CFL real power used TEST

[JohnR66]

I powered some CFLs from a filtered fullwave bridge (820uf filter cap) for DC output and measured current on the DC side x DC voltage to get true power drawn. Since load on DC PSU causes voltage to vary a bit, I checked for each lamp. I allowed the CFLs to stabilize for a few minutes.

Feit BPESL13T 13 watt daylight = 12.1w
Meijer Ultramini 13w natural white = 12.7w
n:vision EDXO14 14w 3500k = 14.4w
n:vision EDXO23 23w 3500k = 21.4w
Meijer Ultramini 23w natural white = 20.0w
n:vision 8TCO3 3w dimmable = 4.7w

Conclusion: It is a toss up. Some lamps were over rated power and some under. The meijer 23 watt lamp was 3 watts under. The n:vision 3 watt lamp was well over. A couple were in spec (within .5 watt - my arbitrary value). Power loss in lamps internal fullwave bridge was not accounted for, but amounts to a couple tenths of a watt max.

 

[fyrstormer]

Any reason why a simple Kill-A-Watt meter wouldn't be accurate?

 

[jtr1962]

Any reason why a simple Kill-A-Watt meter wouldn't be accurate?

Those are fine up to a point, but they don't read to tenths of a watt, and I'm not really sure how accurate they are with small loads. The way JohnR66 did it seems to be a more accurate method.

 

[broadgage]

Those are fine up to a point, but they don't read to tenths of a watt, and I'm not really sure how accurate they are with small loads. The way JohnR66 did it seems to be a more accurate method.

Agree, the plug in power and energy monitors are not very accurate at low loads such as a single CFL.

Here in the UK I have used a plug in energy monitor to accuratly measure the consumption of CFLs.
I connected at least 10 lamps at once so as to put a greater load on the device and increase accuracy.
I found that reputable brands of lamp were allways within 10% of the stated consumption, and often within 5%.

Cheap unknown brands varied widely, with one "20 watt" lamp only being 11 watts.

 

[TorchBoy]

Agree, the plug in power and energy monitors are not very accurate at low loads such as a single CFL.

With all this talk of accuracy, why is a sample size of 1 considered accurate anyway? Wouldn't taking a sample of 10 random CFLs of the same sort and dividing their total power consumption by 10 to provide their average power consumption be more representative of each model of CFL? How much variation do they have? :shrug:

(Edit: I see broadgage addresses some of these points.)

Actually I'd be more interested in knowing which the most efficient brands were.

 

[JohnR66]

A complete test would include the light output of each lamp. The Meijer Ultramini 13w natural white, which seems to be about 4100K, appears to be the brightest to my eyes of the "60 watt" replacements. The Meijer branded bulbs use a narrower diameter tube of longer length (extra twist in the tube) so it might be more efficient.

It would be nice for someone with the equipment who could check for light output to further test the bulbs.

 

[TorchBoy]

Hm, finding an average wouldn't give an indication of spread. http://www.consumer.org.nz/reports/energy-efficient-lighting/test-results-2 has some more test results.

 

[ponygt65]

Im curious. Why the tests? Is it to settle curiosity or accuracy of brands? Not to be a jerk, but what exactly is the point? If it's just tinkering and doing stuff that interests you then.....:thumbsup: Otherise, i fail to see the poit.

 

[blasterman]

Im curious. Why the tests?

Because....there have been many claims that CFLs use more power than documented on the box. I also give JohnR66 kudos for taking the time and effort to make these tests. Not sure why you are even posting.

Even though JohnR66 didn't have multiple samples of the same bulbs, often a relative comparison like this tells you more information. Testing deviations between the same bulbs would be interesting, but given the nature of manufacturing in China and the time span by which we replace CFLs this wouldn't really tell us much as we initially think.

Looking at the actual power usage of different brands here tends to indicate a trend that manufacturers overall stroke the ballasts in these bulbs to utilize slightly lower than advertised wattage. My theory on this is it's basically to avoid advertising litigation; Lumen values are far more abstract and subjective in terms of formal empirical testing than wattage use. So, it only makes sense that the CFLs are more likely to use less than advertised wattage because it's easier to fudge lumens.

