10watts for all 4, will it work

[Zero999]

I would like to disagree to most of the LED bulb horror stories told here and ask everybody to think how it could work, not how it can fail. - Because after all original poster is doing good job - he is trying to make our world better place.

My two cents: 10W power is so low that I would not bother about sequential switching, any LED switching at all. I would run all the LED bulbs in parallel - to save on wiring in many installations and create 10W PWM-dimmable AC square wave inverter with input current limiter. Current limiter as safeguard against inverter overload and output short. Also it is worth to consider optional frequency control in 50-400Hz(?) range. Maybe it can help with some (too smart or too dumb) LED bulbs or reduce flicker. What's more or less clear here - lot of testing with various LED bulbs needed.

Reducing the duty cycle using PWM might not work. If the DC:DC converter inside the LED has a smoothing capacitor, it'll take large current spikes on the rising edge of the pulse, with the magnitude and length of the pulse, depending on the impedance and the voltage on the capacitor. If the output of the internal DC:DC converter is constant current, then the power consumption of the LED will not change, until the PWM drops below a point. This might put extra stress on the lamp, leading to premature failure.

If the LED's DC:DC converter has no filter capacitor, then it will probably work, but it will flicker at double the mains frequency, unless the inverter works at a high frequency.

How do you propose PWMing the AC? It's not possible to do it in increments of less than double the mains frequency with SCRs. Another option is to use two MOSFETs back-to-back and chop the entire AC waveform at a much higher frequency, but that can lead to issues with RF interference and high surge currents, as mentioned above.

Phase control is an other option, in which case leading edge should be used with LEDs, rather than trailing edge, but again MOSFETs need to be used, rather than SCRs.

[ogden]

> Reducing the duty cycle using PWM might not work.

My guess that it shall work with dimmable LED bulbs.

> If the output of the internal DC:DC converter is constant current

I would like to see any <= 10W low-end LED bulb (where OP is aiming at) which have constant current converter inside. Again we CAN find particular LED's which will not work with such approach, BUT again - we shall use common sense here too.

> This might put extra stress on the lamp, leading to premature failure.

10W total power, 2.5W for each bulb. Again I would like to see - does LED bulbs really fail. After all hard edges can be easily filtered on the inverter output at let's say 1KHz cutoff freq or so.

> but it will flicker at double the mains frequency, unless the inverter works at a high frequency.

Didn't I say about 50..400 Hz frequency control which could reduce flicker?

> How do you propose PWMing the AC? It's not possible to do it in increments of less than double the mains frequency with SCRs.

I did say "I would not bother about sequential switching, any LED switching at all". My proposal is to PWM-control inverter.

> Another option is to use two MOSFETs back-to-back and chop the entire AC waveform at a much higher frequency

Too complex. Better chop inverter at higher frequency - as I proposed initially.

> Phase control is an other option, in which case leading edge should be used with LEDs, rather than trailing edge

Square wave inverter cannot have any leading or trailing edge phase control because it's square wave.

[edit] Just looked at available PW controllers. UCC28089 looks like good choice. Additional op-amp and potentiometer needed to augment current sense for PWM-control.

[ogden]

[edit] Just looked at available PW controllers. UCC28089 looks like good choice. Additional op-amp and potentiometer needed to augment current sense for PWM-control.

Obviously alternative option is digitally controlled inverter. Attiny24 is good enough for the job, especially considering it's 20x gain amplifier for ADC which is perfect match for current sense measurement.

[Zero999]

> Reducing the duty cycle using PWM might not work.

My guess that it shall work with dimmable LED bulbs.

That depends on the bulb. Some dimmable LED bulbs look at the incoming waveform and adjust the LED current according.

> If the output of the internal DC:DC converter is constant current

I would like to see any <= 10W low-end LED bulb (where OP is aiming at) which have constant current converter inside. Again we CAN find particular LED's which will not work with such approach, BUT again - we shall use common sense here too.

Plenty, just look up LED driver ICs, using a search engine. Here's one example. Many more exist.
http://www.mouser.com/ds/2/302/SSL21082AT-843512.pdf

> This might put extra stress on the lamp, leading to premature failure.

