Home    Bloggers    Messages    Polls    Resources   
Tw  |  Fb  |  In  |  Rss
Charlotte Erdmann

New IC for Driverless LEDs

Charlotte Erdmann
Page 1 / 2   >   >>
Adrian
Adrian
4/14/2014 2:47:16 PM
User Rank
Photon
Re: The devil's in the details
Ron, as soft as I can:

a-> take your time, read some patents, see a movie, romance a girl;  will find plenty of people doing half wave. My message is: as a devil's advocate yes, the switching segments solution is a major breaktrough in terms of flicker compared with simple half wave/full wave rectified direct LED solutions still in production and still employed in commercially available luminaires.

b-> "my" no light time is indeed dependant on string length. For even strings, the more strings the shorter they are; so the no light time. But this have impact on electric efficiency. Not very complicated, but still. Non linear dependance so, counter-intuitive. The 3 strings commercially available solutions turning on @ some 200Vdc exhibit efficiencies around 97% at the expense of the 3.7ms no light time. Low no-light time [0.5ms] solutions need 8+ segments and their efficiencies can "barely" reach 90%. The name of the game is trade off.

c-> I am not assuming anything. Remember, the assumption is the mother of all failiers. Look carefully to the series resistors [which are sensing resistors in ohmic region] and compare their value with the output impedance of current sources ranging from ohm to kohms while the voltage drop vary from 0 to 70V. Just do the math eyes wide closed.

d-> slow switching to match EMI standards = sweet the slope[s] to some 30microsec. These 30microsec as compared to "standard" nanosec MOSFET switching times increase the power dissipated by switching element by below 1%. Yes, it came with that tiny price. The beauty of life is there is no free lunch :D Yes, I am seeking elegance, sometimes at any price

e-> is a free world, you can care [or not] about anything you like. This particular solution is veeeeery similar by all means the TI solution... including but not limited to the efficiency

f-> try to look at: Figure 1 in the article and page 5 on the datasheet./ You will notice the symbol of an active element [a bipolar transistor]. In fact the series current source is built by a small power IGBT or just a MOSFET and a square box named "control box". So, this active semi + the series sensing resistor makes a current source whose effect can be seen by bare eye on figure 1, where the LED current is depicted as a series of horizontal lines.. The current limiting is dynamic, done by current sources.

Agree, the implementation is very poor. As this poor implementation is veeeery similar to TI one, tell them about.

Afritgo, untill seing CEOs, you have to endure engineers. Oops, Rons.

 

 

50%
50%
goafrit
goafrit
4/13/2014 5:00:19 PM
User Rank
Photon
Re: Yawn
I do question why it is okay to call nameless faceless people off the sight "weasels", yet referring to someone on the sight as technically arrogant and ignorant (which arguable was factual) is not allowed? 

Good comment . It was not technically civil. There are ways to make points in ways that support any ecosystem.

50%
50%
afritgo
afritgo
4/13/2014 3:56:43 PM
User Rank
Luminaire
Re: Yawn
@Ron>> Expertise, brutal honest technical or business analysis, no holds barred. Even our cartoons can be hard hitting at times.

I hope you are not a judge in the day and LED expert in the night. That said, I am not sure this forum is an IEEE Journal where we need that level of analysis. This is a blog where you share the high level stuff to a cross-section of audience. When you move deep into your analysis, you lose the passers-by. Blogs are not Nature, IEEE Transaction Journals, etc that require months to plan and write. This a general LED forum that anyone can come and learn something. I will like the CEOs to come here even as the engineers stay around. There needs to be a balance.

50%
50%
semiman
semiman
4/12/2014 8:51:17 PM
User Rank
Candle
Re: The devil's in the details
The require soft switching to pass FCC makes a minimal on efficiency.

50%
50%
Ron Amok
Ron Amok
4/12/2014 2:46:34 AM
User Rank
Blogger
Re: The devil's in the details
a) who on earth is doing half wave rectification for LED lighting SYSTEMS? You are saying full wave rectification produces less flicker than half wave. The was reported in a 1930 paper by Frau Heifer's Institute of the Painfully Obvious. full wave/half wave has NOTHING to do with switched LED strings.

b) your "no light" time is now dependent on string length

c) You are assuming perfect zero ohm bypass switches in your theoretical efficacy numbers. Look carefully, there are series resistors to limit current. Do the math, the efficacy is less than 80% - it has to be, as those switches are limited to 70V across them (no more than 20 LEDs - you CANNOT have 32 LEDs as you postulate with this supplier) or they go to zero power dissipation and the system then will have to rely upon SEDs (Smoke Emitting Diode). I believe my math is correct FOR THIS IMPLEMENTATION.

d) Slow switching comes at a price - power dissipation. Operating your switches in the linear region creates heat - throw your 90% efficacy out the window. The beauty of SSL design is that THERE'S NO FREE LUNCH - as soon as you get smug about an elegant solution, you get bit in the a$$ by something else in the system because of what you did.

e) I don't care about "such switching solutions" - I know those numbers. I analyzed THIS ONE, saying it was lame compared to the TI solution, which uses 100V switches for starters, and uses binary weighted switching and LED strings.

f) Show me the current sources, you speak of, in THIS implementation. There aren't any - current limiting is done here with resistors. The analysis is of this one, not of the architecture as an academic paper. This one leaves a LOT to be desired and is a VERY POOR implementation, IMO. Both technically and beancounterally.

