Dimming LED

Status
Not open for further replies.

winnie

Senior Member
Location
Springfield, MA, USA
Occupation
Electric motor research
My goodness, look at the devils lurking in the details.

Gar: thanks for generating the plots of current on the LED bulbs. I had no idea that a 120V 9.5W bulb would have a peak current of more than 2A. As you note, this is >2A for a very short time...but put enough of them together and you could have a headache!

-Jon
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
160308-2238 EST

George Stolz:

In post #1 you stated
I am looking to dim 650W of LED lamps on one switchleg. From what I can discern, the best i can get for Lutron is 250W.
From this I assume you want to provide dimming control of about 68 LED bulbs that have a steady-state actual power consumption of 9.5 W each or their equivalent in some larger size. This 650 W dimmer might allow about 10 incandescents rated at 60 W each. Lutron is saying that you can only control about 26 LEDs of 9.5 W each on the 650 W dimmer. You are still way ahead on light output for one dimmer.

Based on field data and probably laboratory tests Lutron has apparently decided that this dimmer should only be rated for 250 W of LED lights. When simply lumping all LEDs into one LED category, then as a manufacturer you need to take a conservative approach to rating your product. Of the LEDs I have tested (very few) I can not judge whether the Lutron rating is reasonable or not. There are many different LEDs and CFLs in the market and there may be great differences between them. The dimmable Feit I have has a much shorter. but higher amplitude repetitive inrush currrent than does the CREE I tested.

It would be good to run tests on large groups of LEDs, dimmers, and circuit breakers to determine what are reasonable operating limits and why. I am not going to do this.

Somewhere along your posts you seemed to have brought up a breaker problem. What was the breaker load and the breaker specification? What caused the breaker to trip? Why was the customer so upset? Does this relate to LEDs and dimmers? How does this relate to post #1?

.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
160309-1149 EST

More on the LED effect on a solid-state switch (the dimmer Triac or whatever is used).

A previous statement I made may be somewhat misleading on the relationship to the energy per cycle dissipated in the dimmer between the repetitive transient and the more steady current during a half cycle.

Why is the measurement of I^2*T of importance? If you look at the power equation for a resistive load one form is P = I^2*R. Energy is power times time. For a given resistance, then Energy is proportional I^2*T. A mass takes time to dissipate energy. So there is a time constant to the dissipation of energy from a mass. The temperature of a mass is a function of the energy state of the mass.

If energy is put into a mass very quickly compared to its dissipation time constant, then the temperature rise is proportional to the energy put in. If the mass is changed, then the temperature rise is inversely proportional to the input energy.

The life or abrupt failure of a semiconductor device is a function of its temperature.

Semiconductor switching devices are somewhat constant voltage drop devices below their state-state rating, but above they look more resistive. Power to a constant voltage load is proportional to current, but to a constant resistance load power is proportional to the square of current.

Thus, for short duration periods one considers wires, fuses, resistors, and semiconductors to have a failure point relative to I^2*T.

The following tests relate to the repetitive transient current thru a dimmer switch (Triac) from turn-on at the AC line voltage peak.

A 1 ohm carbon comp resistor is used to measure current.


CREE 9.5 W bulb..

See Graphs in this post below http://forums.mikeholt.com/showthread.php?t=175292&page=2&p=1721580#post1721580



The first plot shows the major portion of a power cycle. You see a big transient of 2.25 A peak for less than a 0.1 millisecond, and then a current of about 0.1 A for about 4 milliseconds. All of the 0.1 A might not need to be treated as I^2*T
compared to I*T. This really depends upon the thermal time constant of the semiconductor chip. But I will treat all as I^2*T.

The second plot is expanded in time so that the transient details can be seen. For calculations I am using amperes and microseconds.

These calculations are based on visual guesstimates.

2.25*2.25*25 = 126 for the transient
0.1*0.1*4000 = 40 for the steady portion
Total for the CREE for 1/2 cycle = 166


Power or energy dissipated in the switch is not the same as what is transferred to the load. The repetitive transient current contributes little to the power to the load.




Feit 9.5 W bulb.

