After having some power issues, I've decided to investigate the largest power drains in our company, and I've arrived at a realization that 128 lights adds up to a lot of power (and that's not even all of them!).

Overall, I'm looking at the feisability of replacing all lamps with LEDs, but I'd like to avoid surprises of actual power usage being significantly higher than rated. As I started calculating the power usage, I've realized that there is a fairly large difference between the actual power draw and rated power.

The measured current at the grid connection was 0.31667 A / lamp - lowest result was used as "currently most efficient", though there might have been a busted lamp among them. Measurements were made in different areas for reliability and I made sure all other power users were accounted for. In this case, all lamps are linear fluorescent lamps, T8 58W 150cm, using starters.

Using the "standard" 220 V, I get the following result:

0.31667 A * 220 V = 69.6674 W

Using the calculated 233.33 V from the 35 kW, 3 x 50 A grid connection, I get this instead:

0.31667 A * 233.33 V = 73.8886 W

However, both are a far cry from the lamps' rated 58 W

0.31667 A * 183.1560 V = 58 W

Am I calculating something wrong? Is the discrepancy due to lamp deterioration or starters? Is there something else obvious that I'm missing?

Is this discrepancy also something I should worry with the planned LED replacements? Considering current lamps are 58 W and LED replacements would be 24 W.

I'm no expert and although I'm well aware that A * V = W isn't how it always works, I presumed that it should be accurate enough for lamps.

  • \$\begingroup\$ The rating you think about is the one of the lamp (58W), but the ballast uses some power too. Electronic ballasts (those without a starter) use less power, maybe 5W vs 10+ W of the old ones. Still, not economical to replace the ballasts now, just go LED. \$\endgroup\$
    – FarO
    Commented Jul 17, 2019 at 7:02
  • 1
    \$\begingroup\$ @FarO We've ended up switching practically all lamps to LED about 10 months ago and although there's a few caveats to keep in mind, I can confidently say that the investment has already been covered by power savings alone. Nowadays unless your power is extremely cheap, if your lamps are running at least 6 hours per day, you are very likely losing money by not switching to LED. \$\endgroup\$
    – Sašo
    Commented Jul 17, 2019 at 13:35

5 Answers 5


Also consider that many LEDs can bring up the power quality problem. LEDs are bringing higher harmonics to the system, and cutting those out may need designing a low-pass filter. [1]

If you want to read more about Power Factor correction, you can look at [2].

Also your electricity provider may want to measure the power readings from time to time if it's a company building, and if it discovers that you're flooding the grid with harmonics high fees may be brought up on your company, as you're decreasing grid power quality.

For AC circuits use P=U* I* cos(phi) to measure the real power. If you want to measure the cosinus(phi) you can measure the real power(P), then measure RMS values of current and voltage, and from that calculate the apparent power (S) by multiplying I(RMS) times U(RMS). Cosinus phi would be a P/S. There are also measuring instruments that can measure cos(phi) by them own.



  • \$\begingroup\$ Using 0.86 cosφ power factor I've found quoted on Electrical Installation Wiki for flourescent lights, my numbers end up coming close to what I'd consider a margin of error (69.66 W measured vs. 67.44 W calculated). \$\endgroup\$
    – Sašo
    Commented Jul 9, 2018 at 12:22

The measured power at the grid connection was 0.31667 A / lamp

No, that's not power that's current. If you want to measure power then use a wattmeter. Reason: power is voltage x current and if the phase angle between voltage and current is not zero (unity power factor) then measuring current will give an over inflated perception of power (assuming that your voltage supply was fixed at 220 volts).

However, you know that if current goes down significantly when you replace one lamp with LEDs, then in all likelihood power will also reduce. So just get your boss to sign into changing one lamp and get one of those cheap watt meters you can get off ebay and set up an experiment.

LEDs will win every time so don't delay - do a single experiment, tell your boss the results and calculate payback period. Then hopefully you'll see enough evidence to convince your boss to make the big change: -

enter image description here

Picture source.

