aesthetics are more important than data, so I have been evaluating the sounds of bat detection techniques to find the most sonically engaging method.

I have a few questions about the project so far:

- I have been thinking a lot about the AUTO_TE mode. After listening to longer recordings at .10 speed, I have always wanted to experience such sounds in near realtime, using the gaps between calls to expand the playback. I wonder if it would it be possible to start the x0.10 playback as soon as the detector thresh is passed and the buffer begins to fill? Then, in the background we still monitor the thresh for another event. If another event occurs we stop playback, flush the buffer, and start again. That way all calls are preserved, though some will be truncated. I should probably simulate this via editing to see how it really sounds!

- I have been experimenting with ultrasound recording in stereo, and there are some very interesting spatial effects, especially with slightly spaced capsules and directive horns. Buffer length would be halved but perhaps there is still enough RAM?

- Teensy 4 should have more RAM for audio buffering, but is that currently enabled in the Audio Library? (recent commits seem to suggest that it has the same max audio delay as the Teensy 3.6) Also, the pinout is different, hence the new rev. of the Audio Board.

- Edwin: I am thinking I should just get one of your boards and start building the current version of the detector before I hack too much on my variation of AUTO_TE. My only question is whether you are planning an update for Teensy 4? I haven't looked at the changes in pinout, maybe I could just cut some traces and re-route by hand?

Thanks -Zach
 
This is my first time post, but I have been following this thread for a long time. I'm so impressed with the work so far! I'm interested in recording and monitoring bat sounds in an art and activism context, to share my fascination with the species that live around us (above us!) by making their communication audible. So far I have built a very simple bat detector using a Teensy 3.2 + Audio board + ECM mic at 192kHz, running the granular synth object in pitch-shift mode (modded the lib to reduce minimum pitch multiplier to .10).

I use this as my realtime monitor while making recordings on a professional sound recorder, which I later slow down for recording and analysis. For an upcoming project I'm trying to develop an online livestream so people can tune into the ultrasonic world of bats. In this case aesthetics are more important than data, so I have been evaluating the sounds of bat detection techniques to find the most sonically engaging method.

I have a few questions about the project so far:

- I have been thinking a lot about the AUTO_TE mode. After listening to longer recordings at .10 speed, I have always wanted to experience such sounds in near realtime, using the gaps between calls to expand the playback. I wonder if it would it be possible to start the x0.10 playback as soon as the detector thresh is passed and the buffer begins to fill? Then, in the background we still monitor the thresh for another event. If another event occurs we stop playback, flush the buffer, and start again. That way all calls are preserved, though some will be truncated. I should probably simulate this via editing to see how it really sounds!

- I have been experimenting with ultrasound recording in stereo, and there are some very interesting spatial effects, especially with slightly spaced capsules and directive horns. Buffer length would be halved but perhaps there is still enough RAM?

- Teensy 4 should have more RAM for audio buffering, but is that currently enabled in the Audio Library? (recent commits seem to suggest that it has the same max audio delay as the Teensy 3.6) Also, the pinout is different, hence the new rev. of the Audio Board.

- Edwin: I am thinking I should just get one of your boards and start building the current version of the detector before I hack too much on my variation of AUTO_TE. My only question is whether you are planning an update for Teensy 4? I haven't looked at the changes in pinout, maybe I could just cut some traces and re-route by hand?

Thanks -Zach

Hi Zach,

Welcome to the forum. Its always nice to see new ideas/users actively participating. On to your ideas/questions:

The current setup of AUTO_TE is set to start an AUTO_TE replay (from memory and not from SD !!) as soon as an ultrasonic is detected. I have been thinking about the possibility to have two buffers. Both waiting until one gets triggered and starts replay whilst the other keeps collecting data (instead of playing) and can take over immediately after the first has stopped. That more or less is in line with your thinking I guess ?

Moving to Teensy4 is something for the future, more ram/more speed but a different pinout and a new Audioboard will have different demands on both hard/software.

regards
Cor
 
Zach, I have no plans to change to the Teensy 4, the layout is different and also the sd card socket is not on the teensy 4 although it seems people started building extension boards which do make things available.
I also have a small stack of boards which I do not want to throw away so I will stick with those.

The boards were all tru-hole boards, but the latest has an option to place an SMD SOIC8 dual opamp or still use the DIP8 dual opamp.

Kind regards,

Edwin
 
It seems this tread is going into hibernation, just like the bats.

I don't seem to be ready for hibernation, found a bat sound that I could not quite place and wondered how "flat" the response of my microphone could be.
I tried to test with an audio sweep but the output of my ribbon tweeter on my dac or signal generator probably also has better performance on certain frequencies than others.

