Dimmable CFLs? Spectacular Failures?

[Don Klipstein]

thanks for a great response... information that I'm familiar with, though
probably not backed by the knowledge you have. To continue the conversation
back to the original thought - can you make a similar statement concerning
the luminous efficacy of fluorescents when supplied 50% reduction in
wattage? I often give training seminars that are product related at
electrical distributors, and when this subject hits the floor, I'm somewhat
at a loss to give useful information.

50% power reduction for fluorescents:

Overall luminous efficacy should not change much, and could change in
either direction.

I remember from a few classic books (maybe Waymouth or Elenbaas or
both), and from a fluorescent lamp catalog from the late 1970's that
fluorescent lamps in general, and especially T12 ones, operate slightly
more efficiently when slightly to moderately dimmed - if the shape of the
current waveform stays the same.

There may be some exceptions.

Then comes the question of what happens when the temperature of any
given lamp changes as a result of change in input power, and the mercury
vapor concentration in it changes. The temperature could become closer to
or farther from that at which luminous efficacy is maximized.
Ideally, the lamp's temperature is optimal when the lamp gets full
power, so when power input is reduced enough to make much change in the
lamp temperature and the lamp cools, I would think that its efficiency
should normally decrease slightly. With only slight dimming, the
temperature could easily remain close enough to optimum to have less
effect on efficiency in producing light than the change in current density
within the lamp - so overall luminous efficacy could very slightly
increase.
However, for minor to moderate dimming, I would think that this usually
does not change luminous efficacy much in either direction.

Dimming ballasts may provide extra filament/cathode heating so that the
cathodes do not operate at too low a temperature during dimming. That
would indicate some decrease in overall luminous efficacy from dimming,
since dimming requires an increase in percentage of input power being used
to keep the cathodes hot.

Overall, I would say overall luminous efficacy could make a minor change
in either direction if a fluorescent lamp is operated at half power and
working OK when dimmed to such an extent. And I would not expect the
change to be the same at every installation.

- Don Klipstein ([email protected])

 

[[email protected]]

| Overall luminous efficacy of a tungsten filament varies greatly with its
| temperature. As a result, incandescent lamps have overall luminous
| efficacy severely compromised when they are dimmed.
|
| Put 50 watts into a 100 watt 120V A19 incandescent that produces 1710
| lumens with 100 watts, and what comes out is about 20% of that 1710
| lumens. At half power, overall luminous efficacy is about 40% of that at
| full power.
|
| This effect is a characteristic specific to incandescent lamps,
| including halogen lamps. Fluorescent lamps and LEDs have overall luminous
| efficacy not changing as much as input power is varied.

What is the luminous efficacy _change_ of an LED when it is operated by
pulsing the DC power to it, at levels full and zero, with a 50% duty
cycle for each level, at a rate of 1000 Hz?

Same question, but change lamp to incandescent, and change frequency to
0.001 Hz.

 

[Don Klipstein]

| Overall luminous efficacy of a tungsten filament varies greatly with its
| temperature. As a result, incandescent lamps have overall luminous
| efficacy severely compromised when they are dimmed.
|
| Put 50 watts into a 100 watt 120V A19 incandescent that produces 1710
| lumens with 100 watts, and what comes out is about 20% of that 1710
| lumens. At half power, overall luminous efficacy is about 40% of that at
| full power.
|
| This effect is a characteristic specific to incandescent lamps,
| including halogen lamps. Fluorescent lamps and LEDs have overall luminous
| efficacy not changing as much as input power is varied.

What is the luminous efficacy _change_ of an LED when it is operated by
pulsing the DC power to it, at levels full and zero, with a 50% duty
cycle for each level, at a rate of 1000 Hz?

Unchanged, except very slight increase due to the LED running cooler.

Same question, but change lamp to incandescent, and change frequency to
0.001 Hz.

Basically unchanged - extremely slight decrease for a tiny bit of that
1,000 second cycle with the filament being at temperature significantly
below "full" but significantly above ambient.

The decrease is significant when the frequency gets high enough for the
filament to have significant operation at in-between temperature.