See Graphs in this post below http://forums.mikeholt.com/showthread.php?t=175292&page=2&p=1721580#post1721580



The first plot shows the major portion of a power cycle. You see a big transient of 3.5 A peak for less than 0.005 millisecond, another lower pulse, and then a declining current with an increasing oscillation.

The second plot is time expanded to see the oscillation.

The third plot is expanded in time so that the transient details can be seen. For calculations I am using amperes and microseconds.

These calculations are based on visual guesstimates.

3.5*3.5*5 = 61 for the tall transient
1.5*1.5*10 = 23 for the medium transient
0.2*0.2*3000 = 120 for the somewhat steady portion with oscillation
Total for the Feit for 1/2 cycle = 204

The frequency of oscillation is around 30 kHz.


An easy way to get answers is to use the actual SSR temperature rise and to use the real world device as its own analog computer to analyze the interaction of the complex current waveforms, variation of switch voltage drop, and thermal time constants.

.
 
Last edited by a moderator:

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
160309-1654 EST

There are problems with this forum's software and pictures. So this is a redo.

160309-1149 EST

More on the LED effect on a solid-state switch (the dimmer Triac or whatever is used).

A previous statement I made may be somewhat misleading on the relationship to the energy per cycle dissipated in the dimmer between the repetitive transient and the more steady current during a half cycle.

Why is the measurement of I^2*T of importance? If you look at the power equation for a resistive load one form is P = I^2*R. Energy is power times time. For a given resistance, then Energy is proportional I^2*T. A mass takes time to dissipate energy. So there is a time constant to the dissipation of energy from a mass. The temperature of a mass is a function of the energy state of the mass.

If energy is put into a mass very quickly compared to its dissipation time constant, then the temperature rise is proportional to the energy put in. If the mass is changed, then the temperature rise is inversely proportional to the input energy.

The life or abrupt failure of a semiconductor device is a function of its temperature.

Semiconductor switching devices are somewhat constant voltage drop devices below their state-state rating, but above they look more resistive. Power to a constant voltage load is proportional to current, but to a constant resistance load power is proportional to the square of current.

Thus, for short duration periods one considers wires, fuses, resistors, and semiconductors to have a failure point relative to I^2*T.

The following tests relate to the repetitive transient current thru a dimmer switch (Triac) from being turned-on at the AC line voltage peak.

A 1 ohm carbon comp resistor is used to measure current to minimize unrelated high frequency ringing..


CREE 9.5 W bulb.

DS2_QuickPrint81-I.JPG


DS2_QuickPrint82-I.JPG


The first plot shows the major portion of a power cycle. You see a big transient of 2.25 A peak for less than 0.1 millisecond, and then a current of about 0.1 A for about 4 milliseconds. All of the 0.1 A might not need to be treated as I^2*T
compared to I*T. This really depends upon the thermal time constant of the semiconductor chip. But I will treat all as I^2*T.

The second plot is expanded in time so that the transient details can be seen. For calculations I am using amperes and microseconds.

These calculations are based on visual guesstimates.

2.25*2.25*25 = 126 for the transient
0.1*0.1*4000 = 40 for the steady portion
Total for the CREE for 1/2 cycle = 166


Power or energy dissipated in the switch is not the same as what is transferred to the load. The repetitive transient current contributes little to the power to the load.




Feit 9.5 W bulb.

DS2_QuickPrint83-I.JPG



DS2_QuickPrint84-I.JPG





DS2_QuickPrint85-I.JPG


The first plot shows the major portion of a power cycle. You see a big transient of 3.5 A peak for less than 0.005 millisecond, another lower pulse, and then a declining current with an increasing oscillation.

The second plot is time expanded to see the oscillation.

The third plot is expanded in time so that the transient details can be seen. For calculations I am using amperes and microseconds.

These calculations are based on visual guesstimates.

3.5*3.5*5 = 61 for the tall transient
1.5*1.5*10 = 23 for the medium transient
0.2*0.2*3000 = 120 for the somewhat steady portion with oscillation
Total for the Feit for 1/2 cycle = 204

The frequency of oscillation is around 30 kHz.


An easy way to get answers is to use the actual SSR temperature rise and to use the real world device as its own analog computer to analyze the interaction of the complex current waveforms, variation of switch voltage drop, and thermal time constants.

