Announcement

Collapse
No announcement yet.

OT Question About Current Limiting

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    The diodes should prevent the problems I mentioned.
    Free software for calculating bolt circles and similar: Click Here

    Comment


    • #32
      Originally posted by Evan View Post
      The diodes should prevent the problems I mentioned.
      Ahh yes. Without the diodes, more than one e-match will fire with a single button press. Current can flow "backwards" through matches on other slats that have not been "turned on" without the diodes.

      Comment


      • #33
        For those who have been wrapping their brains around the problem of current limiting without large voltage losses, there's a method I've used several times to overcome this. It involves adding some voltage to the source voltage for the 'protection' transistor to reference to. Most power supplies I've built with a limiting feature have another winding added to the transformer to give this extra voltage. Comes to mind that a 555 can easily make this extra voltage, as long as the current draw on it is low.

        The idea is to be able to raise the emitter voltage of the protection transistor by up to about .7 volts higher than the supply voltage. In this condition the base will be forward biased even if no current is flowing into the load. As you reduce this extra voltage, the current sense resistor has to develop some voltage in order to make the protection transistor conduct and control the power transistor. You no longer need to develop .7 volts across the sense resistor (and lose that voltage to the load). You could set it up so you only needed say .2 volts across the sense resistor in order to get the current limiting action.

        In practice, you can get a limiting action with only .1 wasted volt. That means your sense resistor can be .01 ohms, and now it can be made from a fairly short piece of hookup wire or magnet wire, or a metal strip of fairly small dimensions.

        The best results overall are now going to be obtained by using a mosfet for the power transistor. Because it's driven by voltage and not current, the protection circuit can still have a fairly sharp action at the 'knee' point of the limiting, and the 555 circuit will easily be able to supply the extra operating voltage for it.

        Now I have one other thought- if you include the 555 circuit for the extra voltage, you would now have the ability to drive an N channel mosfet to a gate voltage of about 11v or so higher than the source pin. You would still be able to design in a good limiting action, and you'd be able to take advantage of the higher ruggedness of the N channel device over the P channel device, with its lower on resistance as well. It's entirely do-able to have limiting at 10 amps while losing less than a quarter of a volt.

        N channel mosfets are made with amazingly high current capabilities, and are fairly cheap. They can safely handle running a lot hotter than a bipolar transistor can, and this could make the whole package smaller. The entire circuit could be made on less than 2 sq inches of pc board.
        Last edited by darryl; 09-18-2012, 10:57 PM.
        I seldom do anything within the scope of logical reason and calculated cost/benefit, etc- I'm following my passion-

        Comment


        • #34
          You have to watch the listed current ratings of mosfets. While the die might be rated for 120 amps the package and especially the leads may not be. This is common in the TO220 package as the maximum rating for the leads is 50 amps regardless of heat sinking or chip rating.
          Free software for calculating bolt circles and similar: Click Here

          Comment


          • #35
            The sketch, such as it is, is here


            The left hand is the usual high side topology. With P-channel as shown, you have all "pull-down" drive, but with an "N" type, you are faced with a need for an even higher drive voltage.

            P channel are fine, but are not nearly as popular (at lower voltage, such as your 12V/24V they are more available as high side is hot for automotive apps) and are not available in as wide a variety. They have some technical problems as well, most of which won't apply here.

            The right hand is the "below ground" topology, generally an older one, but no worse for that. It allows all ground referenced stuff, NPN or N channel parts with no need for drive voltage over the positive supply, and still provides a standard "ground and plus" output. It is how many many laboratory power supplies have been made, and for the old metal case transistors, it even allowed the case to be at ground potential.

            Mosfets and current.....

            The lead's current rating is generally more than the bonding wire rating, OR the ability of the solder joint to carry the current without overheating. Plus, the usual applications are pulsed, so the true current is the rms value of the pulses, NOT the peak pulse current.

            In general you can NOT use the maximum listed current, not due to leads etc, but because the current is de-rated with temperature, and unless you have a liquid nitrogen cooling system, you can't keep the case (or the die) at the 25C that is the basis of the spec. It's generally nonsense, but may serve to point out some features of the part.

            With IGBTs it is much much worse..... you have more details that cut the current rating..... but the better manufacturers give the really useful specs farther down the page.

            For this application it will NOT matter, so long as a well over-rated part is used
            Last edited by J Tiers; 09-19-2012, 01:01 AM.
            CNC machines only go through the motions.

            Ideas expressed may be mine, or from anyone else in the universe.
            Not responsible for clerical errors. Or those made by lay people either.
            Number formats and units may be chosen at random depending on what day it is.
            I reserve the right to use a number system with any integer base without prior notice.
            Generalizations are understood to be "often" true, but not true in every case.

            Comment

            Working...
            X