  • \$\begingroup\$ Whoops, I forgot about current when rewriting that part - fixed. Speaking of testing, this was measured when nothing else was running on a saturday morning, so there's as little voltage fluctuation as there can be. But does this mean that if I multiply the ratio between rated and measured (1,27ish) with the LED power usage, will I get a realistic result? Or is it best to just work with the 58W - 24W (LED) = 34W (savings) and assume ethat's the worst case scenario? \$\endgroup\$
    – Sašo
    Commented Jul 7, 2018 at 13:42
  • 2
    \$\begingroup\$ Just a note about LED lamps - heat will cause them to fail prematurely. Most read on the packaging that they "should not" be used in sealed fixtures, but this really means they can not. \$\endgroup\$
    – rdtsc
    Commented Jul 7, 2018 at 15:36
  • \$\begingroup\$ @Sašo you have to compare apples with apples so you are looking to compare two technologies that produce the same light output per watt. \$\endgroup\$
    – Andy aka
    Commented Jul 8, 2018 at 10:54
  • \$\begingroup\$ Lumens per watt are not strictly comparable when comparing FL to LED though, since LED replacements are more directional than FL lamps. Since even reputable LED manufacturers are selling 19-24 W LED replacements for 58W FL lamps with brighter ones just not being out there, I have to presume that a 2300 lumen LED lamp is, for the lack of a better word, just as effective as a 4700 lumen FL, keeping in mind the usual height the lamps are installed on. Though lamp effectiveness is a separate issue from actual power draw. The latter can be extrapolated but the first needs actual testing. \$\endgroup\$
    – Sašo
    Commented Jul 8, 2018 at 11:22
  • \$\begingroup\$ You say "LEDs will win every time", then you show an illustration where the LED bulb has a lower efficacy (64 lm/W) than a T8 florescent (80 lm/W). Current CFLs have an efficacy of 70 lm/W and your illustration shows 57 lm/W. You illustration is ancient. I'd guess 10-15 years old. \$\endgroup\$ Commented Jul 8, 2018 at 11:38

Power Factor must be taken into account in AC power systems:

P = V * I * cos(phi)

Voltage can vary during the daytime, so both the values you wrote could be correct.

  • 1
    \$\begingroup\$ This applies especially to Fluorescent lamps as older ones use an inductive ballast to control the lamp current so the power factor may be much less than 1. \$\endgroup\$ Commented Jul 7, 2018 at 14:20
  • \$\begingroup\$ Even led lamps have a quite low power factor (my desk lamp has PF = 0.74), but the load is capacitive. \$\endgroup\$
    – Railectric
    Commented Jul 7, 2018 at 15:29

Modern tri-phosphor tubes 4500’K to 5000’K “can” be as efficient and better CRI than cheap LED lamps 88LPW but both better than incandescent 15 LPW.

The simple quad magnetic ballast is not constant current as you assumed incorrectly. T5’s x 4ft use 28W while T8’s may be 32W and expect 2800 lumens from the best.

There are many choices in better lighting and cost may be a deceptive measure of “better”. Educate yourself in Lumination Engineering if you wish to be wiser.

Also learn how to make true RMS power measurements.

It is easy to retrofit but use the more expensive triphosphor tubes rated for 30kh using 1 switch cycle/day or 50kh is continuous. Your eyes will appreciate the contrast and reading improvement.

  • \$\begingroup\$ Power usage is a greater concern than cost, though LED lamp costs could be covered within 2 years and past that it's all savings. I'm well aware of lower luminance of the so called "equivalent" LED lamps, but that's just something I'll have to test ahead and see if it's sufficient for our requirements. As far as high frequency triphosphorous lamps go, they sound intriguing, though I can't find anything concrete on power savings, but unless it's at least 40% (for same luminance), it's not exactly an option. \$\endgroup\$
    – Sašo
    Commented Jul 7, 2018 at 16:36
  • \$\begingroup\$ net LPW is the factor to compare and CRI \$\endgroup\$ Commented Jul 7, 2018 at 20:00

It's really not about wattage of the light source, it's how much light are you getting per wall watt.

For efficiency light sources for general lighting will have an efficacy rating in lumens/watt. Your common everyday T8 58 W linear fluorescent triphosphor bulb delivers about 80 lm/W minus diffuser panel transmittance loss, and directional loss (light going in to the ceiling).

What type of lighting depends on whether it is an office, retail, or manufacturing setting and ceiling height.

Then there is the color temperature expressed as cool, neutral, and warm which is specified in Kelvins as a correlated color temperature (e.g. CCT=4000K).

Examples of CCT. I took these photos of white paper reflecting 1750K 98 CRI, 3000K 80 CRI, and 2700K 97 CRI LED illumination. Red meat is a specialized LED spectrum for a meat counter to make red meat look more appealing. The red meat LED is very similar to the 1750K spectrum.

enter image description here

Then there is the color rendering index (CRI typ. 70-90) which indicates the quality of the light and how closely it approximates sunlight (e.g. CRI=80). Higher is better and more expensive. 80 CRI is middle of the road. 90 CRI contains more of the red spectrum.

The upper half of this photo is low CRI the bottom is high CRI.

enter image description here

In general lower CCT and higher CRI light sources (other than incandescent) will have lower efficacy ratings (e.g. 2700K 90 CRI ≈150 lm/W max).

