Cortex-M4 mini-evaluation board

05 February 2021
Although I had previously made a host board for the ST Microelectronics STM32L4R5 back in 2018 all I ever got round to doing with it was get together a working linker script with bootstrap code and do some basic GPIO. At the time it got left behind as I put gave priority to getting up to speed with the NXP LPC11(U)xx and later the Kinetis KE0x Cortex-M0 microcontrollers, and then was overtaken by events. Aside from working out how to to get RS232 send & receive working, which in practice would mostly be about working out the idiosyncrasies of the UART's BAUD rate generator, there was little else I could have done with it.

STM32L4R5 mini-eval board

I recently found the time to design and build a more featureful “mini-evaluation” board and the result is shown in the image above. The main additions are fanning out the pins for I2C and USB because getting those two things working would be the first thing I would want to do with this Cortex-M4 chip. With fabrication costs also being less sensitive to board size I decided to add in some extra bits for convenience such as a power jack.

Circuit design

Because the previous host board was designed using KiCad 4 which I have long since stopped using in favour of KiCad 5, this revised board is a complete redesign although some superficial similarities remain. The circuit design follows the power supply recommendations within technical note AN4555, and fans out dedicated connections for programming, I2C, RS232, and USB. In trying to keep it general-purpose some pins power on-board indicator LEDs and there are two connectors intended for GPIO — one 8-way 2.54mm pin socket and one 16-way surface-mount MicroMatch connector.

PCB layout

Fabrication & building

Like many of my more recent PCB designs the KiCad files have been open-sourced and are available from my Bitbucket repository. Due to horrendous intercontinental shipping delays I opted to use Aisler rather than JLCPCB as last time I used the latter the boards got stuck in Belgium and due to travel restrictions was unable to receive them. Although I did order in a solder stencil I ended up hand-soldering the circuit boards instead of reflow soldering them — the trick seems to be to liberal with the flux and conservative with the solder.

List of components

Most components are from recently-ordered stock but since things got a bit mixed up in relocating all my electronics stuff some of the order codes below might be wrong. Based on my experience with the first-generation breakout board a lot of the capacitors could probably be omitted, and in the case of the ferrite bead FB1 there is no recommended value beyond mentioning that it could be replaced with a short. One thing to watch out for is tantalum capacitors being polarised — I only realised this when it came to soldering the circuit, and had I known in advance I would have used ceramic capacitors instead.

Pad Component Manufacturer Part number
U1 ARM Cortex-M4 ST Microelectronics STM32L4R5VITx
U2 3.3-volt power regulator Texas Instruments UA78M33CDCYR
X1 16MHz Oscillator Multicomp MCSJK-6NC2-16.00-25-B
D1,D2,D3,D4 LED Kingbright L-113IDT
R1,R2,R3,R4 10kΩ resistor Multicomp MCWR08X1002FTL
R5,R6,R7,R8 1kΩ resistor Multicomp MCWR08X1001FTL
C1,C4,C5,C9 0.1μF (100nF) capacitor Samsung CL21B104KACNNNC
C10,C11,C12,C13
C2,C3,C6 1μF tantalum capacitor Vishay TMCP1D105KTRF
C7 0.01μF (10nF) capacitor Samsung CL21B103KCANNNC
C8 4.7μF capacitor Vishay VJ0805G475KXXTW1BC
FB1 Ferrite bead TDK MPZ2012S601AT
J1 16-way MicroMatch socket TE Connectivity 1-2178711-6
J2,J3,JP1 3-way pin header n/a
J4 8-way pin header
J5 Barrel jack Cliff Electronic FC681465P
J6 USB Mini-AB Multicomp MC32598
J7 20-way IDC socket Multicomp MC-254-20-00-ST-DIP
SW1 Push button n/a

I have forgotten where I obtained the push button switches but suspect it is from a batch I ordered off E-Bay quite a while back; I am also unsure what order number was used for the pin headers I cut down to size. The price of Cortex-M4 chips has gone up a lot in the last year or two and it alone accounts for the majority of the cost of components.

Remarks

At time of writing I have only done a smoke test of checking whether a programmer can communicate with the chip, as writing firmware to do things like serial communications is a long-term project. Figuring out RS232 and I2C on the NXP microcontrollers took several weeks whereas USB was on-off for about a year. For the time being the Cortex-M0 chips I have experience with are more than adequate for my needs and I only got the Cortex-M4 for when I need to write more demanding firmwares, but that time will come and I want to be prepared.