- Don Klipstein ([email protected])

 

[[email protected]]

|>
|>| Overall luminous efficacy of a tungsten filament varies greatly with its
|>| temperature. As a result, incandescent lamps have overall luminous
|>| efficacy severely compromised when they are dimmed.
|>|
|>| Put 50 watts into a 100 watt 120V A19 incandescent that produces 1710
|>| lumens with 100 watts, and what comes out is about 20% of that 1710
|>| lumens. At half power, overall luminous efficacy is about 40% of that at
|>| full power.
|>|
|>| This effect is a characteristic specific to incandescent lamps,
|>| including halogen lamps. Fluorescent lamps and LEDs have overall luminous
|>| efficacy not changing as much as input power is varied.
|>
|>What is the luminous efficacy _change_ of an LED when it is operated by
|>pulsing the DC power to it, at levels full and zero, with a 50% duty
|>cycle for each level, at a rate of 1000 Hz?
|
| Unchanged, except very slight increase due to the LED running cooler.
|
|>Same question, but change lamp to incandescent, and change frequency to
|>0.001 Hz.
|
| Basically unchanged - extremely slight decrease for a tiny bit of that
| 1,000 second cycle with the filament being at temperature significantly
| below "full" but significantly above ambient.
|
| The decrease is significant when the frequency gets high enough for the
| filament to have significant operation at in-between temperature.

Kind of like the efficacy of dimming with a variable resistor

 

[[email protected]]

| On Sat, 12 Apr 2008 06:21:31 +0000 (UTC), [email protected]
| (Don Klipstein) wrote:
|
|>>
|>>| Overall luminous efficacy of a tungsten filament varies greatly with its
|>>| temperature. As a result, incandescent lamps have overall luminous
|>>| efficacy severely compromised when they are dimmed.
|>>|
|>>| Put 50 watts into a 100 watt 120V A19 incandescent that produces 1710
|>>| lumens with 100 watts, and what comes out is about 20% of that 1710
|>>| lumens. At half power, overall luminous efficacy is about 40% of that at
|>>| full power.
|>>|
|>>| This effect is a characteristic specific to incandescent lamps,
|>>| including halogen lamps. Fluorescent lamps and LEDs have overall luminous
|>>| efficacy not changing as much as input power is varied.
|>>
|>>What is the luminous efficacy _change_ of an LED when it is operated by
|>>pulsing the DC power to it, at levels full and zero, with a 50% duty
|>>cycle for each level, at a rate of 1000 Hz?
|>
|> Unchanged, except very slight increase due to the LED running cooler.
|>
|>>Same question, but change lamp to incandescent, and change frequency to
|>>0.001 Hz.
|>
|> Basically unchanged - extremely slight decrease for a tiny bit of that
|>1,000 second cycle with the filament being at temperature significantly
|>below "full" but significantly above ambient.
|>
|> The decrease is significant when the frequency gets high enough for the
|>filament to have significant operation at in-between temperature.
|>
|> - Don Klipstein ([email protected])
|
| Don't you mean "frequency gets low enough" in the last
| sentence.

I believe he did mean "high enough". At a low frequency like 0.001 Hz, the
light is on for 500 seconds (8m20s) and off for 500 seconds. If the frequency
were to be changed to something like 30 Hz, where it would be on for 1/30th a
second (two halves of a 60 Hz AC cycle) and off for 1/30th a second, the light
would never reach full brightness, and as such, its spectrum would be operating
in a less efficacious range. The total energy output in the 700nm to 400nm
wavelength range would be less as compared to the 0.001 Hz case, when averaged
over long time.