Dennis can you delete the contents of the previous version of this post where the plots were lost, and just inxert a note referencing this post?

.

.
 

Fulthrotl

~Autocorrect is My Worst Enema.~
I didn't want to spoil all the fun; ask for the time, get detailed instructions on how to build a nuclear submarine. Another typical day on the forum. It may prove useful to some.

well, dimming LED's is problematic. not all across the line dimming devices will dim LED's,
and the ones that claim LED dimming ability won't always work with all brands of LED's,
even if the LED lights claim to be dimmable. and sometimes even using a dimmer approved
for a specific brand of LED light or fixture gives varying success, as manufacturers change
internal components depending on supply situations.

if you really want to dim LED's, then 0-10 volt dimmable devices is the most reliable
way to do it, unless you go with something proprietary like lutron's ecosystem, but the
price point has now moved far beyond reality.
 

sparkyrick

Senior Member
Location
Appleton, Wi
I am looking to dim 650W of LED lamps on one switchleg. From what I can discern, the best i can get for Lutron is 250W.

Leviton claims a 1000W dimmer for LED. Is there fine print I am missing, or is Lutron sucking wind on this particular front?

Edison base A19 LED replacements.

Dennis, educate me: power booster...?

Were these all on one switch leg originally? When you say 650W, was that the original incandescent wattage? If it was, your LED wattage will be much less.
 

sparkyrick

Senior Member
Location
Appleton, Wi
I didn't want to spoil all the fun; ask for the time, get detailed instructions on how to build a nuclear submarine. Another typical day on the forum. It may prove useful to some.

laff.gif
laff.gif
laff.gif
 

George Stolz

Moderator
Staff member
Location
Windsor, CO NEC: 2017
Occupation
Service Manager
Were these all on one switch leg originally? When you say 650W, was that the original incandescent wattage? If it was, your LED wattage will be much less.

I don't know how to say it more plainly, but here's another attempt:

72 lamps x 9 watts apiece = 648W.

There does not appear to be a dimmer that will handle this.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
160312-1522 EST

George:

In post #1 you did not explicitly state
72 lamps x 9 watts apiece = 648W
.
You need to know what the specifications are for a control element and how those specs relate to a controlled load.

In town we have what at one time was claimed to be the most powerful laser in the world, and it still might be. I also believe the statement may have been made that its output power was greater than all of the generating capacity in the world. How can this be? It is really a funtion of what the words mean.

Why would you expect or be surprised that a dimmer rated for 650 W of incandescent, and that really only means ordinary incandescent bulbs, be rated at 650 W for any load that has an average power rating of 650 W? An average power rating hides many factors. Peak power alone also hides many factors. For example RFI.

You need to know how devices work, and their characteristics to determine how different devices work together.

The interaction of fluorescents, CFLs, and LEDs with dimmers is a jungle.

Lutron does not provide much quantitative information in the following report, but it is a good qualitive overview of the problems. http://www.lutron.com/en-US/Educati...LED/LFI 2013 Are We There Yet-V1.01 FINAL.pdf

Variation in LEDs --- See bar graphs on various lamp/dimmer combinations http://www.lrc.rpi.edu/programs/solidstate/assist/pdf/assist-TechNote-Dimming-InrushCurrent.pdf

A Triac spec sheet that says little about very short peak current limits. http://www.onsemi.com/pub_link/Collateral/BTA12-600CW3-D.PDF
These short time peak limits will be a function of average chip temperature, instantaneous juntion temperature, and instantaneous current density.

.
 

James L

Senior Member
Location
Kansas Cty, Mo, USA
Occupation
Electrician
I don't know how to say it more plainly, but here's another attempt:

72 lamps x 9 watts apiece = 648W.

There does not appear to be a dimmer that will handle this.
If I may ask, what kind of room has 72 lamps?

Where I'm going is....do these lights get turned on and off frequently? Are they on all the time?

How many times will there be inrush?

I wired a basement last year, put 172 watts of led on a standard 600/150w dimmer and have had no issues.

Probably in part because they don't get turned off and on. They stay on
 
Status
Not open for further replies.
Top