No announcement yet.

Does dimming lights really save energy?

  • Filter
  • Time
  • Show
Clear All
new posts

    Does dimming lights really save energy?

    Does dimming a light with a dimmer switch (X10 or otherwise) really save any energy? Since the dimmer gets pretty warm when the light is dimmed, I'm wondering if it really saves energy or if it just converts electrical energy that would have been converted to light to heat energy instead.

    Yes, it does save some. The dimmer works by delaying the turning-on of the triac until well into the 60 Hz cycle, so no current is drawn through part of the cycle (this also causes the buzzing). But the triac does disipate heat, I don't know how much, but would guess it's no more that 10 watts. I also think that it disipates more when the light is set brighter.

    So if you dim more than about 10 watts below full brightness, you get a net saving. If you run at full brightness, you're losing due to the heat generated by the dimmer.


      So if I operate a 100 watt bulb at 20%, I'm saving quite a bit of energy? Maybe even getting close to the equivalent of a 20 watt bulb?

      I installed X10 dimmers on the lights outside the doors of my shop. I like the effect at 20%. It's just a soft light that enhances the landscaping lighting. So, I think I'll turn them on (20%) at dusk and leave them on until the rest of the landscape lights go off (1:00 AM). But, there are 3 of them so I didn't want to do that if they were really using 100 watts each or more! Thanks for the info.


        Using rocco's figure, a 100W bulb dimmed to 20% would be (100W X .2) = 80W savings less 10W heat loss = 70W or 30% overal savings.

        I use delayed events to turn them lights off when motion detectors have not detected motion for extended periods of time (careful, someone could be sitting still reading) for even more savings.

        Jim Doolittle
        Jim Doolittle

        My Twitter
        My Hardware & Software


          If dimming to 20% means the light is only using 20 watts, and the dimmer itself is using 10 watts to convert to heat energy, then the total consumption is 30 watts for a 70% savings.

          I am going to use motion detectors also. When motion is detected I'll brighten them to 100%. A few minutes after motion ceases, they will dim back to 20%.


            Oops. My intent was not to show my math proficiency (or lack thereof) but to key on rocco's comment about losses from the dimming process itself.

            I wanted to save even more electricity by using 11W fluorescent replacement bulbs but then I would end up spending big bucks on inductive X10 switches to control them [img]/infopop/emoticons/icon_smile.gif[/img].

            This discussion makes me think of modifying my lighting strategy to use motion sensors to begin dimming lights in rooms where motion is not detected for some time. Something like "dim light 10% every 10 minutes of no motion and turn off when dim level reaches 40%".

            Jim Doolittle
            Jim Doolittle

            My Twitter
            My Hardware & Software


              Okay, I've been thinking. I want to switch to 2-way light switches but would also like to use compact florescent bulbs. I can live without dimming capability for those lights. Any other considerations?

              // AJ

              My Links: Website | HomeSeer Status | System Description


                ... turning off lamps in rooms where there is no motion is not an option for me. For one, I don't have motion detectors in every room. But more importantly, my wife's idea of lighting automation is "all lamps on whenever we wake up or arrive home" (yes ALL lamps in the house on, day or night). She likes the house "bright and cheery" (matches her personality). For her, lighting is not just for practical reasons like "so you can see", it's a big part of the decor. We have a lot of lamps (3-5 in every room) but at least she uses very low wattage bulbs (20-40 watts).


                  because light bulbs are non-linear. A 100 watt bulb may draw 100 watts at 100% brightness, but will probably draw more than 20 watts when dimmed to 20%. It doesn't affect any of your reasoning, I just don't want the accountants on my case. [img]/infopop/emoticons/icon_razz.gif[/img]


                    It's always nice to save energy, but from the cost perspective it really doesn't do much:

                    Based on a cost of $.07 per kwh:
                    If you use a 100 watt bulb, 6 hours per day, 300 days per year, it will cost $12.60 per year. So even if you save 80% of that by some dimming tricks, that's only saving $10 per year. You will need at least a few years to make up for the cost of the switch.

                    Of course there are other factors, like the heat generated makes your AC work harder, costing more. I think if you're really trying to have a green house, it's best to invest in smart irrigation and HVAC control. Having said that, if you're already a HA nut, it's nice to have a little justification!