Your typical linear fluorescent is a T12 4' tube. CCT ranges from 4000K to 6500K and efficacy is in the 60 lm/w range. CFL bulbs are about 70 lm/w. T8 Linear fluorescent 80 lm/w

The problem with comparing LEDs to light bulbs is that an LED is nothing like a light bulb. The biggest issue is the lumen measurement. Because light bulbs are an isotropic light source (radiates light equally in all directions) lumens works well. Lumens measures the total light exiting the light bulb in every direction.

LEDs are an anisotropic light source meaning the light exiting the source does so in a directional path. This means you can point LED light in a particular direction where the lumens are concentrated in a smaller area. Linear fluorescent tubes will also have a reflector that attempts to redirect the light towards the area to be illuminated. Some of the illuminance is lost or misdirected with the reflectors.

The graphic below represents a typical anisotropic LED. If it were an isotropic source the lines from the source would be equal in intensity and cover the entire circle.

enter image description here

What matters more than the Lumens is the Illuminance or Lux (lm/m²). An LED fixture, with the same luminous flux (lumens) as a florescent fixture, will likely measure more lux on an illuminated surface than the florescent fixture.

LEDs fixtures have the dimming advantage. If you don't need as much light as the fixture outputs, just dim it and save. Can improve ambiance as well.

Linear fluorescent's efficacy does not include the diffuser panel which can block up to 80% of the light. Diffuser panels will have a transmittance rating which indicates the percentage of light that goes through the panel and is not reflected or absorbed. A clear polycarbonate panel will have about 95% transmittance. 40% transmittance for milky white diffuser is considered good.

I replaced all my incandescent light bulbs with LED light bulbs over 10 years ago. I would not do the same today. I am not a fan of LED light bulbs.

When LEDs get hot they produce less light. When LEDs are concentrated in a small area as is done in a light bulb. An individual LED bulb needs it own power supply and LED driver which lowers efficiency and reliability.

I now use strips of many inexpensive mid power LEDs where the LEDs are spaced about 10mm apart with a 45V forward voltage.

I use highly efficient and reliable Mean Well HLG 48V drivers with power factor correction. I use 48V because it has higher efficiency than lower voltage drivers. Also the reason I have 45V strips.

A good replacement for T12 tubes is Samsung F-Series Gen 3 1120 mm (44") strips. The are also available in 280 mm and 560 mm lengths.
Efficacy 180 lm/W about 2.25x the efficacy of a T8 bulb.

For good looking fixtures I like the Klus Design Fixtures and Extrusions to house the strips. They have a lot of nice hardware for installing strips in ceilings, walls, floors, in hand rails, stair steps, under cabinets and counters.

enter image description hereenter image description hereenter image description hereenter image description here

LINK: Klus Catalog

I make my own 560 mm x 9 mm wide strips to get 2700K 90 CRI and they fit in the narrower (cheaper) 10 mm extrusions.

Below is a 560 mm strip (circled in white) sitting on top my 55" monitor facing upwards so the light is diffused by the ceiling. It is drawing 17 wall watts including the HLG-40-48B driver. It is positioned 40" from the back wall.

You can see the directional intensity. The LEDs are pointed straight up. Directly above the strip (the top edge of photo) it is brightest. The illuminance decreased on the back wall as the light angle changes (there is no back to the bracket the strip is mounted on.

If you click and zoom in on the light you will see what appears to be a blue power indicator in the center of the strip. The strip is screwed on to an aluminum angle bracket which is covered with blue electrical tape.

enter image description here

Below is a Samsung F-Series Gen 3 strip with a Klus micro-ALU extrusion rigged up as a temporary plant light.

enter image description here

  • \$\begingroup\$ Thanks for the overview. The heat is a concern, especially since I'm required to use IP6x enclosures (dust). As far as changing regular bulbs to LED goes, in my experience (in private use) they've improved a lot in the last few years. The fairly recent "retro" filament lamps also seem pretty good but that's a very general observation with the usual YMMV depending on the brand. \$\endgroup\$
    – Sašo
    Commented Jul 8, 2018 at 9:56
  • \$\begingroup\$ There are some Bridgelux EB-Series Gen 2 strips that are a little better than Samsung at heat dissipation but they are wider and very difficult to find an extrusion they will fit into. Klus has many options including IP67 and some IP68, you should download the catalog if you have not yet done so. Heat should not be a big problem. Extrusions act as heatsinks and what I do is reduce current until the temperature is what I am looking for. I ran some Bridgelux 560 mm strips at 1400 mA and the temp was about 42°C in open air. Samsung runs warmer than Bridgelux. \$\endgroup\$ Commented Jul 8, 2018 at 10:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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