Still takes some days for the additional capacitors to arrive. For now i have rebuilt it on a Breadboard and added R5 (57 Ohm). And the high pitch hissing sound is gone. But it still works very badly. Here a sound example: Sound example
According to the already published paper the circuit which i built should have results as attached picture sows (ELOC is the one i am trying to build).
The complete paper can be read here: Locate Elephants using Infrasound
Maybe i got the wrong circuit diagram? I don't know. Communication with these university profs takes always longer than i wish. And i am motivated right NOW
I am not sure if it is too much to ask. But i have the feeling that this forum is the right place to get a much better design for our purpose?
The University in Sri Lanka is building their own preamps and send them here to Sumatra. It would be really good if i can have a better design. So i can compare them in the jungle with elephant sounds later with the ones from Sri Lanka.
It should be cheap (preferably less than 10 USD) and parts easily available.
Infrasound travels the furthest so it should be sensitive to low frequencies.
Another question which i ask myself: Does the low pass filter really make a big difference for sensitivity in low frequencies or can we just use software to filter out the frequencies we need for the AI?
A donation for the forum is the least i can do. It's all volunteering. But i founded a DIY-projector forum in Germany 20 years ago so i kind of feel a connection
According to the already published paper the circuit which i built should have results as attached picture sows (ELOC is the one i am trying to build).
The complete paper can be read here: Locate Elephants using Infrasound
Maybe i got the wrong circuit diagram? I don't know. Communication with these university profs takes always longer than i wish. And i am motivated right NOW
I am not sure if it is too much to ask. But i have the feeling that this forum is the right place to get a much better design for our purpose?
The University in Sri Lanka is building their own preamps and send them here to Sumatra. It would be really good if i can have a better design. So i can compare them in the jungle with elephant sounds later with the ones from Sri Lanka.
It should be cheap (preferably less than 10 USD) and parts easily available.
Infrasound travels the furthest so it should be sensitive to low frequencies.
Another question which i ask myself: Does the low pass filter really make a big difference for sensitivity in low frequencies or can we just use software to filter out the frequencies we need for the AI?
A donation for the forum is the least i can do. It's all volunteering. But i founded a DIY-projector forum in Germany 20 years ago so i kind of feel a connection
Attachments
I have only had time to read a part of the paper so far, but I have one question and one remark:
-What is done when the elephants are localized?
-The conclusion that Eloc outperforms INFRA-20 is not convincing to me, because what matters is the signal to noise ratio rather than the sensitivity - it's easy to increase gain, but there is no point in doing so when the microphone noise already dominates.
There are some similarities to this project: Need Advice for a Weird Project , although Hal is trying to detect birds rather than elephants and therefore wanted a high-pass rather than a low-pass filter.
I couldn't access your sound file, but if the sound was very distorted, that could be due to the LM358's class-C output stage. A 4.7 kohm pull-down resistor from the output (pin 1) of the LM358 to ground could improve that, although a better solution is to use a better op-amp.
-What is done when the elephants are localized?
-The conclusion that Eloc outperforms INFRA-20 is not convincing to me, because what matters is the signal to noise ratio rather than the sensitivity - it's easy to increase gain, but there is no point in doing so when the microphone noise already dominates.
There are some similarities to this project: Need Advice for a Weird Project , although Hal is trying to detect birds rather than elephants and therefore wanted a high-pass rather than a low-pass filter.
I couldn't access your sound file, but if the sound was very distorted, that could be due to the LM358's class-C output stage. A 4.7 kohm pull-down resistor from the output (pin 1) of the LM358 to ground could improve that, although a better solution is to use a better op-amp.
Last edited:
The paper in very short:
The rumbling from elephants can be picked up from mics at max 2 km distance. So there will be a mesh network of microphones maybe 1km apart.
A microprocessor will trigger and sends the data through a LoRa network to our server if a elephant sound is recognized. With this data the life location elephant groups can be calculated and if the AI is good enough even from each elephant because they of course have their own specific "voice".
-What is done when the elephants are localized?
The aim is not to use collars any more (costly, dangerous and not animal friendly). One of the biggest threat of elephants are local farmers. Very often elephants destroy their crops and the farmers shoot them. So it should act as a early warning system (for now electric fences are used or GPS data from collars). Also poachers are caught more easily when there are gunshots near elephants which would be picked up from the mics and sent via LoRa mesh-network to guards. Research of elephant behavior is of course another reason.
-The conclusion that Eloc outperforms INFRA-20 is not convincing to me, because what matters is the signal to noise ratio rather than the sensitivity - it's easy to increase gain, but there is no point in doing so when the microphone noise already dominates.
That is why i am building these preamps and test them first at a elephant camp nearby. To see if they can be used instead of the 400USD INFRA-20. Some researchers already use these for elephnats which are mics made for bioaccoustic research. They also work. But also cost 300USD.