10W total power, 2.5W for each bulb. Again I would like to see - does LED bulbs really fail. After all hard edges can be easily filtered on the inverter output at let's say 1KHz cutoff freq or so.

Possibly. It would certainly exceed the ripple rating of the input capacitor. Another issue is, if the LED bulbs are in standard fittings, people might replace them with other LEDs which aren't compatible. Look up capacative dropper LED, which certainly will fry if operated from too higher frequency.

> but it will flicker at double the mains frequency, unless the inverter works at a high frequency.

Didn't I say about 50..400 Hz frequency control which could reduce flicker?

Good idea.

> How do you propose PWMing the AC? It's not possible to do it in increments of less than double the mains frequency with SCRs.

I did say "I would not bother about sequential switching, any LED switching at all". My proposal is to PWM-control inverter.

The trouble is, he's already has the inverter and doesn't seem to be willing to redesign it or accept any help with the circuit, as he won't post it because it's proprietary information.

> Another option is to use two MOSFETs back-to-back and chop the entire AC waveform at a much higher frequency

Too complex. Better chop inverter at higher frequency - as I proposed initially.

The only option if changing the inverter is not an option.

> Phase control is an other option, in which case leading edge should be used with LEDs, rather than trailing edge

Square wave inverter cannot have any leading or trailing edge phase control because it's square wave.

Actually, if could work, irrespective of the waveform, just delay the turn on, after

[edit] Just looked at available PW controllers. UCC28089 looks like good choice. Additional op-amp and potentiometer needed to augment current sense for PWM-control.

Good, if you're redesing the inverter.

[ogden]

>Plenty, just look up LED driver ICs, using a search engine. Here's one example. Many more exist.
>http://www.mouser.com/ds/2/302/SSL21082AT-843512.pdf

Existence of drivers does not prove that they are used in low end LED bulbs. You can more or less count on that. BTW you are showing one which for sure will work with "PWM dimmable square wave inverter" 

>Possibly. It would certainly exceed the ripple rating of the input capacitor.

Ripple rating of the input capacitor?  BS. This lesson comes from bigclive LED academy? Crude LED bulbs have that poor series resistor for exactly that reason. - To reduce EMI and obviously ripple [inrush] current spike when minimum forward voltage of LED string is reached (check IV curve of the led). Most "dumb" LED bulbs conduct quite late in the AC cycle. Also we shall take in account that peak voltage of square wave inverter will be just 0.7x of AC mains peak for same RMS voltage output, so current spike difference could be minimal. In short - don't bother about wearing out capacitor or series resistor.

>Look up capacative dropper LED, which certainly will fry if operated from too higher frequency.

I would not be so certain. 400Hz is not that high frequency after all. Also remember about power budget we are talking here about. Most likely everything will be underpowered and very far from frying.

>The only option if changing the inverter is not an option.


Then somebody else will save the world

[orolo]

I would not be so certain. 400Hz is not that high frequency after all. Also remember about power budget we are talking here about. Most likely everything will be underpowered and very far from frying.

Not high frequency at all. Take this cheap "3W" dropper bulb from Big Clive's Academy:
 


At 50Hz, the impedance of the 820nF is 3882 Ohms, so the RMS current is 52.5mA, and the 30V LED string dissipates a puny 1.6W, certainly underpowered for an alleged 3W bulb. Now crank the frequency up to 400Hz. The impedance is 485 Ohms, the current 359mA, and the power 10.68Watts. A x10 6.6x increase in power, and 3.3x more power than the rating. I'm sure Big Clive would approve  .

[ogden]

At 50Hz, the impedance of the 820nF is 3882 Ohms, so the RMS current is 52.5mA, and the 30V LED string dissipates a puny 1.6W, certainly underpowered for an alleged 3W bulb. Now crank the frequency up to 400Hz. The impedance is 485 Ohms, the current 359mA, and the power 10.68Watts. A x10 increase in power, and 3.3x more power than the rating. I'm sure Big Clive would approve  .

