Alternatives for microcontroller
Hi, my name is Franco Dal Farra, I am an engineering student in Universidad Nacional de Río Negro, in Argentina. I am currently in a project group working in cooperation with the biology area, and we are trying to develop an audio data logger for getting statistical analysis on local frog species. During our investigation we stumbled upon your open device AudioMoth and we found it to be really a great implementation so we decided it could be really helpful to work based on that project.
However, we had some difficulties with the purchase of the microprocessor EFM32WG380. It is no longer available on market in any of the distributors from which we can import to Argentina. Currently we are looking into two different options: buy either EFM32WG280 or EFM32WG980 that have both same encapsulation and outputs than the one you are using in AudioMoth, but don´t have the USB bootloader on the hardware features; or as a different approach buy either EFM32WG332 or EFM32WG942 that have QFP64 encapsulation but do have USB bootloader on the hardware.
We are deciding on what path to take for moving forward but it could be really helpful for us if you give us some advice. Is it better to modify all the external hardware for a microcontroller with different pinout, or is it better to modify the software and do the configuration of the device in a different way than using the USB controller that is on the internal hardware of the EFM32WG380? Could it be possible to use some external hardware to solve the communication interface instead of changing the software?
In any of the cases above we would have to change some part of the software, but as it looks like a really good functional project we would like to make as few changes as possible. Maybe during the development stages your group had to face the decision of which microcontroller to use and all these may have come into discussion so we were thinking that you had some better experience to tell us so we can learn from that. Thanks!
I would recommend starting with the AudioMoth Dev which is intended to make it easy to add additional sensors through the GPIO pins (https://github.com/OpenAcousticDevices/Datasheets/blob/main/AudioMoth_Dev_Datasheet/AudioMoth_Dev_Datasheet.pdf). AudioMoth provides temperature measurements through the internal temperature sensor in the Wonder Gecko (https://github.com/OpenAcousticDevices/Application-Notes/tree/master/AudioMoth_Temperature_Measurements). You could add any additional serial or I2C sensors here, or have a port expander with an additional board for all the sensors. This would allow you to focus on the things that are different from AudioMoth without having to re-engineer everything. Andy (andy@openacousticdevices.info) would probably be able to provide them direct to Argentina. If you wanted to design all the hardware from scratch, I would base it on the EFM32WG380 or EFM32WG980 so you could use as much of the existing firmware as possible. If you use another Wonder Gecko, or another Gecko processor, converting the firmware is probably doable. If you use a non-Gecko processor you'll have to rewrite all the firmware from scratch using the libraries and language supported by the microcontroller distributor. This is quite a big job. If low power isn't a major concern, then using a single-board Linux computer, like a Raspberry Pi, makes it much easier to prototype and you can program in a high-level language. The power requirements would probably be about ten times that of AudioMoth, so it depends on the deployment setting whether that is feasible or not. Overall, it depends on the length and size of your project really, and whether it is intended to as an electronic engineering project or a conservation project. Engineering a device from scratch is generally quite a big project. AudioMoth at the time of its release was the end result of 2 x 3 years of PhD research, and it has developed a lot since then. The AudioMoth firmware is quite complicated as it caters to lots of applications though. Writing the firmware for one specific use case is much easier.