Better snubber?

Discussion in 'General Electronics Chat' started by MikeML, Jun 22, 2014.

  1. MikeML

    Thread Starter AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    I am using a magnetic gate latch to hold a swing gate closed. It is rated at 24Vdc . It has a resistance of ~200 Ohms and a huge inductance. I am using it on 13Vdc, and it is taking about 3 sec before the gate can be pulled away from the latch after the current is interrupted.

    I am looking for a way to control the inductive spike as the magnetic gate latch is switched off and shorten the delay before the gate can pull away and begin moving. Here is an LTSpice sim of both the Original Snubber and my proposed fix:

    In the Original, only a Si Rectifier is inverse connected across the coil (Green traces for V(a) and I(L2)). The Better circuit replaces the Rectifier with a complex network consisting of the same Rectifier, a Capacitor, and two Resistors. Note the voltage V(b) and the current I(L2) which reverses slightly, but first gets to zero in about 1/4 of the time in the original circuit.

    Does anybody have a snubber circuit that would dissipate the energy in the latch magnet even faster while still keeping V(a) below Vce(max) for the switching transistor and while keeping the dV/dt on the cabling to a reasonable rate (so that I don't glitch the Arduino)?
     
  2. crutschow

    Expert

    Mar 14, 2008
    12,991
    3,227
    The "Best" (meaning the fastest current reduction) snubber is a zener diode in inverse series with the diode since it maintains the maximum reverse voltage until the inductor current stops. You pick a zener so its voltage plus the diode forward drop, plus the supply voltage is comfortably less than the transistor max Vce rating. With an 18V zener my simulation showed a turn-off time of 80ms (to less than 1ma inductor current). That compares with about 88ms that I measured for your "Better" circuit. So it's Best but not by much.

    Note that my simulation showed some high frequency oscillations after the turn-on which I snubbed with a 10k resistor across the zener diode.

    I don't understand the 3 sec it takes to release with just a diode snubber. :confused: The simulation shows only about a half second.
     
  3. MikeML

    Thread Starter AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    Thanks for the suggestions. I also simulated the diode+zener snubber, but the dV/dt at turnoff makes me queasy, since the gate motor, latch magnet and control signals are routed in an underground conduit for several tens of feet between the controller and the gate, and I am worried by inter-wiring crosstalk coupling stuff into other wires that go to inputs of an Arduino.
    The RC network seems to round off the dV/dt nicely.

    I have not measured the actual inductance of the latch magnet. I just went out and timed how long it takes to release the latch plate more accurately, and it actually takes 2s. The 30H I used as the inductance in the simulation was just a WAG! I guess that means that the inductance is actually higher...

    btw, I can see the huge inductance in action if I power the magnet with a metered lab supply. It takes > two seconds for the current in the magnet to become stable...
     
  4. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    I would do some empirical testing first, using a big'ol switch to just break the current to the electromagnet, and no snubbing.

    If it still takes 2 sec for the magnetic field to die down and open the latch (which it probably will) then no type of snubbing will fix that.

    If so, the only fix will be to hit it with a reverse current to force faster field decay.

    You could try powering the latch with a half-wave rectified DC instead of pure DC. That may stop the core saturation getting so bad when latched.
     
  5. MikeML

    Thread Starter AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    I thought about driving the Latch with a L298 H-bridge module so I could actively reverse the current.

    I doubt that it is core saturation because of the air gap (albeit small) between the latch and the "keeper" plate... Besides, I am utilizing the latch at half its design voltage...

    I effectively tried your switch idea with a clip-lead before installing it on the gate. It makes a nice arc as the clip-lead is disconnected. I did notice that it released the "keeper" plate much quicker than it does with the snubber diode in place, so I think I am on the right track.
     
  6. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,282
    1,233
    Not sure you can do any better. The voltage rises to Vce. You could use just a zener, but then the Dv/Dt will be high.

    Wow, a lot of posts after I stared to look at it.
     
