Gooser3

After the failure of GooserCS - OSHWLab I decided to start over with a two-motor version using 6PWM on FD6288Q like Lepton. In addition to reducing the number of potential problems, it also gives more procesing power per motor. After settings up some motors on B-G431B-ESC1, I found that the FPU and higher clock rate does not result in as much speed increase as I expected over Lepton's G031, so four motors would have been a bit RPM-limited anyway.

While this is designed as a dual-motor driver, it is tileable for more motors. The I2C communication lines and power lines run across the width of the board so you can copy-paste the whole thing for as many motors as you like on a single board. Each pair of motors adds another 20mm of width. Or you can make the whole board 20mm wide if you supply power to each motor's rails separately. Or cut off everything south of the programming connector as well to make it a single-motor board. Or paste two of those side-by-side if you want one CPU per motor.

Changes from GooserCS:
1. Full 3-channel current sense instead of just two per motor.
2. CPU is moved to the back side of the board due to lack of space, increasing the difficulty of finishing it yourself if starting from single-sided assembly by JLC.
3. Driver voltage can optionally be supplied via the programming connector, enabling operation up to the 30V maximum of the mosfets.
4. Changed to a smaller 3.3V regulator. The higher voltage rating of the larger one wasn't useful because it was thermally limited by the relatively high current of all the sensors, so it required a separate input voltage from the driver anyway.

With these changes you can use up to 7S lipo, and tap into the first and third cells via the balance connector for the low-voltage and driver voltage inputs.

Interestingly, the main input voltage has no lower limit now. You could run on 1S lipo with an external boost converter for the driver voltage. Or even lower with a boost converter for the 3V regulator input too.

Going by Valentine's recommendation of >1000uF for Lepton this should have >2000uF of bulk capacitance. There are various holes in the power rails which can be used for soldering through-hole capacitors and/or as position pins if soldering copper bars on the back. Or if you don't use copper bars, you can solder SMD electrolytics on the back. With or without copper bars, you can solder a bunch of 1206 capacitors on the back, but not enough for the full 2000uF.

Motor 0:
TIM8 ch1 PB6, ch2 PB8, ch3 PB9, ch1n PB3, ch2n PB4, ch3n PB1
Current sense ADC1 ch2 PA1, ch1 PA0, ch10 PF0
Hall encoder ADC2 ch5 PC4, ch4 PA7

Motor 1:
TIM1 ch1 PA8, ch2 PA9, ch3 PA10, ch1n PB13, ch2n PA12, ch3n PB15
Current sense ADC2 ch17 PA4, ch13 PA5, ch3 PA6
Hall encoder ADC1 ch11 PB12, ch14 PB11

SWD: io PA13, clk PA14
UART3: tx PC10, rx PC11
I2C1: sda PB7, scl PA15
SPI3: mosi PB5, miso pc11, sck PC10, nss PA15
Note: SPI is not entirely usable since its pins overlap with UART and I2C. But PB7 (SDA) is UART1_RX and PC4 (encoder 0A) is UART1_TX, so that can enable UART and SPI together.

There are a few other pins broken out as solder pads on the back of the board, but they don't have functions that combine particularly well with any others. Just whichever pins had enough space nearby to fit a pad after all critical routing was complete.

The aux connector is primarily for joining the I2C and power lines between two boards stacked on top of eachother, but may be useful for other purposes as well. The 3V hole should typically not be connected between boards. Each board has its own 3.3V regulator, and only the inputs should be joined. However it is possible to bypass the onboard regulator and supply 3.3V directly via this hole, which can be convenient if you have an external 3.3V supply but nothing in the 3.6-6V range for the regulator input.