Isolated Power MosFet from PIC ?

Discussion in 'Embedded Systems and Microcontrollers' started by rsisson, Nov 2, 2007.

  1. rsisson

    Thread Starter New Member

    Nov 2, 2007
    I am working on a Neighborhood Electric vehicle...

    Lots of issues, but the big one is the lack of a robust affordable controller.

    All of the controllers in the "DIY" or "Hobby" environment use medium power Mosfets in a multiple Bridge format depending on if the are Brushed or brushless motors and the number of phases.

    The Problem I am looking at at the moment is going from a packaged controller that run on 5.6-18v and getting it to drive power MosFets on 50+V rails.

    Controlling the low-side of the bridge is easy, but the high side needs additional devices and "boost" circuitry.

    I was wondering if optical isolators could eliminate all of the extra "stuff" as it could be connected almost directly between the rail and the gate with minimal other "Stuff" due to the isolation. Adding a similar isolator for the bottom side would allow you to "Separate" the power section and the logic section completely, possibly putting them on separate boards...

    How practical is this?

    For a controller I was looking at

    I only see a handful of components, and nothing that requires special shielding, coil winding, or surface mounting.

    The standard MosFet in most controllers is a

    If I combined a controller like the above, and a few indestructible MosFets like

    and Optically Isolate them on their on sub-assembly, how hard would it to build a affordable "robust" controller that could take the the mistakes that hobbyist throw at them... even if I do fry a MosFet, the control circuitry will be safe....

    Looking for some guidance... Once I get close I was going to use one of the on-line board designers and build a to be cheaper than frying multiple controllers at $250 each.

    Bob Sisson
  2. beenthere

    Retired Moderator

    Apr 20, 2004
    It's hard to be sure of an answer with no schematic to work from, but the opto-isolation scheme shouldn't be too mind-boggling to come up with. Just look for drive capability to see if you need to buffer the output to get a better switching time.

    High side drivers are a bit more critical. The FET or IGBT needs to have the gate at 0 volts with respect to the source to be turned off, or within 20 volts of the source for turn on. When the load causes the source to change potential, using only some divider scheme that is referenced to the supply rail may allow the gate to go beyond the 20 volt limit and lose all its factory smoke.

    Linear Technology, Maxim and IRF all have lines of high side drivers. Using one should save some grief. If you use some of those humongo IGBT's with 600 volt and 350 watt ratings, it would need a serious goof to fry one.
  3. shenta

    New Member

    Dec 2, 2008
    Your choice of IC are used in most e-bike from China, you do not need an opto to isolate, there are very little noise even under high load, as long as you design your ground plane properly. Be careful about your mosfet driver, you will blow up alot of mosfet before you can get the right driver (Speaking from experience). Also current limit is a big problem, during Motor start, Motor will take app. 25x of current, if your current limit is not fast enough, Mosfet will get fried again and again.

    As for your choice of Mosfet, even tho you have 130A mosfet, in our experience, you can only do app 30A max of 20Khz switching (and it get really hot even with big heat sink ), if more than that, it blow up again. You can try to parallel more Mosfet (the very max we tried and success is 6 on high side and another 6 on low side), but when 1 of the 6 start to run away with themal, the rest is fried also.

    Let us know your progress, and if you meet problem, I am happy to help out.

  4. eblc1388

    Senior Member

    Nov 28, 2008
    Isn't MOSFET Rds has positive temperature coefficient different to Vce of bi-polar transistors? If current in one increases, it heats up and Rds increases and then current reduces or shifted to the remaining MOSFETs.

    Why would thermal runaway happen on one of the parallel MOSFETs if they are mounted on properly designed heatsink?
  5. John Luciani

    Active Member

    Apr 3, 2007
    Using optical isolators will work provided that you have sufficient current to drive
    the gate. The longer the transition from off to on takes the more power you
    dissipate in the FET. The thermal time constant of the die is in mS so it does not
    take long. You can get an idea of the time constants from the transient thermal response
    curves. To calculate the current required use the formula Q = t * I where Q is the
    total gate charge (datasheet) and t is the desired time.

    You will need an opto, possible a gate driver and an isolated supply. I would check
    the high-side driver ICs first since they were designed specifically for this type of

    (* jcl *)
  6. John Luciani

    Active Member

    Apr 3, 2007
    Properly mounted devices on a heatsink will have the same thermal resistance from junction to ambient (Rja) but may not dissipate the same amount power due to
    mismatches in Rds(on).

    In switching applications this is usually not a problem since you have sufficient gate drive to turn all of the FETs completely on. In linear applications (i.e. electronic loads) this can be a big problem especially if you do not provide an independent control loop for each FET.

    (* jcl *)