As you did your homework, then I have no other option than to agree. - In case we talk about 100% duty cycle. But as I suggested not only PWM control but also input current (effectively output power) limiter, then RMS figures will be ok. Thou it is not good to 10x exceed current and I shhall agree that 400Hz is too much. Specs shall be revised 

[dave_j_fan]

I am at a total loss to understand the point of this.  Rather than trying to dim a 10W lamp so it only consumes 2.5W, why not start with a 2.5W lamp?

How are you intending to switching a TRIAC on/off at a frequency greater than the zero crossings of the 50/60Hz line frequency?

Perhaps the inverter generates a higher frequency than the normal mains?

yes it does,inverter is at leat 10x more ....

hero999 inverter scheme is  fined tuned one . but the switcher i will draw later

[dave_j_fan]

I would like to disagree to most of the LED bulb horror stories told here and ask everybody to think how it could work, not how it can fail. - Because after all original poster is doing good job - he is trying to make our world better place.

My two cents: 10W power is so low that I would not bother about sequential switching, any LED switching at all. I would run all the LED bulbs in parallel - to save on wiring in many installations and create 10W PWM-dimmable AC square wave inverter with input current limiter. Current limiter as safeguard against inverter overload and output short. Also it is worth to consider optional frequency control in 50-400Hz(?) range. Maybe it can help with some (too smart or too dumb) LED bulbs or reduce flicker. What's more or less clear here - lot of testing with various LED bulbs needed.

world better place.--yes .... we will save on led lights power usage ,atleast cut bills to 1/3
poor gets backup power,solar lamps need smaller battery ..so on

but pwm dimming is of no use as hero999  outlined

the SCR gives flicker to 5watt LEDS , but 4watts work flawlessly. I will try with triac

REading this https://www.digikey.in/en/articles/techzone/2012/oct/avoiding-flicker-in-led-lighting-applications

[Zero999]

>Plenty, just look up LED driver ICs, using a search engine. Here's one example. Many more exist.
>http://www.mouser.com/ds/2/302/SSL21082AT-843512.pdf

Existence of drivers does not prove that they are used in low end LED bulbs. You can more or less count on that. BTW you are showing one which for sure will work with "PWM dimmable square wave inverter" 

It's not that simple as PWMing the power. Read the datasheet, especially the part about  t_on control.

At 50Hz, the impedance of the 820nF is 3882 Ohms, so the RMS current is 52.5mA, and the 30V LED string dissipates a puny 1.6W, certainly underpowered for an alleged 3W bulb. Now crank the frequency up to 400Hz. The impedance is 485 Ohms, the current 359mA, and the power 10.68Watts. A x10 increase in power, and 3.3x more power than the rating. I'm sure Big Clive would approve  .

As you did your homework, then I have no other option than to agree. - In case we talk about 100% duty cycle. But as I suggested not only PWM control but also input current (effectively output power) limiter, then RMS figures will be ok. Thou it is not good to 10x exceed current and I shhall agree that 400Hz is too much. Specs shall be revised 

Yes that's pretty much what I was getting at.

No one here is being negative for no reason. There are reasons why people here have recommended things such as a higher power inverter or 12VDC LEDs, but the original poster is either, unable or unwilling to implement them.

[dave_j_fan]

no worries with lowering of impedance here , smps bulbs only.people will  be instructed to buy standard LEDs all are SMPS
 ... we specify make and watt
there are Khz frequency inverters in market , running LEDs for long periods.


power-slicing : i call it .....its the TDM  ( time division multiplexing) of  AC 

Personally i try to avoid complexity in implementation. Bringing special chips for the job are last options.
I will use just 4017like  sequencing ICs so that parts are available everywhere .

>Plenty, just look up LED driver ICs, using a search engine. Here's one example. Many more exist.
>http://www.mouser.com/ds/2/302/SSL21082AT-843512.pdf

i already have non-isolated driver based bulbs with similar chips . they too will be used in tests .

i may be noob here but Qn arises
where PWM dimming has come here ? we dont want to dim the lamps  and reduce power .
we want same light for a much lesser power

[mikerj]

>Plenty, just look up LED driver ICs, using a search engine. Here's one example. Many more exist.
>http://www.mouser.com/ds/2/302/SSL21082AT-843512.pdf

Existence of drivers does not prove that they are used in low end LED bulbs. You can more or less count on that. BTW you are showing one which for sure will work with "PWM dimmable square wave inverter" 

>Possibly. It would certainly exceed the ripple rating of the input capacitor.