Also, while lumen tests are interesting, they are difficult to do with CFLs because of all the variables in play. Also, you need to test bulbs in the same kelvin range, which means the same phosphor balance. At least subjectively, a 23watt 5500k bulb will always appear brighter than a 23watt 2700k.

Again, good tests.

 

[TorchBoy]

Not sure why you are even posting.

I'd say he's posting in an attempt to understand the rationale behind the tests. I had similar questions/doubts when I first read the local Consumer tests - they seemed a bit of a waste of time to me because they didn't consider actual power use. I'm glad this test does, but like the Consumer test it too doesn't test everything.

Lumen output changes during the life of a CFL, so I look forward to seeing some lumen data, or even just brightness readings. It would be good to have them done from multiple angles.

 

[JohnR66]

I have 3 of the Meijer Ultramini 13w natural white CFLs, all with <20 hours. Two came up with 12.6w and one shows 11.8w. It would be nice to test at least 10 from different lots, but I only have the three.

My reason for testing was to satisfy my own curiosity of the true power used since volts * amps (current as printed on the lamp) is not true power used.

Since I did not see this test performed here, I thought I'd share the results.

 

[jtr1962]

Lumen output changes during the life of a CFL, so I look forward to seeing some lumen data, or even just brightness readings. It would be good to have them done from multiple angles.

Well, I recently made a test jig to start testing some LED bulbs for someone. I can also test CFLs (in fact I did test one). I'll start a thread on it as soon as I find the time (very busy lately). However, I can't publicly release the LED bulb results until the person I'm testing them for gives the say so.

 

[blasterman]

I'd say he's posting in an attempt to understand the rationale behind the tests.

If somebody posts beam shots of trees and weeds at night of a $100 vs $200 flashlight, and shoots those pictures with a digicam in full automode with no corresponding control over camera white balance, and says 'take my word about the color', nobody complains. 50 additional responses commenting on the "throw" of the light and how they need to buy one. :thinking:

JohnR66 takes the time to test the actual power consumption of CFLs, and he's asked for 'rationale'? Good grief...

 

[TorchBoy]

My reason for testing was to satisfy my own curiosity of the true power used since volts * amps (current as printed on the lamp) is not true power used.

Since I did not see this test performed here, I thought I'd share the results.

Fair enough, and thanks. I think it is useful data, as far as it goes, although I don't think any of those brands are available here.

I can also test CFLs (in fact I did test one).

Was there much directionality in the light output?

I have mixed feelings about white balance, blasterman. I forgot to fix the white balance when I took before and after shots of my car dome light mod, and after kicking myself (figuratively) I figured that our eyes compensate even more. :shrug: And I think that someone wanting to know why a test was run is good. It's not always obvious since we all approach these things in different ways, and at least they're interested in finding out.

 

[jtr1962]

Was there much directionality in the light output?

Not really. The intensity fell to only 90% at around 95° off-axis and 50% by the time you got to 140° off-axis. But that's to be expected because the base was blocking part of the light from the tube. This was a globe CFL BTW. A spiral would probably have a less even distribution. I'll show all my results as soon as I have time to make a thread.

 

[ponygt65]

Because....there have been many claims that CFLs use more power than documented on the box. I also give JohnR66 kudos for taking the time and effort to make these tests. Not sure why you are even posting.

Even though JohnR66 didn't have multiple samples of the same bulbs, often a relative comparison like this tells you more information. Testing deviations between the same bulbs would be interesting, but given the nature of manufacturing in China and the time span by which we replace CFLs this wouldn't really tell us much as we initially think.

Looking at the actual power usage of different brands here tends to indicate a trend that manufacturers overall stroke the ballasts in these bulbs to utilize slightly lower than advertised wattage. My theory on this is it's basically to avoid advertising litigation; Lumen values are far more abstract and subjective in terms of formal empirical testing than wattage use. So, it only makes sense that the CFLs are more likely to use less than advertised wattage because it's easier to fudge lumens.

Also, while lumen tests are interesting, they are difficult to do with CFLs because of all the variables in play. Also, you need to test bulbs in the same kelvin range, which means the same phosphor balance. At least subjectively, a 23watt 5500k bulb will always appear brighter than a 23watt 2700k.

Again, good tests.