                      So I conducted a little experiment this morning using a Kill-A-Watt power consumption meter, an LM-465 lamp module, a small table lamp and several light bulbs.
                      The bulbs were: Philips 200W, Philips 100W, GE 60W, GE 40W and a Philips Marathon 23W dimmable CF (all brand new out of the box).
                      The power meter was placed in series between the mains and the test lamp, so any light from the bulbs or heat by-product from the LM-465 was considered "spent" electrical energy.
                      Each bulb was run through from 100% down to 10% in ten percent increments from the HS device console with an "off" command in between each dim level (to eliminate any MR26A or palmpad "dimming" errors). Each test was performed three times to reduce any repeatability errors. The data you see here is the average of the three runs. There were no "oddball" numbers from the original data, so I'm guessing the repeatability is pretty accurate.
                      The measurements were read from the power meter as "instantaneous" readings a few seconds after the meter settled at each dim level.
                      Oh, and mains voltage and frequency were 116VAC and 59.9Hz.

                      200W 100W 60W 40W 23W(CF)
                      100% 182 92 56 36 22
                      90% 179 89 55 35 21
                      80% 169 85 52 33 20
                      70% 143 73 44 28 19
                      60% 124 62 38 24 16
                      50% 101 52 32 20 13
                      40% 80 40 25 16 9
                      30% 55 28 17 11 7
                      20% 25 12 8 6 6
                      10% 10 6 3 2 <1

                      From this data, the answer to DC's original question appears to be - yes, dimming saves electricity even though some of that energy is converted to heat.

                      I've got a 2-way SwitchLinc 2380 staring at me just begging to be wired up to an extension cord to do this whole thing again. Way too much coffee this morning...
                      Attached Files


                        I have been playing with an oscilloscope and some modules, in the course of modifying some Lamplinc units to allow me to use them in the UK. I also tested an LM12U (basic dimmer unit, no preset dim etc).
                        I see you never get the full wattage for any of the lamps you used. This ties in with what I have seen, where the X10 module never presents the full sinewave to the lamp when at 100% (full on) setting, instead it chops the first 20-30 degrees (1-1.2 milli seconds of the half cycle). The LM12 also never shut it off completly when taken to 0%, again it left the last 1 msec of the cycle presented to the lamp. This might help explain the situation where dimming to 'off' still uses a bit of energy 'wasted' though the lamp. The Lamplinc devices I have don't do this though, they shut off the Triac completely.



                          I'm going to graph it to get an idea of the linearity.

                          Another good test would be to do the same bulbs both with a lamp module (at 100%) and without. This would get some idea of the loss (due to heat) in the triac. But if the lamp module doesn't really go to 100%, it may not be a valid test.

                          Thanks, it's nice having someone to keep things in perspective.


                            This time around I used the exact same setup except a 2-way SwitchLinc 2380 took the place of the LM-465. The data was similar in all cases except the 23W dimmable CF. That bulb responded to an "ON" - but once it was on it did not respond to any other X10 command. I know SwitchLincs respond well to inductive loads (fans & motors), but there might be a safety mechanism built in to "latch" the switch on when presented with a capacitive load such as a CF bulb. The switch did dim the CF load locally - just not by X10. I don't really know, I haven't torn apart a SwitchLinc yet - so this is just a SWAG.

                            Anyhow, here's the data:

                            200W 100W 60W 40W 23W(CF)
                            100% 180 90 55 36 22
                            90% 170 85 53 34 *
                            80% 159 80 49 32 *
                            70% 145 73 45 29 *
                            60% 130 65 40 26 *
                            50% 113 57 35 23 *
                            40% 94 47 29 19 *
                            30% 74 37 23 15 *
                            20% 53 27 17 11 *
                            10% 35 18 12 8 *
                            * = Responded only to an X10 "ON" command

                            Rocco, with only a bulb as a load (no module or X10 switch inline) - the data looked like this:
                            200W bulb = 183W
                            100W bulb = 93W
                            60W bulb = 57W
                            40W bulb = 38W
                            23W CF = 21W
                            Notice that the power consumption of the incandescent bulbs (purely resistive load) was higher in all cases *without* an X10 module or switch in series. All but the CF had a power factor of 1.00 - the CF's was .87.
                            When an X10 module or switch is in the circuit, the power factor never goes above .87. As the level of brightness decreases, the power factor also decreases - all the way down to .26 at the 10% dim level. I think this really drives Johnh1's point home about the "level of dim" equating to how much of the sinewave the load actually sees.
                            Attached Files


                              Zoicos, thanks for the additional data.
                              Notice that the wattage with no switch is only a couple of watts higher than with a switch. Like you said, this supports Johnh1's observation, but also shows that a 200W bulb never gets to 200 watts.

                              Though the reading for the dimmer was only a couple of watts less, I wonder how much of that was getting to the bulb, and how much was dissipated by the dimmer. But you would need TWO meters to find that out.

                              lookit:\docs\LampDimmingData.pdf The dimming is kinda linear.