 

[Don Klipstein]

|>
|>| Overall luminous efficacy of a tungsten filament varies greatly with its
|>| temperature. As a result, incandescent lamps have overall luminous
|>| efficacy severely compromised when they are dimmed.
|>|
|>| Put 50 watts into a 100 watt 120V A19 incandescent that produces 1710
|>| lumens with 100 watts, and what comes out is about 20% of that 1710
|>| lumens. At half power, overall luminous efficacy is about 40% of
|>| that at full power.
|>|
|>| This effect is a characteristic specific to incandescent lamps,
|>| including halogen lamps. Fluorescent lamps and LEDs have overall
|>| luminous efficacy not changing as much as input power is varied.
|>
|>What is the luminous efficacy _change_ of an LED when it is operated by
|>pulsing the DC power to it, at levels full and zero, with a 50% duty
|>cycle for each level, at a rate of 1000 Hz?
|
| Unchanged, except very slight increase due to the LED running cooler.
|
|>Same question, but change lamp to incandescent, and change frequency to
|>0.001 Hz.
|
| Basically unchanged - extremely slight decrease for a tiny bit of that
| 1,000 second cycle with the filament being at temperature significantly
| below "full" but significantly above ambient.
|
| The decrease is significant when the frequency gets high enough for the
| filament to have significant operation at in-between temperature.

Kind of like the efficacy of dimming with a variable resistor

My data of severe efficacy loss for dimmed incandescent does not even
consider the dimmer taking on a loss as a result of dimming, while being
lossless without dimming.

It sounds awful enough for a 100 watt 120V "A19" with CC-8 or CC-6
filament and rated to last 750 hours to produce 20% or so of "full"
luminous output when it receives 50 watts.

This gets much worse still with a rheostat dimmer! Thankfully those
are not even the usual way to dim incandescents for stage lighting in
less-modern theaters nowadays, and nearly as rare as hairy eggs (should
they exist at all) in homes!

You wanna know how low the light output and energy efficiency are for
a 120V 100W "A19" with CC-6/CC-8 filament and 750 hour life expectany,
when such a lamp is dimmed by a rheostat for combined power consumption of
the lamp and the rheostat to be 50 watts?
At "half current", this "100A"/"100A19" 120V lamp has voltage drop of
about 33.5 volts. At this rate, about 14 of these 50 watts are going into
the lamp and the other 36 of these 50 watts are producing about 123
BTU/hour of heat in the rheostat.
At 33.5 volts, this 120V "100A19" lamp produces about .38-.39% of "full
light output".
Other 120V incandescents at 33.5 volts can easily produce anywhere
within or a little outside the range of 1/4% - 1/2% of "full output".
Given variations in lamp resistance as a function of voltage, other
incandescent lamps could have light output anywhere from less than 1/10 of
1% to maybe 1-2% of "full" at half its "full current".
Energy efficiency with a rheostat engaged to have power consumption
(including that of the rheostat) is twice the very low figures for
percentage of "full light output" - roughly .76%-.78% of "normal" for a
100 watt 120V "A19" rated to have average life expectancy of 750 hours,
and anywhere from under .2% of "full" to 2-4% of "full" for other
incandescent lamps.

Keep in mind much more mildly bad numbers of energy efficiency,
efficacy, etc. for dimming of incandescents via means that have most or
nearly all of the power going to the filament when dimming is engaged to
the extent to cut lamp power in half!

- Don Klipstein ([email protected])

 

[Don Klipstein]

<In response to my mention of incandescent lamps losing efficiency
bigtime when dimmed>

Don't you mean "frequency gets low enough" in the last sentence.

So far, if I haven't brain-farted too badly, I thought I was falling
into discussion of how overall luminous efficacy of an incandescent varies
from "its normal" when PWM-ed with 50% duty cycle at 1/1,000 Hz, or period
of 16-2/3 minutes. As in 8 minutes 20 seconds on, 8 minutes 20 seconds
off.

I would think that every incandescent I ever saw, when given "full
power" at a 50% duty cycle at 8 minutes 20 seconds "on and 8 minutes 20
seconds "off", has its overall luminous efficacy (ratio of photometric
output to energy consumption) close to that achieved with continuous
operation.

When I said "frequency high enough", I meant on-time and off-time
getting down to a couple seconds or fraction of a second, so that the
filament does some significant part of its glowing while not fully warmed
up. As in frequency being increased from .001 Hz to at least a goodly
fraction of a Hz, especially a few Hz.