I started to use running water from a faucet with aerator that seems to have al lot of ultrasound in it but I have no idea how much to expect on what frequencies. It would be nice if someone could make some recordings with a high end professional recorder so I could compare audio spectrums.

Here are some of my experiments that I conducted sofar in this PDF file.
https://drive.google.com/open?id=1Zwt5ezAhhttOtyjX2ilaADhomfn-UOoM

I anyone can think of a nice ultrasound source that gives reasonably flat audio over a large part of the spectrum we are looking for that would also be nice.

Kind regards,

Edwin
 
Edwin, thanks for the Teensy 4 advice. For now I will stay busy assembling the bat detector based on your guide and a Teensy 3.6. I just placed an eBay order for your board + several of the MEMS mic boards to experiment with. (Do you use a reflow oven? I have wondered about DIY techniques for soldering MEMS mics since the pads are hidden.) Concerning the cone and weatherproofing tests in your recent document: Have you seen this? Assembling cheap, high-performance microphones for recording terrestrial wildlife: the Sonitor system This paper compares several capsules, cone designs, and weatherproofing membranes. (I think this link has not yet appeared on this thread, but apologies for re-posting if it has!)

Cor, thanks for the clarification about AUTO_TE. 2 buffers makes sense. If the speed reduction factor was small enough then it may be nice to have the "background" buffer interrupt foreground playback if another event occurs, so the timing between events would be preserved. Of course this means cutting off the "tails" of closely-spaced events, but 2 buffers means that the interrupted event could be faded out while the new event starts. Lots of interested aesthetic decisions to make. I'm looking forward to contributing once I get my detector built!

I am also interested in experimenting with stereo AUTO_TE, but I see Edwin's tests indicate that the Teensy Audio Board line input has a hidden "noise gate" of some sort. Perhaps there's a register in the SGTL5000 codec that the lib isn't setting properly? (Note that the auto volume control has been messed up for years due to a similar problem.)

-Zach
 
Hi Zach,

Considering 2 buffers and interupting the current play, I dont know if that would be wise. If you have picked up a signal and start slowing it down with TE when can we infer that initial signal is "finished" and can we switch to the 2nd buffer ? Not too easy I think, but its something we can at least try to have a look at. My initial guess was simply, keep recording raw data in a 2nd buffer when playing from the 1st buffer and when that play is "finished" see if the 2nd has interesting data. If so, play that also slowed down. But this is all a bit too much conceptual, the real coding will probably have limitations (memory !).
For stereo AUTO_TE you need two working data-processing pipe-lines with incoming data ... using a single teensy looks challenging to me. A pair of teensies 3.6 might be the easiest way to ensure you have enough computing power and you will need to synchronize L and R microphones.

Hope you can find time to build using the nice boards Edwin has made.

regards
Cor
 
Zach,

I just packed the board and microphones. Expect a small refund because of the combined shipping.

The soldering can be tricky, I have seen people soldering by clamping the microphone with pre-tinned pads in an alligator clip and heating it with a hot air soldering device.
I am using a kind of hot plate, I put a pre-tinned microphone om the thin pcb, but the pcb on the hot plate. The soldering is done within 10 seconds and I am sure the tempereature stays well below the 260C max peak temperature.

All additional parts (capacitors) are soldered by a soldering iron.

I have seen the document you mentioned, they use special Gore(R) acoustic vents. I was not aware of newer versions of this document.
I used simple household cling film to see what it does and was quite amazed how well ultrasound seemed to pass.

I would like to do some more cone tests, the document does not describe what te cone looks like.
The cone in the picture seems quite shallow which should give less directional effect than the 90 degree cone I tested.

I woud love to have more info in acoustic cones in the ultrasound range, the documents referred to are not free and I rather do some fun experiments than buy a book.

Kind regards,

Edwin
 
I wonder if the z-axis layering of the 3D-printed horn causes audible artifacts since the ultrasound wavelengths are very short.

Some simple observations about ultrasound horns:

Peterson's M500 detector looks like it uses a Knowles FG capsule. Their removable horn looks like a very narrow conical design.

The Dodotronics USB ultrasound microphones use FG sensors too, but with a very shallow conical horn.

I've never used the Echo Meter touch from Wildlife Acoustics, but the horn looks much more flared, likely exponential instead of conical. Not sure what's up with those slots either.
-Zach
 
A few weeks ago v100 of the code appeard on CorBee's Github.

There is a nice feature added, "Autorecord"
This function starts a recording for a defined time followed by A defined pause after a sound was heard above a minimum frequency and minimum volume.

I tried to compile this in the latest Arduino/Teensyduino this seems to work fine.
Cor needed to add some functions to the audio library to get this to work.

To be sure people have the least problems compiling I created a zip file that contains all the necessary files to compile in windows.
This zip file can be unzipped into the "My documents" folder.