Many good microphones are able to record till 10Hz or so. But i think they are not sensitive enough in these low frequencies?
Sorry about the sound file. Here the working link.
The rumbling from elephants can be picked up from mics at max 2 km distance. So there will be a mesh network of microphones maybe 1km apart.
A microprocessor will trigger and sends the data through a LoRa network to our server if a elephant sound is recognized. With this data the life location elephant groups can be calculated and if the AI is good enough even from each elephant because they of course have their own specific "voice".
-What is done when the elephants are localized?
The aim is not to use collars any more (costly, dangerous and not animal friendly). One of the biggest threat of elephants are local farmers. Very often elephants destroy their crops and the farmers shoot them. So it should act as a early warning system (for now electric fences are used or GPS data from collars). Also poachers are caught more easily when there are gunshots near elephants which would be picked up from the mics and sent via LoRa mesh-network to guards. Research of elephant behavior is of course another reason.
-The conclusion that Eloc outperforms INFRA-20 is not convincing to me, because what matters is the signal to noise ratio rather than the sensitivity - it's easy to increase gain, but there is no point in doing so when the microphone noise already dominates.
That is why i am building these preamps and test them first at a elephant camp nearby. To see if they can be used instead of the 400USD INFRA-20. Some researchers already use these for elephnats which are mics made for bioaccoustic research. They also work. But also cost 300USD.
Many good microphones are able to record till 10Hz or so. But i think they are not sensitive enough in these low frequencies?
Sorry about the sound file. Here the working link.
Last edited:
The farmers don't shoot them if they don't destroy their crops.
The units will be installed all around the jungle. Not only at farms. (This has been done already OFLLINE in Congo with expensive SWIFT recorders. For research purposes only. See the OFFLINE mics in the attachment)
So with ONLINE mics we are able to see and predict their movement way BEFORE they reach the farm and send a ranger there to scare them away or tell the farmers to scare them away. This has been done so far but without the mics.
We also need to know when/if they leave the national parks.
All the ideas how to solve that with other solutions for example drones... satellite pictures... seismic activity sensors, etc... It's another topic and many things has been tried already. That would go too much off topic if we'd discussed all that now
The units will be installed all around the jungle. Not only at farms. (This has been done already OFLLINE in Congo with expensive SWIFT recorders. For research purposes only. See the OFFLINE mics in the attachment)
So with ONLINE mics we are able to see and predict their movement way BEFORE they reach the farm and send a ranger there to scare them away or tell the farmers to scare them away. This has been done so far but without the mics.
We also need to know when/if they leave the national parks.
All the ideas how to solve that with other solutions for example drones... satellite pictures... seismic activity sensors, etc... It's another topic and many things has been tried already. That would go too much off topic if we'd discussed all that now
Attachments
Last edited:
There aren't any wild elephants in the Netherlands, so I didn't know you could scare them away. It's clear to me now.
I agree with epicyclic's remark about further increasing R5. The smaller values suggested by me and others were only meant for stability, but then you haven't got a useful low-pass filter yet.
The Panasonic WM-61A and WM-61B have an equivalent acoustic noise level of 32 dB(A), while a PUI Audio AOM-5024L-HD-R only has a noise level of 14 dB(A). It might be worth looking into that if its price of $3.16 at Digi-Key is acceptable and if wind noise doesn't dominate anyway.
For his bird monitoring project, Hal used the same 5 V supply that supplied his single-board computer (see the thread I linked to in post #23), but you use 9 V batteries for the preamplifiers. Why do you prefer 9 V batteries?
Hal had some difficulty with moisture ingressing in his preamplifier unit, but he managed to solve that. Maybe it's a good idea to ask him exactly how, so you don't run into the same issue that he already solved - unless you already know how to prevent that, of course.
Last edited:
the sound samples weren't the best but the LM358 can be powered with as much as 32 volts why not try two 9V batteries in series? at double the operating voltage that would mean less current for the same operational conditions, sound in the samples seemed like the mic signal was clipped due to lack of supply voltage.
is use of small diaphragm mic really the way to go to capture low low frequency information?
How to Convert a Speaker into a Dynamic Microphone | Recording, Engineering, and Music Production - YouTube
is use of small diaphragm mic really the way to go to capture low low frequency information?
How to Convert a Speaker into a Dynamic Microphone | Recording, Engineering, and Music Production - YouTube
Last edited:
Some more testing was done and it looks really promising already:
Removed C2 and R5 became 1K Ohm. Also added a 4,7K Ohm from pin 1 to GND. This is how it sounds now:
Sound example
The humming noise is at 50Hz which i guess is interference from the 220V so a proper shielding should get rid of that.
My tiny cheap subwoofer (which was only 40cm away) is not really efficient if it comes to 20Hz or lower i guess. So need to test infrasound otherwise.
is use of small diaphragm mic really the way to go to capture low low frequency information?
