I am building a street lamp controller that should be able to set a street lamp to on, off, or dim state. I have successfully built an incandescent lamp dimmer (phase dimming - using zero cross detector, AVR microcontroller, and controllable TRIAC driver). Now I am interested in building a similar thing for street lamps.

From my reading most street lamps still use HID lamps (metal halide, mercury, high pressure sodium) and they are not purely resistive so that some people in other forums state that they cannot be dimmed using phase dimming method. For example see http://www.edaboard.com/thread115980.html

My question is how should these kind of lamps (metal halide, high pressure sodium) be dimmed. This seems possible since I have seen such dimmer products in the Alibaba store but the manufacturer cannot tell me their method. What waveform should be expect?

Update 26/3:

  1. Based on NEMA standard, HID lamps can be dimmed: step level or continuous (phase control, high frequency shifting). The limitation is: power level and RMS voltage: cannot below specified value (determined by ANSI or lamp manufacturer)
  2. Phase control is somehow unsuitable (leading phase control may introduce flickering while lagging phase is dangerous for inductive load such as HID ballasts)
  3. High frequency shifting is a method that is so far, according to NEMA, seems safe enough for dimming.
  4. So far, the recommendation is to use step level switching (be aware of minimum power requirement) and, most importantly, consult lamp manufacturer

My question is, according to Russell's link here, HID lamps is operated at low frequency to avoid acoustic resonance (may ended in damage or lamp explosion). So, if I apply high frequency shifting on the ballast input, this will not affect the low frequency output of the ballast. Is this true?

  • 3
    \$\begingroup\$ I've heard tales from engineers who spent a while trying to make an after-market HID dimmer. Their conclusion was that it physically can't be done. Their story is that the arc approximates a constant-power load, and draws more current if you try to starve it in voltage. That, and the fact that gas flow is somewhat chaotic, makes them very difficult to control. But that's not first-hand experience! \$\endgroup\$ Mar 25, 2013 at 11:26
  • \$\begingroup\$ Nothing on dimming but some good commnents on HID generally in this link from ?Tony? \$\endgroup\$
    – Russell McMahon
    Mar 25, 2013 at 16:06
  • \$\begingroup\$ Well, so far I can only imagine 2 possible waveform modification: 1. Phase control, which is forbidden according to Osram's guide here 2. High frequency switching on/off, which is according to Russell's link is not feasible because at high frequency the lamp will most likely ended in lamp damage/explosion. \$\endgroup\$
    – bonchenko
    Mar 26, 2013 at 1:22
  • 1
    \$\begingroup\$ @HikeOnPast - High/Low beams are more complicated then just changing the brightness. It also changes the pattern of radiated light. If you had high-beams that were simply dimmed, they would still be high-beams, and a pain-in-the-ass for other drivers. The mechanical complexity is not from the fact that they couldn't be dimmed, it's from the fact that they need to change the light radiation pattern for the different modes. \$\endgroup\$ May 22, 2013 at 5:59
  • 1
    \$\begingroup\$ Why are we suddenly moving from street lamps to car headlights? \$\endgroup\$ Dec 14, 2013 at 3:36

2 Answers 2


As I understand it, dimming HID / High Pressure Mercury / High Pressure Sodium / Metal-Halide is like dimming flourescant, in that a special ballast does the work and you simply provide a control signal such as 1-10v, DALI or DSI.

Google "dimmable hid 1-10v" and you should find some examples.


I'd say just use PWM. Use a 555 timer to generate the PWM signal, then use a solid-state relay or some similar device to control the line going into the lamp.

  • 5
    \$\begingroup\$ Go look into how HID bulbs work. They can NOT be simply pulse-width modulated. \$\endgroup\$ May 22, 2013 at 5:59
  • \$\begingroup\$ @Connor Wolf Well, actually, if you passed it through RC first, you'd be fine. \$\endgroup\$ May 22, 2013 at 22:43
  • \$\begingroup\$ Alright, it was just an idea. It wouldn't have broken anything either way. So you didn't have to be an ass about it. \$\endgroup\$ May 23, 2013 at 22:58
  • \$\begingroup\$ Uh, you're really not helping your case. HID bulbs have complex electronic ballasts, and it very well could damage the bulb by trying to run it on a consistent under-voltage, or cycling the power very rapidly. And yes, I did have to be an "ass" about it, after the second response that showed you had still done no research, or even bothered to read the other comments. \$\endgroup\$ May 24, 2013 at 3:17
  • \$\begingroup\$ Actually, I did do the research, and well, nothing really would have been actually damaged if the bulbs had been under-volted. If you built your own custom ballast and just purchased the bulb itself, you could pretty easily have your custom ballast allow PWM to be used for dimming the bulb. You really don't have to be condescending if you disagree though, you could simply say that you disagreed and why you disagree. \$\endgroup\$ May 24, 2013 at 5:34

Not the answer you're looking for? Browse other questions tagged or ask your own question.