Ripple rating of the input capacitor?  BS. This lesson comes from bigclive LED academy? Crude LED bulbs have that poor series resistor for exactly that reason. - To reduce EMI and obviously ripple [inrush] current spike when minimum forward voltage of LED string is reached (check IV curve of the led). Most "dumb" LED bulbs conduct quite late in the AC cycle. Also we shall take in account that peak voltage of square wave inverter will be just 0.7x of AC mains peak for same RMS voltage output, so current spike difference could be minimal. In short - don't bother about wearing out capacitor or series resistor.

You seem to be somewhat confused.  It's true the very cheap, low power lights use capacitive droppers, but when you are talking about a 10Watt LED lamp for a conventional lamp fitting then it is almost certain to be using a constant current switching regulator.  I took apart some cheap GU4 bulbs recently and even those used a switching regulator.

If it has a bridge rectifier and smoothing cap, then any attempts to dim the LED through e.g. phase control or PWM will greatly increase the peak current through the rectifier and the ripple current through the smoothing cap.

The OP is not saving the world, he's trying to  create a very complex solution to what appears to be a simple problem, and refuses to consider any other way of solving it.

[Zero999]

>Plenty, just look up LED driver ICs, using a search engine. Here's one example. Many more exist.
>http://www.mouser.com/ds/2/302/SSL21082AT-843512.pdf

Existence of drivers does not prove that they are used in low end LED bulbs. You can more or less count on that. BTW you are showing one which for sure will work with "PWM dimmable square wave inverter" 

>Possibly. It would certainly exceed the ripple rating of the input capacitor.

Ripple rating of the input capacitor?  BS. This lesson comes from bigclive LED academy? Crude LED bulbs have that poor series resistor for exactly that reason. - To reduce EMI and obviously ripple [inrush] current spike when minimum forward voltage of LED string is reached (check IV curve of the led). Most "dumb" LED bulbs conduct quite late in the AC cycle. Also we shall take in account that peak voltage of square wave inverter will be just 0.7x of AC mains peak for same RMS voltage output, so current spike difference could be minimal. In short - don't bother about wearing out capacitor or series resistor.

You seem to be somewhat confused.  It's true the very cheap, low power lights use capacitive droppers, but when you are talking about a 10Watt LED lamp for a conventional lamp fitting then it is almost certain to be using a constant current switching regulator.  I took apart some cheap GU4 bulbs recently and even those used a switching regulator.

If it has a bridge rectifier and smoothing cap, then any attempts to dim the LED through e.g. phase control or PWM will greatly increase the peak current through the rectifier and the ripple current through the smoothing cap.

The OP is not saving the world, he's trying to  create a very complex solution to what appears to be a simple problem, and refuses to consider any other way of solving it.

I agree. Unless a special type of LED controller is used this won't work.

no worries with lowering of impedance here , smps bulbs only.people will  be instructed to buy standard LEDs all are SMPS
 ... we specify make and watt
there are Khz frequency inverters in market , running LEDs for long periods.


power-slicing : i call it .....its the TDM  ( time division multiplexing) of  AC 

Personally i try to avoid complexity in implementation. Bringing special chips for the job are last options.
I will use just 4017like  sequencing ICs so that parts are available everywhere .

If you want to control the sequence of some lamps, then why not use a microcontroller? It will be more flexible than one of those old ICs and doesn't require as many components.

>Plenty, just look up LED driver ICs, using a search engine. Here's one example. Many more exist.
>http://www.mouser.com/ds/2/302/SSL21082AT-843512.pdf

i already have non-isolated driver based bulbs with similar chips . they too will be used in tests .

i may be noob here but Qn arises
where PWM dimming has come here ? we dont want to dim the lamps  and reduce power .
we want same light for a much lesser power

If it was possible to produce the same light for less power, then don't you think the LED lamp manufacturers would have already done that?