I had exactly the same thoughts and looked into bigger diaphragm mics but got the answer from a pro that this is not the case. Open the first comment on this Youtube video and look for EDsteve to see a more detailed answer.
Removed C2 and R5 became 1K Ohm. Also added a 4,7K Ohm from pin 1 to GND. This is how it sounds now:
Sound example
The humming noise is at 50Hz which i guess is interference from the 220V so a proper shielding should get rid of that.
My tiny cheap subwoofer (which was only 40cm away) is not really efficient if it comes to 20Hz or lower i guess. So need to test infrasound otherwise.
is use of small diaphragm mic really the way to go to capture low low frequency information?
I had exactly the same thoughts and looked into bigger diaphragm mics but got the answer from a pro that this is not the case. Open the first comment on this Youtube video and look for EDsteve to see a more detailed answer.
here's what i know to be the bench mark for low frequency infrasonic capture...you won't want to request the price without a bottle of scotch at hand https://www.bksv.com/en/transducers...-cartridges/4193?gclid=EAIaIQobChMI-Oup5v238A
hey just listened to the sound file now i'm thinking that the input filter and lack of high frequency decoupling was simply creating an oscillator you've got wider band operation but at the end of the day your not interested in the higher frequencies unless the harmonics of the input are significant to the low frequency prime for direction analysis.
i'm no good with spice and other sim's so hopefully someone who's better equipped then i am can help you, it could be as simple as the right filter order on the front end.
i'm no good with spice and other sim's so hopefully someone who's better equipped then i am can help you, it could be as simple as the right filter order on the front end.
Last edited:
but from my understand your only interested in the mid 200's and down no?
Mostly yes. There are already bioacoustic recorders which seem to be good enough for the whole frequency range. The SWIFT recorders for example are used in Congo for elephant infra sound and other environmental sounds.
Sumatran elephants have never been recorded so far and our team is basically using different recorders to compare them. The main goal is to have something cheap and open sourcewhich can be used for the final live tracking system which i explained earlier)
The next months we will use the expensive SWIFT recorders, another all purpose low cost recorder and the recorder i am building right now which should be made more sensitive for low frequencies. So in the field we can see which one is best for the final project . Or maybe a combination of those.
who's your pro?
The pro i was talking about is from this youtube video i linked in my last post. Search for Phil Hofman in the comments. At least he sounds like a pro
Mostly yes. There are already bioacoustic recorders which seem to be good enough for the whole frequency range. The SWIFT recorders for example are used in Congo for elephant infra sound and other environmental sounds.
Sumatran elephants have never been recorded so far and our team is basically using different recorders to compare them. The main goal is to have something cheap and open sourcewhich can be used for the final live tracking system which i explained earlier)
The next months we will use the expensive SWIFT recorders, another all purpose low cost recorder and the recorder i am building right now which should be made more sensitive for low frequencies. So in the field we can see which one is best for the final project . Or maybe a combination of those.
who's your pro?
The pro i was talking about is from this youtube video i linked in my last post. Search for Phil Hofman in the comments. At least he sounds like a pro
Last edited:
don't get me going on the who's a pro thing....i just wanted to offer you a low cost way of looking at low frequency measurement capability if you have not looked into my profile i can assure you that the understanding of what it takes do be in your domain of interest is not outside the scope of my interest or knowledge...
@turk 182
Don't get me wrong. You are definitely a pro too in my eyes. But it seems even experts have different opinions and i am in no position to say which one is right or wrong. At the end i can't try it all. But the BK mic didn't look as a cheap solution to me
MarcelvdG also suggested a mic with larger diaphragm(Thanks and sorry that i forgot to look into that actually).
So i guess i followed the wrong path and "larger" might be the right way to go. 4USD is totally in the budget. Unfortunately i can't get it in Indonesia. So i will use what i have so far and a future version will use a different mic.
Shipping to Sumatra takes time and "special" import taxes (my brother sent me a picture from his son and even that cost me 2,5 USD "import tax"...)
Already curious about how these two mics will compare
Don't get me wrong. You are definitely a pro too in my eyes. But it seems even experts have different opinions and i am in no position to say which one is right or wrong. At the end i can't try it all. But the BK mic didn't look as a cheap solution to me
MarcelvdG also suggested a mic with larger diaphragm(Thanks and sorry that i forgot to look into that actually).
So i guess i followed the wrong path and "larger" might be the right way to go. 4USD is totally in the budget. Unfortunately i can't get it in Indonesia. So i will use what i have so far and a future version will use a different mic.
Shipping to Sumatra takes time and "special" import taxes (my brother sent me a picture from his son and even that cost me 2,5 USD "import tax"...)
Already curious about how these two mics will compare
Last edited: