Design Rationale and Selection
For this design project, I wanted to produce a Bluetooth speaker capable of projecting audio to a medium-sized room in my parents’ house, in a package roughly 8” x 4” x 4” that can be powered externally (with battery considerations being something for a later iteration of the project). When defining the circuitry needed for this project, I wanted to locate the following components:
- A microcontroller with a wireless data transmission module
- An RF receiver and transmitter module
- A speaker
- An audio amplifier IC
- A power switching circuit for sequencing of different power domains
For the first two components, I knew that the ESP32 series of modules would fulfill most of my requirements with minimal fuss. At first I had chosen the ESP32-C6-WROOM-1-N8 module, but I soon found a problem with this: this module was designed for circuits that use Bluetooth Low Energy (BLE), not Bluetooth Classic. Bluetooth Classic is the high-bandwidth version used in audio applications, and has many more examples written of how to use in audio applications with the ESP32, so I changed my microcontroller selection to the ESP32-WROOM-32E-N8 midway through the hardware design.
Speakers aren’t something I knew much about before this project, but I gathered that three important criteria to use for filtering were low-end resonant frequency, sensitivity, and audible frequency range. With my novice understanding, I filtered on Digi-Key for speakers with a resonant frequency below 200Hz, a good frequency response from ~20Hz to 20kHz, which are in-stock, have a datasheet available on Digi-Key, and cost less than $20/unit at low quantities. (link to search filters). I ended up going with the PUI AS07808AS-R (Fs = 120Hz) for the first prototype, which has a 82dB sensitivity and a frequency range from 60Hz - 40kHz. Some time after the first prototype I had a call with PUI engineers to review my design choices; in this call, they recommended the PUI AS11508AR-R (Fs = 80Hz) for its better low-end frequency response, which has an 84dB sensitivity and a frequency range from 80Hz - 15kHz.

I then looked at amplifiers that could power this speaker. I wanted to make sure that the amplifier had an I2S interface for digital audio that would allow for abstraction of the microcontroller-to-amplifier interface, the capability to drive a 8Ohm load at a rated power of 15W, and a leaded SMD package (SOIC, TSSOP, QFP e.g.) to enable for easy hand-soldering and probing (link to search filters). I decided to use the TAS5711PHPR amplifier by TI, which met my requirements and was relatively cheap. The speaker has a rated power of 15W and a max power of 30W, whereas the amplifier has a power capability of 20W in the bridge-tied load configuration. This leaves no margin, but I don’t intend to operate the speaker at >15W due to its purpose as a relatively low-power, medium-volume speaker for use in quiet medium-sized rooms. The TAS5731M and TAS5713 ICs also meet the requirements, but have slightly higher power capabilities and compatible pinouts with the TAS5711. In the event that I need a higher power output from the amplifier, I’d evaluate one of those drop-in alternatives first.

In order to drive an 8 Ohm load at the 15-30W range, it seemed most fitting to use an 18V nominal supply range for the PVDD net of the amplifier. I looked for off-board AC-to-DC power supplies with a >20W power output capability (link to selection filters) and chose the following Mean Well part: GST25U18-P1J. Looking back at the search filters now, I’m not entirely sure why I bought the 25W version instead of the 36W version of this power supply. Maybe it came down to which parts were in stock at the time.
The last major component that required some in-depth design decision was power switching to the PVDD and DVDD domains of the amplifier. I wanted to choose a load switch with some overcurrent protection and the ability to be switched from a microcontroller’s 3.3V signal. Toshiba has a nice eFuse IC series of load switch that seemed suitable. I chose the TCKE905ANA,RF part for switching the 18V and 3.3V circuits. These weren’t in the initial v1 design - I had used a combination of PMOS and NMOS fets instead - but they’re a drastic improvement in the ease of control of the switching circuits.
Sean Morton
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.