Your scheme will not give you the same light for less power. It will use more power to give you the same light and poorer reliability or slightly less power for much less light, again with poorer reliability.

Simply switching the LEDs on and off at a high frequency will not lead to any increases in efficiency, even if such a scheme were compatible with the integrated driver electronics. The human eye perceives the average brightness of a light flashing at a high frequency. If you flash a 10W bulb at 1kHz, equal time on and off, it will appear to give off the same amount of light as a 5W bulb.

In practise, if you tried to do that with a ready made LED, with a constant current DC:DC converter, the results will be unpredictable. It might: produce the same brightness but draw double the current, causing potential reliability issues, flicker horribly or just be dimmer.

Another factor is the non-linear response of the human eye. A 5W lamp may not appear to be half as bright as a 10W lamp, so even if you think you're successful by using half the power, for a slight drop in apparent brightness, you'll find the 5W lamp would have been a better option.

[mikerj]

i may be noob here but Qn arises
where PWM dimming has come here ? we dont want to dim the lamps  and reduce power .
we want same light for a much lesser power

If you have found a magical (and it would require magic) scheme to quarter the power consumption of an existing LED lamp for the same light output by manipulating the line voltage then you need to get a patent sorted out ASAP.

[Zero999]

world better place.--yes .... we will save on led lights power usage ,atleast cut bills to 1/3
poor gets backup power,solar lamps need smaller battery ..so on

The prices of solar panels and batteries are falling and are becoming a more attractive option, especially in areas with high electricity prices. There are other options, such as wind and energy from waste (look up biogas and gasification) which can be used to provide power on a small scale.

but pwm dimming is of no use as hero999  outlined

the SCR gives flicker to 5watt LEDS , but 4watts work flawlessly. I will try with triac

REading this https://www.digikey.in/en/articles/techzone/2012/oct/avoiding-flicker-in-led-lighting-applications

That article discusses LED DIMMING, using phase control. It will not help you achieve higher efficiency.

[Damianos]

As making basic EI transformer inverter is super easy  i am planning a thing

  i am adding a triac based switcher to make sure the apparent LED load is within 10W in the inverter
  so a bulb is made on at a time and fast switch is done so that every bulb look steady glow but inveter can be used is very low watt


 
   
  Experts please suggest will it work  ?

clarification :::
It is not a DIMMER
it is NOT a XMAS light
I dont want to use 12V LED

just DIY $5inverter should be able to do double duty that is 10W inverter drives 20W

Of course it will do!
You can also expand it. At each output from the switcher, connect another switcher ... and, continuing this, you can have unlimited power/energy to lighten up the entire planet.
But not only this. If you start combining the outputs after a few billions of switchers, there will be plenty of power for trains, factories, homes, etc.
Still, if from some of the outputs re-feed the input, you can save and battery ... After start-up, you can remove it and your invaluable inverter!
Watch out! There will be many who will try to steal your designs ... ...


To be serious!
Are you insisting that a fast switching, with TRIACs or anything else, can cheat on the rules of physics?

[Connoiseur]

If it was possible to produce the same light for less power, then don't you think the LED lamp manufacturers would have already done that?

Your scheme will not give you the same light for less power. It will use more power to give you the same light and poorer reliability or slightly less power for much less light, again with poorer reliability.

Simply switching the LEDs on and off at a high frequency will not lead to any increases in efficiency, even if such a scheme were compatible with the integrated driver electronics. The human eye perceives the average brightness of a light flashing at a high frequency. If you flash a 10W bulb at 1kHz, equal time on and off, it will appear to give off the same amount of light as a 5W bulb.

In practise, if you tried to do that with a ready made LED, with a constant current DC:DC converter, the results will be unpredictable. It might: produce the same brightness but draw double the current, causing potential reliability issues, flicker horribly or just be dimmer.

Another factor is the non-linear response of the human eye. A 5W lamp may not appear to be half as bright as a 10W lamp, so even if you think you're successful by using half the power, for a slight drop in apparent brightness, you'll find the 5W lamp would have been a better option.

