BLDC driver v6

Powerful BRUSHLESS Motor Driver version 6 is a small brushless motor driver intended to drive motors up to 300W, a great fit for many robotics projects.

Product Highlights:

• Input voltage: 7-25V (30V surge) 
• Continuous Output Current without heatsink: (TBC) 20A (Peak), 14A (RMS)
• Connectors for quick and easy wiring, no soldering required.
  Power: XT30, Motor: MR30, CAN: GH-2P, I2C: GH-4P, DEBUG: SH-4P.
• Small size: L: 58mm, W: 26mm, T: 8.5mm.
• Very low power dissipation, less than 7W (TBC) power dissipation when operating at 20A peak motor current (14A RMS).
• RDS(on) based current sensing with temperature compensation for high efficiency.
• Very low idle power, less than 0.5W with motor holding position under no load.
• I2C (1Mbps) and CAN bus (8Mbps with SIC) is available for communication with the main controller.
• Up to 1A at 3.3V or 5V (independantly selectable by software) can be sourced from the I2C and sensor ports to power external circuitry.
• Two M2 screw holes (max screw head diameter 4mm) are available for mounting.
• Maximum PWM frequency: 50KHz
• Maximum Motor Speed: 160000 ERPM
• Supported Encoders: Sin/Cos Encoder (can be implemented using linear hall sensors, or magnetic encoder chip such as AS5115)
• Supported Motors: Robomaster M2006 P36, Robomaster M3508 P19, more motor and encoder combinations coming soon
• PCB specs: 14 layer, 2oz copper on all layers, 2.0mm thickness, epoxy filled vias, min via hole 0.2mm, components on both sides.

Issues noticed:

• There was a small oversight when designing the TCXO circuit and the output biasing network was missed in version 6.0 and causes the TCXO to be unusable. Version 6.1 should fix this issue. (Note: Version 6.1 has not been manufatured or tested yet.)
• Any further issues noticed will be posted here when known.

Comparison with Powerful BRUSHLESS Motor Driver version 5:

The new version has been redesigned for better ease of use, with the return of the screw holes for mounting using M2 screws. In addition, the PCB width has been reduced while slightly increasing the thickness. The PCB length has increased, but in reality, the space of the connectors takes up was neglected in the previous design. Once you account for the connectors, the new version takes up less space than the previous version due to the connectors being packed closer together.

The PCB surface area has been increased slightly by fully utilising the space under the connectors, and combined with the thicker 14 layer PCB and upgraded MOSFETs, leads to noticeably improved thermal performance.

The current sensing accuracy should be improved as the DRV8353 should not have offset calibration issues seen in the DRV8323 (TBC). Additionally, the current sensing circuit PCB layout has been redesigned for higher accuracy, and the thicker PCB leads to better thermal coupling, for more accurate temperature compensation of RDSon. The current sensing and gate drive performance at low input voltages has also been improved as the MOSFET gate voltage is regulated more accurately.

The idle power consumption has also been significantly reduced by a new highly efficient power supply architecture providing 10V to the gate driver and 1.8V to the microcontroller. The full load efficiency is also slightly increased due to the thicker PCB having lower resistance, new MOSFETs, and better thermals leading to MOSFETs running cooler and more efficiently.

A new high current software switchable voltage output is available on the I2C and sensor ports, which can output either 3.3V 1A or 5V 1A. This lets you provide power to much larger devices such as a Raspberry Pi without using an external regulator, improving convenience and saving more space in many applications. The 3.3V option is still available for powering microcontrollers such as ESP32, leading to more flexibility in use case. The outputs also feature reverse current blocking, which allows multiple outputs to be put in parallel safely, and is less likely than the previous version to suffer permanent damage in case of over load.

Manufacturing Notes:

The blind slot depth is approximately 0.8mm and should hit the center of the Inner5 layer which is blank. The sections above this layer with the text "BLIND SLOT" must not be removed, otherwise it will cause large board thickness variations. The blind slots are needed so the connectors do not collide with the PCB. If you skip the blind slots to save costs, you can sand the connectors manually so they do not collide with the PCB.

The 14 layer 2oz 2.0mm stackup JLC142022-106 is actually 2.6mm. This will cause issues soldering the MR30 connector as the lead length is 2.0mm, also will cause via aspect ratio to be too high, and cause board thickness to exceed the advertised thickness. To fix it, you need to slightly modify the stackup by replacing 3 layers of 106 prepreg in the stackup with 2 layers instead (in total removing 7 layers of 106 prepreg). This will reduce the board thickness to 2.2mm, which is the thinnest possible.

When ordering you must use a panel, and make sure that the edge rails and space between the boards is completely filled with copper. The reason is to ensure balanced copper density  and reduce board thickness variations. The provided PCB panel template has some test features on the edge rails which intentionally violate some DFM rules.

JLCPCB SMT Assembly seems to have trouble achieving proper barrel fill for the XT30 and MR30 connectors. See Version 5 for sample images. When ordering please leave a remark that the connectors in wave soldering need higher temperature than normal otherwise they may not be soldered properly. The poor barrel fill will cause the high current connectors to not make a proper connection to the inner layers.

In case manual rework is required for this issue, the method involves preheating the board to slightly below the solder melting point, and using a soldering iron to rework the connector pins. You should also apply flux to both sides of the board near the connector pins. It is recommended to preheat the board from the bottom side with a hot air gun because the plastic connectors on the top side are easily deformed when exposed to high temperatures. I found that hot air at 270C and the soldering iron at 370C works best, it takes around 30-60 seconds for the solder to flow properly.