It basically contains the code by Cor renamed to a .ino file and set "compiler ARDUINO" in stead of PIO to get around a problem with Time.h and the necessary library files.

If you already have special library files in "My documents"/Arduino/libraries there is a chance you overwrite them when extracting this zip file.

If I have anything interesting, like new code, building instructions or other interesting experiments I will try to store them in this folder on my google drive.
The code still has some experimental settings in it and the manual is not quite updated yet but I think the added feature speaks for itself.

Like the bats, probably most bat-lovers on the northern hemisphere are in hibernation but it is nice to know there still are some interesting developments in the code.

Thanks a lot for this code Cor.

Kind regards,

Edwin
 
good evening

I came across this forum by chance and I am very interested in this project.

I saw on Ebay that it was possible to get this PCB and a microphone. Is this the latest version?

Regarding the case, has anyone already made a case with a 3D printer?

Finally last thing, can we connect an earphone or a small speaker to listen to the signal live?

Thank you for your details and sorry for my bad English!

Manu
 
Hi Manu,

I can not see any bad English, just looks fine.

Where are you form?


The PCB and mic on ebay is the latest version, I do not know if anyone crated a 3d print file for a box. maybe someone did but I have not heared about it.
You probably cannot print for the price of the used enclosure but I have to agree cutting the hole takes some efforts.
I also had questions about complete kits, i am sorry there is not complete kit (yet) I might assemble partial kits.......

A complete build would probably cost near 100 euro's I do not want to be stocking a buch of 100 euro kits.


The teensy audio board has a connector for a headphone, if you want a speaker, you need some kind of amplified speaker.
Yo can listen listen live in Time expansion mode but also heterodyne, auto-heterodyne and timedivision modes.


On youtube there are a few video's I made, not with the latest software but they give a good impression.
I am sure you will find it when searching for teensy bat detector


I also have been working on a new user manual for the v100 sofftware, I did not have the time to check it and probably do have some additions but I put it on my google drive.



I think this answers most questions, maybe someone has already been working on a 3d printed box....


Kind regards,

Edwin
 
Hello Edwin,

Thank you for the answer.

I live in Alsace, 50km from Starsbourg.

I have been living in a new house for a year near a river and I noticed a lot of bats in the evening last summer.
I am therefore interested in this project, it seems to me that it is the best for a reasonable price.

I'm going to order the PCB and the microphone I think this week.

Regards

Manu
 
Oh I see, probably France as I see the way you are writing the names, but quite close to the border of Germany I think. Anyway I think your English writing is fine.

If you want to have a more than a Frequency Division or Heterodyne detector I think this is a good choice.
Time division does sound a lot better than FD and Heterodyne (my opinion). Recording is very nice if you want to determine the bat species.

I guess you should expect a bit more from the commercial available devices but with the same features these cost at least ten times more than this detector (not taking building time in account)

There are some people that think other microphones and microphone amplifier circuits should have some benefits but the circuit as is performs very nice.


Just for fun I added an image of a recording I made. There are a lot of echo's behind this call.
It probably is some kind of Myotis with very wide band FM signal. It looks like a species I should not expect here.

Anyway, I think the frequency range it covers is very good. On my google drive you can see I also conducted some microphone tests. If looks like this is a fine set-up.

Myotis.jpg


Edwin
 
Edwin,

you saw well, I live in the east of France very close to germany.

I just ordered the PCB on Ebay!

Manu
 
good evening

I came across this forum by chance and I am very interested in this project.

I saw on Ebay that it was possible to get this PCB and a microphone. Is this the latest version?

Regarding the case, has anyone already made a case with a 3D printer?

Finally last thing, can we connect an earphone or a small speaker to listen to the signal live?

Thank you for your details and sorry for my bad English!

Manu

I created a 3D printed lid that replaces the lid from the 116x68x36mm Project Case as described by Edwin in his instructions when using his PCB. The 3D model can be found here: https://www.tinkercad.com/things/7gZ0wq9Fpcm

Here are some pictures:

TeensyBat_Lid.jpg

Lid.JPG

I hope this is useful to someone. Note that the distortion on the edges of the lid was due to non-optimal printing conditions. This lid is designed to fit the bottom part of the 116 x 68 x 36mm black project box.

Adrian
 
hello, I'm new to the forum, I went through it, thinking to get a more precise idea on a bat detector. Well, it's more confusing than before, but I learned things !!! :)
I had the idea of making a detector with an esp8266, wifi, and a mobile for the display.
I do not know the teensy, but apparently it is programmed as an arduino, therefore as an ESP. (the price is not the same)
my question why a teensy?
can we adapt the program to an ESP?
thanks for reading.
 
Back
Top