I would be glad to know the lm/W output of such a contraption.


Boy! the OP should watch more free energy videos on YouTube.

[ogden]

You seem to be somewhat confused.  It's true the very cheap, low power lights use capacitive droppers, but when you are talking about a 10Watt LED lamp for a conventional lamp fitting then it is almost certain to be using a constant current switching regulator.  I took apart some cheap GU4 bulbs recently and even those used a switching regulator.

Before you label someone as confused, you shall comprehend what he is talking about. I shall remind that main topic was not "capacitive droppers" as such but exceeding ripple rating of input capacitor and it's consequent failure. Even in case of constant current switching regulators square wave input shall not accelerate failure rate unless design is flawed.

[edit] Knowing target customer and assuming that he will buy 10W LED bulb with current switching regulator inside is kinda... uneducated thinking. If total power provided for 4 bulbs is 10 watts - then what's the point of paying for 4x10W bulbs?

If it has a bridge rectifier and smoothing cap, then any attempts to dim the LED through e.g. phase control or PWM will greatly increase the peak current through the rectifier and the ripple current through the smoothing cap.

If LED is designed as dimmable, it will work. Try to think of solution too, not only just dig up flaws. Actually latter is much easier.

The OP is not saving the world, he's trying to  create a very complex solution to what appears to be a simple problem, and refuses to consider any other way of solving it.

Oh, Mr.Obvious. Do you know concept of sarcasm?

[Zero999]

You seem to be somewhat confused.  It's true the very cheap, low power lights use capacitive droppers, but when you are talking about a 10Watt LED lamp for a conventional lamp fitting then it is almost certain to be using a constant current switching regulator.  I took apart some cheap GU4 bulbs recently and even those used a switching regulator.

Before you label someone as confused, you shall comprehend what he is talking about. I shall remind that main topic was not "capacitive droppers" as such but exceeding ripple rating of input capacitor and it's consequent failure. Even in case of constant current switching regulators square wave input shall not accelerate failure rate unless design is flawed.
[edit] Knowing target customer and thinking that he will buy 10W LED bulb with current switching regulator inside is kinda... uneducated thinking.

The power dissipation in the capacitor/inrush limiting resistor will be higher with a 50% duty cycle square wave, than a sine wave. If the inrush limiter has a negative temperature coeffient, then it might be fine, since the increased current might reduce the resistance enough to avoid the power dissipation getting too high, otherwise it could fry.

See the attached simulation. With a 500Hz 50% duty, 325V (the peak voltage for 230VAC), the power dissipation in R1 (the surge limiting resistor) will be 500mW. When V1 is changed to a 500Hz square wave, the power in R! increases to over 2.5W. I'll attach some screenshots later.

If LED is designed as dimmable, it will work. Try to think of solution too, not only just dig up flaws. Actually latter is much easier.

It might work or it might not. It depends on how the dimming is implemented. With the IC data sheet I linked to earlier, it certainly wouldn't work.

Oh, Mr.Obvious. Do you know concept of sarcasm?

Unfortunately sarcasm often gets lost in the depths of the Internet.

[ogden]

See the attached simulation. With a 500Hz 50% duty, 325V (the peak voltage for 230VAC), the power dissipation in R1 (the surge limiting resistor) will be 500mW. When V1 is changed to a 500Hz square wave, the power in R! increases to over 2.5W. I'll attach some screenshots later.

Did not able to interpret .asc file provided, sorrry. Your beginners error is that square wave inverter will not output "325V (the peak voltage for 230VAC)" but 230VAC RMS voltage which is essentially 230V peak. So, RMS power dissipation in R! in case of square wave will be identical. Where's the catch?

[edit] Actually I was beginner here, apologies. 230V peak of square wave inverter at 100% duty cycle will give same power as RMS 230VAC only for resistive loads. LED with rectifier inside needs 325V peak disregarding it is sine or square wave.

With the IC data sheet I linked to earlier, it certainly wouldn't work.

Detailed explanation please [grin].