50%
50%
Nancy Golden
Nancy Golden
4/11/2014 3:15:35 PM
User Rank
Luminaire
Re: Yawn
Sigh...I have to agree that there is a natural dichotomy between engineering and marketing:

My posting was a generic marketing vs engineering bashing, not aimed at anyone in particular - at least that was the intent.

Every engineer feels this way at one time or another, and sometimes more often than not...


I just wanted to reiterate that our dear Ron's statement was not a personal attack on the contributing editor -while brilliant, Mr. Amok is not known to be tactful at times and thus his posts may be misconstrued. Nor is he attacking a female writer - while I am female and have felt the brunt of his wit on more than one occasion - he does not reserve it for one gender or ethnic group (well, maybe Texans...) but is pretty fair about unleashing his words on anyone.



@Keith - I am with you - much to learn here as an observer watching the debates...


50%
50%
Davidled
Davidled
4/11/2014 12:10:59 PM
User Rank
Luminaire
EL01 Chip – AC source
When checking the datasheet in the attached link of blog, datasheet illustrates only one test circuit using DC voltage, not AC source. I wonder how the chip made the testing. Most time, customer would like to review the datasheet in detail before purchasing the product.

50%
50%
kdawson
kdawson
4/11/2014 10:22:32 AM
User Rank
Editor
Re: what do I think?
Where did you got the 72?

Made it up. I don't have the background to do the calculations you folks are employing, so I think I will sit back and learn what I can while you debate the issues.

50%
50%
Adrian
Adrian
4/11/2014 3:20:17 AM
User Rank
Photon
Re: what do I think?
Keith, no and no.

No need to be a circuit designer. If read the datasheet, on page 5 will get EXPLICIT indication that with 500mm2 heatsink the drain current can go max to 88mA.

At this current, the whole schematic [72? LEDs + 4 ICs] sink some 10W. If want 20W need to put another COMPLETE schematic i.e. 2 x 72LEDs + 2x4ICs, if want 100W need to put 10 complete schematics i.e. 10x72LEDs and 10x4 ICs.  It is like saying if need a 70W luminaire you can use 10 x 7W bulbs; this is way different from using a single 70W device

note: in there are 72LEDs for 230Vac operation, then the individual voltage drop on each LED shall be some 3.75V; this is pretty un-likely. Where did you got the 72?

50%
50%
Adrian
Adrian
4/11/2014 3:03:02 AM
User Rank
Photon
The devil's advocate
Ron: let me be that advocate. [disclaimer: pls excuse my poor English]

when comparing the time domains switched segments solutions with the half cycle or bridge rectified direct LED solutions yes, a lot of licker get's resolved. In the half cycle simple rectified led string the no light is some 10+ms and the flicker index above 0.7; in the bridge rectified solution the no light time goes to some 5ms and the flicker index to some 0.5

So, the described solution is indeed a flicker fixing solution. Point is ALL segment switching solutions exhibit pretty the same parameters, dramatically decreasing the flicker.

Other topics:

On 230Vac there must be some 90LEDs in series, does NOT matter if you separate them in 10LEDs per string, or 18LEDs per string, 32LEDs per string, or make other un-even combinations; the total voltage drop must stay between 270 and 280V.

THD is poorly related to the number of strings! can get below 10% by just 2 strings!

EMI is not of concern if the strings switching is done slow [and they are switched slow]. I have measured many 3-4 segments solutions which exhibited noise@150kHz well below 55dBuV.

On electric efficiency you are a bit wrong. I can prove [and math demonstrate] the efficiency of such string switching solutions easily go to 98%. Of course, at the expense of PF [and flicker]. But they do. A reasonably balanced design can offer efficiencies in excess of 90% for no light time below 0.5ms and flicker indexes below 0.24

The most of the [electric] power is wasted NOT in the series resistors, BUT in the active elements [transistors] consisting the current sources. Power lost in the resistors is really marginal.


The suggested solution have HUGE margins for spikes and surges. Depending how fast are the ICs [the current sources], most of these spikes will be absorbed by/will drop on them. In the real life I have seen many more bridge rectifiers fried by grid spikes than current sources.. Fuses do not save the bacon, they just protect the grid in amonte!


What bothers is the triumphalistic tone on novelty, while these particular chips were around since .. 2+ years.

50%
50%
Page 1 / 2   >   >>
More Blogs from Charlotte Erdmann
Those who originate a technology aren't always the ones to enjoy its revenues.
Transparent and translucent optical ceramics offer the prospect of cooler, longer-lasting, and more efficient white-light LEDs.
The sort of lighting that employees prefer, and that is best for them, is most often honored in the breach.
It's not just about color rendering. It's about control and respect.
The science behind how gradual light helps us to awaken.
flash poll
follow us on twitter
like us on facebook
All LED Lighting
About Us     Contact Us     Help     Register     Twitter     Facebook     RSS