Motor power board preliminary discussions

Specifications

  • Compliant with RE25 and RE40 motors
  • 24V (RE25 or RE40) power supply
  • 15A current limitation
  • High PWM precision (1/255)
  • size : 5x8 cm

Conception

  • To follow 15A, H-Bridge made with discrete components in S08 package to fit into the board
  • To get a high precision PWM and to limitate power dissipation, use of low transistors with a low Rds and transition time under 200ns

  • Use of an integrated MOS driver :

    • Infineon TLE6284G : full-bridge, all integrated, PWM/DIR command, too slow (300ns), hardware current limitation but at 40-80A with a 12mOhm Rds MOS
    • Fairchild FAN5109 : half-bridge, shoot-trough protection
    • Maxim MAX15012 : half-bridge, best performance !
    • Maxim MAX5063 : idem previous. Seems to be the best compromise, but... unavailable by Farnell
    • Maxim MAX5064 : idem previous with programmable dead-time, but in QFN package (no handcraft soldering)
    • IRF IR2010 : quick half-bridge acceptable, but a bit slower than MAX5063. Will heat a bit more
    • Intersil HIP4080A : full-bridge, programmable dead-time, but a bit heavily complicated

  • MOSFET used :

    • SO-8 MOSFETS are used. They are a good compromise between low Rdson, small size, and ability to use un a handcraft process
    • The best compromise between price and Rdson by Farnell is the PHK24LQ04 (Philips). It has a 5mOhm Rdson and costs 1,4 euro when ordered by more than 10 units

  • Current limitation. The limitation is done by the following architecture :

    • A shunt resistor is in serie with each leg of each H-bridge
    • A low pass filter cleans a bit the voltage accross this resistor
    • A reference voltage is generated by the AVR. This voltage is proportionnal to the amount of current allowed to flow in the bridge leg
    • A comparator checks that the shunt voltage is lower than the reference. If not, the comparator put the output low, generating an interrupt to the AVR
    • When receiving this interrupt, the AVR has to release all MOSFETs of the concerned bridge until the next PWM cycle starts again.

  • Power and thermal considerations :

    • Spice simulations has been done with an equivalent and have shown that the power dissipated is 1,2W maximum for a full switching MOS, at 15A. The simulator is LTSpice, it is freely distributed by Linear Technology (www.linear.com)
    • Thermal simulation using the power datas given by Spice simulation has been done with Vishay's thermal simulator. It has shown that the maximum PCB temperature was 80° to 90°C in steady-state. It is shown that even by this very high temperature, the maximum allowable die temperature of MOSFETs (150°C) was not reached.
    • It has been decided to implement thermal limitation on the board. The max PCB temperature will be 70°C wich allow a margin for MOSFETs and a security for user (a 70°C PCB is hot enough for causing injury, we don't need more hot !).
    • The sensors are places the closest of the MOSFET's area. There is one sensor for each channel. The AVR measures the temperature through ADC. The ADC resolution is 20mV (8bits, 256 dots, Vref=5V), whitch correspond roughly to 1°C for temperature sensor's output.
    • The software shall measure temperature regularly and limit the current proportionally to temperature according to the following rule :
    • under 60°C : limit = 15A (max absolute rating of the board)
    • between 60°C and 70°C : linear decreasing of the current
    • over 70°C : limit = 3A (normally, the power at 3A is 1/25 of the power at 15A. It should really cold down the board)
    • over 75°C : total shutdown the channel until the temperature falls down to 70°C (the channel is believed to be in hardware default)

  • TODO :

    • input logic which shall be generated by the AVR

Manufacturing & tests

  • Manufacturer is Olimex. Technology is 2-sides board.
  • Tests can be done in Kiki's work laboratory