    Last edited: Jun 22, 2014
  7. crutschow

    Expert

    Mar 14, 2008
    12,991
    3,227
    If the actual inductance is higher, than the peak voltage with your "Better " snubber circuit may exceed the transistor Vce (with 120H I measured 47V) so you would need to adjust the values accordingly.
     
  8. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,773
    1,103
    Couldn't sim any improvement on your 'better' snubber topology. It's a tradeoff between dI/dt and dV/dt.
    The 100uF electro sees voltage both ways. Will it survive?
     
  9. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    It's probably still saturation of the iron core itself. Some are meant for full wave rect DC with no cap, or even AC. Running from pure DC causes the problem.

    Try running it from full wave rect DC or half wave rect DC with no filter cap. I bet that will fix it.
     
  10. crutschow

    Expert

    Mar 14, 2008
    12,991
    3,227
    Good catch. Might want to place a diode across the cap (anode to V+) to limit the reverse cap voltage.
     
  11. crutschow

    Expert

    Mar 14, 2008
    12,991
    3,227
    Not clear to me what saturation would have to do with the delay in the release of the solenoid. :confused: All saturation does is limit the maximum magnetic flux from the solenoid. Are you perhaps thinking of magnetic hysteresis (which is not directly related to saturation)?
     
  12. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    Sorry I may have used the wrong word?

    The iron core or structure of the device gets magnetised when using pure DC and takes a couple of seconds for the magnetism to die down once power is removed.

    Running it from pulsed DC means the iron will not get as magnetised, so it will release quicker. I saw this many years ago in industry with some "DC" solenoids that were actually designed to be run from rectified AC, with no filter cap.

    That "2 seconds" symptom sounded familiar. If the latch holds good enough from full or half wave rect DC then I'm sure it will release quite fast compared to releasing when using pure DC.
     
  13. Brownout

    Well-Known Member

    Jan 10, 2012
    2,375
    998
    Hysterisis?
     
  14. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    Magnetic buildup in the iron parts? Core magnetising effect? It's a slow phenomenon.

    The problem is that the iron core parts get slowly magnetised from a constant field strength and acts like a poor permanent magnet for a short while. If you use pulsed DC the magnetic field strength is constantly going up and down when latched, so the little iron particles don't get that same magnetised effect, or only to a very small degree in comparison.
     
  15. Brownout

    Well-Known Member

    Jan 10, 2012
    2,375
    998
    We call it magnetic hystersis. It's what makes the B-H curve a squar-ish loop rather than a line.
     
    THE_RB likes this.
  16. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    So if I wrap some wire around a nail, and the nail remains slightly magnetised for a while after the current is stopped, that is called magnetic hysteresis?

    Cool, thanks I'll remember that in future.
     
  17. MrAl

    Well-Known Member

    Jun 17, 2014
    2,421
    488
    Hi, The core gets 'magnetized' and when the current is reduced to zero it stays magnetized to some degree, and that magnetization is called the residual magnetism. When the core is taken through a positive and negative cycle, we see a hysteresis loop.

    Some ideas...
    1. Use a high voltage transistor and a zener, use a capacitor to damp any line spikes, not the electromagnet spikes.
    2. Use a resistor instead of a zener. Have to experiment a little to get the right value and power rating.
    3. Most latching electromagnetic circuits work partly on proximity to the secondary piece as well as current. The closer the secondary piece is to the core pole face the stronger the pull in force, and the stronger the pull away force. Insert a piece of insulating electrical paper or mylar between the pule face and secondary piece and the forces are reduced. That's if the application can take a reduction in force and still work decently.
    4. Use the normal circuit for the pull in, then switch to a circuit with a resistor in series with the coil after a few seconds. This will reduce the current in the coil and thus make turning it off faster. Again this is if the application can take a reduction in normal operating force and still work acceptably.
    5. Use a counter force. A strong solenoid will push it open regardless of the current in the latch coil. Reduce the current, energize the release solenoid.
     
    Last edited: Jun 28, 2014
Loading...