At maybe around 1 Hz to a few Hz frequency, it appears to me that
50%-duty-cycle operation of most incandescents will have the filament
spending a fair amount of time glowing at "less-than-full" temperature,
and accordingly with decreased "overall luminous efficacy".

- Don Klipstein ([email protected])

 

[[email protected]]

[...]

|>Kind of like the efficacy of dimming with a variable resistor
|
| My data of severe efficacy loss for dimmed incandescent does not even
| consider the dimmer taking on a loss as a result of dimming, while being
| lossless without dimming.

And that would be hard to do, considering the variety of dimmers. That
would have to be an independent set of data. Then one would have to add
up the "wasted energy/power". Note that IR, UV, RF, and acoustic emission,
as well as thermal convection, when such are not desired, all constitutes
waste from dimmers, lamps and the wiring in between.


| It sounds awful enough for a 100 watt 120V "A19" with CC-8 or CC-6
| filament and rated to last 750 hours to produce 20% or so of "full"
| luminous output when it receives 50 watts.
|
| This gets much worse still with a rheostat dimmer! Thankfully those
| are not even the usual way to dim incandescents for stage lighting in
| less-modern theaters nowadays, and nearly as rare as hairy eggs (should
| they exist at all) in homes!

But some use variacs. BTDT.


| You wanna know how low the light output and energy efficiency are for
| a 120V 100W "A19" with CC-6/CC-8 filament and 750 hour life expectany,
| when such a lamp is dimmed by a rheostat for combined power consumption of
| the lamp and the rheostat to be 50 watts?
| At "half current", this "100A"/"100A19" 120V lamp has voltage drop of
| about 33.5 volts. At this rate, about 14 of these 50 watts are going into
| the lamp and the other 36 of these 50 watts are producing about 123
| BTU/hour of heat in the rheostat.
| At 33.5 volts, this 120V "100A19" lamp produces about .38-.39% of "full
| light output".
| Other 120V incandescents at 33.5 volts can easily produce anywhere
| within or a little outside the range of 1/4% - 1/2% of "full output".
| Given variations in lamp resistance as a function of voltage, other
| incandescent lamps could have light output anywhere from less than 1/10 of
| 1% to maybe 1-2% of "full" at half its "full current".
| Energy efficiency with a rheostat engaged to have power consumption
| (including that of the rheostat) is twice the very low figures for
| percentage of "full light output" - roughly .76%-.78% of "normal" for a
| 100 watt 120V "A19" rated to have average life expectancy of 750 hours,
| and anywhere from under .2% of "full" to 2-4% of "full" for other
| incandescent lamps.

I wasn't even able to get any kind of visible glow, observed in total darkness,
out of a 120 volt 4 watt bulb, when powered with 3 volts. I had to put the
3 volts bulbs back in

An no, I have no plans to test the 3 volt bulbs on 120 volts. It is kind of
sad that both use the same screw in socket.


| Keep in mind much more mildly bad numbers of energy efficiency,
| efficacy, etc. for dimming of incandescents via means that have most or
| nearly all of the power going to the filament when dimming is engaged to
| the extent to cut lamp power in half!

Maybe incandenscent bulb packages should have been required to list the power
needed to make them operate at various percentages of light output. Since
incandescent lights are soon going to underground markets, that option is no
longer viable.

 

[[email protected]]

| You are correct. I thought you were starting at 1000 Hz and
| moving down to where the mattered, instead of starting at
| 1/1000 Hz and moving up.

The original question I asked started the LED at high frequency and the
incandescent at low frequency. Sorry for the confusion.

I had asked a similar question a couple years ago regarding flipping on and
off three very high wattage lights (arrays) that would draw more current
than the circuit would allow. Drawing 36 amps on a 15 amp circuit, but only
having the lamp on for a few seconds. Then switch a different one on that
is on a different circuit on phase B. Then switch the third on from phase C.
Then back to A and so on. The point in that case was a shared neutral circuit
(three phase) to illustrate how the neutral could still be overloaded even
though none of the breakers would see any long term averaged overload. It was
an attempt to illustrate the effect of harmonics on neutral current.