Actions
Forum Thread Details
  • State Suggested Answer
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 80 subscribers
  • Views 911 views
  • Users 0 members are here
Related

Parallel, redundant, Mosfets on DC motor control board

Catwell
imageimage
 
I was having a issue with some CNC machinery of mine, and I had to take apart the factory power systems in each. I noticed that the powering scheme has several paralleled mosfets. One of the mosfets, a D8020LD8020L, is more than enough for powering in the motor. But there were 5 in parallel. Is this a common practice for driving DC motors in machinery?
 
The above, blurry, pictures are from the machine's DC motor drive board.
 
Cabe
  • 0

    Hi Gabe,

     

    This issue here is reliability and thermal management. Let's start with thermal management, a spec sheet might say that a transistor is good for 5A or 25W of heat dissipation, however if you look carefully in the spec sheet your going to find a derating curve, which shows that as tempature increases the amount of power that can be dissipated decreases. Since these transistors are going to be running CNC motors and are inside an enclosure that probably doesn't have the most abundant air flow, the designers of the controller derated the transistors accordingly and gave a some margin for reliable operation(assuming good design practice).

     

    The general rule of thumb for reliability is that tempature and reliability tend to follow an inverse square relationship. If your tempature doubles then you can expect the product lifetime to be cut by a quarter. By paralleling the FETs each transistor handles only a portion of total current which yields a lower temperature rise, which in turn yields an increase in reliability minimizing the number of failures in the field.

     

    As an aside MOSFETs are very good for parallel applications, their on resistance has a positive temperature coefficient. What this means is that if one of the mosfets starts drawing more current, its temperature will increase. The increase in temperature will yield an increase in the on resistance, which will in turn lower the amount of current drawn by the transistor. Such behaviour acts as form of feedback to share the load current equally among FETs.

     

    Hope this answers your question.

     

    Best Regards,

    Jorge Garcia

  • 0

    Hi Cabe,

    There must be a reason for using 5 devices in parallel.
    If the motor is not running, the impedance is very low and the start current might be very high.
    As a result of the inductance of the motor, the peak drain Voltage might be higher than the supply voltage.
    To operate the FET’s within the SOA (Safe Operating Area) the current per device must be derated.
    (See the SOA curve in the datasheet of the fet.)
    Another reason could be the dissipation in worst case conditions. The designer can use multiple standard fet and heatsink modules or heatsinks with different sizes. The larger size heatsinks cannot be mounted on a PCB because they are too heavy.
    Mosfets can easily be used in parallel because they have a positive temperature coefficient. The hottest device will draw the lowest current. This way the circuit will stabilize itself.
    Bipolar transistors  have a negative temperature coefficient, causing the hottest device to draw the highest current.
    This could lead to thermal runaway and failure of the device with the lowest resistance.
    When transistors are used in parallel, resistors are necessary for equal distribution of the current over the devices.
    A disadvantage of using multiple fets in parallel is the larger drain – gate capacity, that requires a higher drive current. Replacing a single fet by multiple fets in parallel might lead to instability and oscillations unless a separate drive circuit is used for each fet.
  • 0

    Hello,

     

    one important differentiation we need to make here is whether it is a linear or switch-mode drive. Today, and especially in this power range, pretty much all drive circuits are switch-mode. The motors being used are brushless DC motors in many cases, for which a "frequency inverter" is needed. The minimum number of phases for an electrical motor to effectively determine rotational direction is 3, so you should be seeing three half bridges, or 6 sets of switches in the inverter.

    For lower-voltage motors (think forklifts), the currents are very high for the same power, as the voltage is lower. So you'll need more current-handling capability in the switches. Moreover, the worst case condition for the current is determined by the mechanical parts the motor is driving, e.g. when it is blocking, or at startup. Third, the time until any overcurrent protectio ncircuit can react needs to be considered as well. These factors lead us to rather large design margins that need to be implemented, hence the paralleling of so many MOSFETs. In fact, in forklifts it is not uncommon to see 30-40 MOSFETs in parallel.

    Key to successful paralleling in linear applications are given in the previous posts. For switching applications, some of this applies as well. The most important factor here is a "good enough" match of the threshold voltage of the MOSFETs. Imagine at turn on: The MOSFET with the lowest threshold voltage will turn on first, and as the gate voltage continues to rise the other MOSFETs will follow. However, that first MOSFET will see high currents, potentially leading to destruction.

    On the temperature coefficient, yes MOSFETs have a positive tempco if operated above a certain current level (which is almost always the case in switching power applications). If you look at the diagram of drain current versus gate voltage for different temperatures, you will see all these lines cross in one point, where the temperature coefficient is literally zero.