If somebody posts beam shots of trees and weeds at night of a $100 vs $200 flashlight, and shoots those pictures with a digicam in full automode with no corresponding control over camera white balance, and says 'take my word about the color', nobody complains. 50 additional responses commenting on the "throw" of the light and how they need to buy one. :thinking:

JohnR66 takes the time to test the actual power consumption of CFLs, and he's asked for 'rationale'? Good grief...

I'll respond later....but for now; IDK why you felt the need to post with such hostility. I even gave him the thumbs up for doing it.

In short, if these tests were being done to 'prove' something, it's an inaccurate test on teh sole basis of variation between production among other things. That's like saying all 04 mach 1 mustangs dyno 275RWHP because one did on a certain day. Ain't gonna happen. They have what's called cam timing variance, so even with the same tune, motor, built specs, they can still be different.

But, john did clarify why he did it, and as you can see; I gave a thumbs up for it. And I'll do it again.

I have 3 of the Meijer Ultramini 13w natural white CFLs, all with <20 hours. Two came up with 12.6w and one shows 11.8w. It would be nice to test at least 10 from different lots, but I only have the three.

My reason for testing was to satisfy my own curiosity of the true power used since volts * amps (current as printed on the lamp) is not true power used.

Since I did not see this test performed here, I thought I'd share the results.

:thumbsup:

 

[broadgage]

In my experience, the variation in actual watts used by different lamps of the same type is very small.
I have tested 20 Sylvania CFLs each 230/240 volts, 15 watt and found less than 0.1 watt difference between examples.
I would not expect any significant difference since they all come of the same production line having been manufactured with the same components.
In the longer term, perhaps from one year to the next, larger variations might be expected as the factory uses up component stocks and perhaps starts useing alternatives with slightly different tolerances.
All the lamps that I tested come from two cartons obtained at the same time from the same source.

I have also tested some 12 volt CFLs, the reputable ones were all close to the stated wattage, but cheapo "11 watt" ones from fleabay were only about 6 watts, though gave quite a good light for the 6 watt input.

 

[brickbat]

...it's an inaccurate test on teh sole basis of variation between production among other things. That's like saying all 04 mach 1 mustangs ...

You're drawing an inaccurate analogy.

Go back and read his conclusion. The OP never made any generalization about his measurements applying to ALL CFLs.

And I'll take actual measurements on a sample of 1 over guesses or assumptions, or hearsay any day.

I'm glad for the OP's post. Some people have been making an AC current measurement, multiplying that by 120V, and (not knowing about PF) concluding that CFLs draw twice their rated power. This post pretty well dispels that. :thumbsup:

 

[broadgage]

You're drawing an inaccurate analogy.

Go back and read his conclusion. The OP never made any generalization about his measurements applying to ALL CFLs.

And I'll take actual measurements on a sample of 1 over guesses or assumptions, or hearsay any day.

I'm glad for the OP's post. Some people have been making an AC current measurement, multiplying that by 120V, and (not knowing about PF) concluding that CFLs draw twice their rated power. This post pretty well dispels that. :thumbsup:

Agree, a sample of more than one would be better, but a single sample is indeed much better than guesses, assumptions or hearsay.
And I measured 20 similar lamps and found no significant difference.

Reputable makes of CFL seem to be at least as accurate as incandescent lamps as regards wattage.

 

[ponygt65]

You're drawing an inaccurate analogy.

Go back and read his conclusion. The OP never made any generalization about his measurements applying to ALL CFLs.

And I'll take actual measurements on a sample of 1 over guesses or assumptions, or hearsay any day.

I'm glad for the OP's post. Some people have been making an AC current measurement, multiplying that by 120V, and (not knowing about PF) concluding that CFLs draw twice their rated power. This post pretty well dispels that. :thumbsup:

I did not make an inaccurate analogy as my response was to blasterman. please show in my original post where I stated teh OP was making a generalization. I, in fact, asked the question as to why because it was not clear whether it was to generalize or merely show that he was interested in 'tinkering'. And the OP's answer was, in fact, not a generalized one. Therefore I gave him the thumbs up. Also, measurements on a basis of even 1 % is still an assumption to some degree since it is such a small amount by comparison. Not to say I don't agree that I'd take factual information over 'it should' based on calc.s etcetera.

I really don't see how this turned into such an issue. I agree with the OPs post and am glad to see it. I merely asked what the reasoning behind it was. :thinking: