Incandescent light bulb inductance

Discussion in 'The Projects Forum' started by AnalogKid, Nov 10, 2015.

  1. AnalogKid

    Thread Starter Distinguished Member

    Aug 1, 2013
    4,538
    1,251
    A normal light dimmer using a TRIAC usually has some form of EMI suppression because of the current spike when the filament is turned on part way through each AC half-cycle. But if I use a pair of power MOSFETs I can turn the bulb on and off anywhere in the cycle. So I could turn the bulb on at the zero crossing and off at some phase delay point, the opposite of TRIAC operation. This should eliminate the turn-on current surge, but now could create an inductive kick at turn-off.

    How inductive is an incandescent light bulb that is on? When the filament is hot its resistance is high (relative to its off resistance), so if there is any inductance it should have a low Q. How low?

    ak
     
  2. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    If you just look at the filament, there cannot be much inductance there. I see a few turns of tungsten, about 0.2" in diameter, about 2" long.
    This calculates out to less than 1uH.

    fil.gif

    I would say "insignificant" at 60 or 120Hz, and even at the dV/dt as the switch turns on...
     
    Last edited: Nov 10, 2015
    AnalogKid likes this.
  3. alfacliff

    Well-Known Member

    Dec 13, 2013
    2,449
    428
    I dont know of any light bulbs with a spec for inductance. if they had any apreciabloe inductance at 50 or 60 hz, they would probably spec it.
     
  4. DickCappels

    Moderator

    Aug 21, 2008
    2,653
    632
    You would have to measure the particular bulb that interests you. Inductance depends upon the design of the filament, and they can vary widely.

    [​IMG] [​IMG][​IMG]
     
    #12 likes this.
  5. crutschow

    Expert

    Mar 14, 2008
    13,009
    3,233
    I would think that would make little difference in the EMI. A current surge at turn-off will generate about the same amount of EMI as a similar current-surge at turn-on.
    A step function has the same EMI generation (Fourier components) independent of whether it goes positive or negative.
     
    #12 likes this.
  6. ian field

    Distinguished Member

    Oct 27, 2012
    4,415
    784
    Some household light bulbs have "coiled-coil" filaments, so the inductance can mount up.

    Quite different to a car headlamp bulb that has a single spiral of fairly thick filament.
     
  7. AnalogKid

    Thread Starter Distinguished Member

    Aug 1, 2013
    4,538
    1,251
    True, but a light bulb filament does nto have constant properties throughout each half-cycle. According to other threads here and elsewhere (and personal experience), the cold turn-on current of a standard 60 W or 100 W incandescent light bulb is many times the average running current. 5x, 8x, 10x, whatever. So while the turn off current of a 120 W bulb would be 1 A, the turn on current is higher. And while it is not that great a difference after the bulb warms up, the filament is off for the first part of each half-cycle of a TRIAC-dimmed bulb, making for a small but noticeable current surge and thermal shock. So if my FET-based dimmer actually works, and I adjust it to turn on at 45 degrees and off at 135 degrees, the change in current at those two points will not be merely equal and opposite; the turn on delta should be larger.

    Much of the turn-on current spike comes from the fact that the voltage is not at a zero crossing at turn-on, so it is at least partly a rapid step function rather than an overshoot kind of spike. Still, my scope shows a burst of ringing that travels back and forth along the voltage sinewave as I adjust a dimmer, and ringing usually means inductance. Hmmm... Still chewing...

    ak
     
    cmartinez likes this.
  8. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    The "delay" is due to thermal-time constant, and negative tempco of the filament; not "inductance"!
     
  9. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,798
    1,103
    Even with a coiled-coil arrangement I would guess the filament inductance is only a fraction of the supply wiring inductance.
     
  10. ian field

    Distinguished Member

    Oct 27, 2012
    4,415
    784
    If I can still find any bulbs - I might get around to measuring it.
     
  11. wayneh

    Expert

    Sep 9, 2010
    12,123
    3,047
    Surely this has been done? See post #10 here. He reports 190nH for a 60W incandescent.

    See here for a spice model. It's also worth noting that lightbulbs can be used as dummy loads for VHF and UHF transmitters. That wouldn't work if they had much inductance.

    Here's another napkin analysis.
     
    AnalogKid likes this.
  12. crutschow

    Expert

    Mar 14, 2008
    13,009
    3,233
    I would think that, due to the long thermal time-constant of the bulb, the resistance value at the start of the half-cycle is not that different from the end value, at least not enough to make a significant difference in the EMI generated.
    But I could find no good data on that.
     
  13. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,388
    1,605
    Unless you are putting the dimmer inside the light bulb's base don't forget all the wiring between your controller and your controlled.
     
    nsaspook likes this.
  14. ian field

    Distinguished Member

    Oct 27, 2012
    4,415
    784
    The triac is in series with the bulb, only minimal circuitry is ground referenced like the CR bit for the diac. So unless you enclose the dimmer in screening around the lamp base and include mains in filtering - it doesn't make any difference where in circuit the dimmer is.
     
  15. eetech00

    Active Member

    Jun 8, 2013
    650
    112
    How about not turning it completely off? Maybe keep a small amount of current flowing just to keep the filament
    warm?
     
  16. cmartinez

    AAC Fanatic!

    Jan 17, 2007
    3,572
    2,537
    That can only be accomplished with a mosfet. A triac can't be shut off until it reaches 0V and requires a minimum amount of current to start/stay on
     
  17. AnalogKid

    Thread Starter Distinguished Member

    Aug 1, 2013
    4,538
    1,251
    Thanks for all of the responses. If I actually build this, I'll post the results. It turns out that a super-efficient or more controllable light dimmer based on power MOSFET(s) is not an original thought. ESP has a power MOSFET dimmer project, and there are several patents. So I'm brilliant, but late.

    ak
     
  18. Brownout

    Well-Known Member

    Jan 10, 2012
    2,375
    998
    The OP is using MOFSETs
     
  19. ian field

    Distinguished Member

    Oct 27, 2012
    4,415
    784
    Its easily accomplished by bypassing the triac with a bleed resistor.

    But you have to keep the bleed current low to avoid upsetting triggering close to the zero crossing points.
     
    cmartinez likes this.
Loading...