[Zero999]

See the attached simulation. With a 500Hz 50% duty, 325V (the peak voltage for 230VAC), the power dissipation in R1 (the surge limiting resistor) will be 500mW. When V1 is changed to a 500Hz square wave, the power in R! increases to over 2.5W. I'll attach some screenshots later.

Did not able to interpret .asc file provided, sorrry. Your beginners error is that square wave inverter will not output "325V (the peak voltage for 230VAC)" but 230VAC RMS voltage which is essentially 230V peak. So, RMS power dissipation in R! in case of square wave will be identical. Where's the catch?

No.

For a start, in order to work properly, the inverter would have to output a voltage of 325V, since that what will be on the capacitor, after the rectification of the 230VRMS mains, which has a peak voltage of 325V. Don't believe me? Connect a 12VAC transformer to a bridge rectifier and capacitor. You'll find that the voltage on the DC side, is equal to about 1.414 times the AC side, minus the diode losses. Even if the voltage were reduced to 230V, the power dissipation would still be higher, than for a sine wave.

With the IC data sheet I linked to earlier, it certainly wouldn't work.

Detailed explanation please [grin].

Did you refer to the data sheet, paying attention to the part about  t_on control, pin 8, the TONMOD pin? The dimming function is dependant on the timing of the sine wave on the input to the rectifier.
http://www.mouser.com/ds/2/302/SSL21082AT-843512.pdf

[kalel]

As (little) as I know, multiplexing does decrease the brightness we perceive, since each LED is only on for some time. Even a single LED switching "on off" faster than we can see will produce less light than a constantly on LED, because it will be some "on and off" output average that we see rather than just "on" output.

As far as capacitive droppers, I'm almost certain I have seen capacitive droppers in cheap claimed 9/10W LEDs. I haven't measured the wattage to see what it outputs really, but they are quite bright, and don't last too long (up to a few years). Sometimes the LEDs get burned (lack of proper heat-sinking?) - they first get darker spots, still work for a while - and finally fail, sometimes the electrolytic capacitor fails (it does get moderately heated, I don't know how a better capacitive dropper design could mitigate this issue).

[ogden]

Why you are using 50Hz for sine and 500Hz for inverter?

[edit] Anyway - you just completely ignore all the other "specs" of the project including 10W power limiting of inverter output, it's duty cycle which for sure will not be 100%, you put hi-voltage 47uF capacitor into low cost LED (!).

Also this:

>Did you refer to the data sheet, paying attention to the part about  ton control, pin 8, the TONMOD pin?

is not detailed explanation but more like trolling.

[dave_j_fan]

Simply switching the LEDs on and off at a high frequency will not lead to any increases in efficiency, even if such a scheme were compatible with the integrated driver electronics. The human eye perceives the average brightness of a light flashing at a high frequency. If you flash a 10W bulb at 1kHz, equal time on and off, it will appear to give off the same amount of light as a 5W bulb.

In practise, if you tried to do that with a ready made LED, with a constant current DC:DC converter, the results will be unpredictable. It might: produce the same brightness but draw double the current, causing potential reliability issues, flicker horribly or just be dimmer.

The point of getting reduced light like 5Wled bulb from 10w  with 1khz frequency is not valid , in khz range the LED gives almost same brightness.I have tested it, with few brands like hevells, philips ..

For the point "flicker" "dimmer"   is also not fully true .
my Jig already achieved same light for 4W LED . I have tested with SCR as active element .
 

yes the 5W led gived a breathing like fliker mey be because SCR have reduced the voltage .

 I have reduced the off period to 1sec and the SCR switch is giving flicker , means
..TRIAC tests will ... hopelfully make the jig work
The discussion is very informative. I am going baby steps way to make the jig flicker free .
Jig will feed from a small diy inverter .... DC current i will measure and find the truth .
 
 remember JOULE ringers running 10LED bulbs ...so possibility is always there
 

[dave_j_fan]

  will see ...

 for your information i have a DC led tube , i added PWM and reached 12hours light ... originally it was giving 3-4hour
 dont bring physics if you dont know pulsed and switched led lighting saves lot energy .

 however i  have less test knowledge on AC so asked experts