New constant-current, constant-voltage LED driver

PhotonBoy

Flashlight Enthusiast
Joined
Mar 11, 2003
Messages
3,304
Location
Annapolis Valley, Nova Scotia, Canada http://tinyu
EDN article

"LEDs usually take their drive from a constant dc-current source to maintain constant luminescence. Most dc/dc converters, however, deliver a constant voltage by comparing a feedback voltage to an internal reference via an internal error amplifier. The easiest way to turn a simple dc/dc converter into a constant-current source is to use a sense resistor to convert the output current to a voltage and use that voltage as the feedback. The problem is that 500 mA of output current with a 1.2V drop—the typical reference voltage—in the sense resistor incurs relatively high power losses and, thus, a drop in efficiency.

One approach is to use an external op amp to amplify the voltage drop across a low-value resistor to the given reference voltage. This method saves converter efficiency but significantly increases the cost and complexity of a simple converter by using additional components and board space. A better approach is to use the LT1618 constant-current, constant-voltage converter, which combines a traditional voltage-feedback loop and a unique current-feedback loop to operate as a constant-voltage, constant-current dc/dc converter. Figure 1 (PDF) shows the LT1618 driving a 1W, white Lumileds ( http://www.lumileds.com ) LXHL-BW02 Luxeon LED...."
 

balrog

Enlightened
Joined
Oct 28, 2003
Messages
438
Location
Sydney, Australia
Thanks for posting that article. Any idea how much efficiency is lost by using the LT1618's constant-current circuitry?
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
Not exactly new news. The LT1618 is the workhorse guts of a number of products familiar to CPFers including the ARC-LS, the Badboy, Downboy, and Wizard drivers.
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
[ QUOTE ]
balrog said:
Thanks for posting that article. Any idea how much efficiency is lost by using the LT1618's constant-current circuitry?

[/ QUOTE ]
About 2% driving a single die white emitter such as a Lux I or III. It varies slightly with Vf and drive current.
 

cgpeanut

Enlightened
Joined
Jan 7, 2004
Messages
660
Location
tennesse, usa
Re: New constant-current, constant-voltage LED dri

Thanks for the article Doug, /ubbthreads/images/graemlins/goodjob.gif the author mentions the use of the Iadj (current adjustment) pin which the arc-ls board does not use. Injecting a PVM signal to this pin will allow you to dim the led during normal operation. Does this mean an overdriven arc/BB/DB will not work? Wonder at what frequency the PVM should be set at to drive the LT1618 Iadj pin to dim or to go bright? I know our eyes cannot distinguish frequencies above 42Hz (flicker rate) so a PVM above 42Hz should be ok right?
 

CM

Flashlight Enthusiast
Joined
Sep 11, 2002
Messages
3,454
Location
Mesa, AZ
Re: New constant-current, constant-voltage LED dri

The Iadj pin is used to reduce the output from the nominal bias. So if the converter is set to 500mA, tickling the Iadj pin will reduce the current to the LED.

Now what they need to do is make a synchronous buck/boost converter with a current sense amplifier built in like the LT1618. There is a buck/boost converter made by LTC also that can achieve over 95% efficiency. It uses four FET's (two on each side of the inductor) so that it can be made to operate in buck or boost mode (with buck only or boost only efficiency) with the chip taking care of the transition. Some buck/boost converters have mediocre efficiency because they actually use a boost converter that is referenced to the input voltage instead of ground to be able to operate above or below the load voltage. It typically cost 15% in efficiency right off the bat doing it this way with efficiency going down hill from there the load demand increases or the battery voltage decreases. Several lights we know of use this trick to attain buck/boost operation without resorting to SEPIC's using a cheap switch mode boost converter and a single inductor. We need someone like dat2zip or Newbie to try out this chip or find something similar since the pervasive use of Li-Ions with high power Lux III's will dictate a buck/boost converter with high efficiency.

CM
 

billw

Enlightened
Joined
Nov 2, 2002
Messages
295
Location
SF Bay Area
Re: New constant-current, constant-voltage LED dri

So just what is this "flash" capability I keep seeing the chip
vendors touting? I don't recall seeing any "white LED flash"
devices...

BillW
 

NewBie

*Retired*
Joined
Feb 18, 2004
Messages
4,944
Location
Oregon- United States of America
Re: New constant-current, constant-voltage LED dri

Luxeons, check out the app notes on the LumiLEDs site from a couple of years ago on it, and their new section on their site.
 

RobD

Newly Enlightened
Joined
Jul 6, 2003
Messages
40
Re: New constant-current, constant-voltage LED dri

I took a quick look at the Linear chip and it the efficiency is anything to rave about.
What I want is 90+ with constant current from 2.7 to 4.2 volts.
Has anyone done it? I was thinking of an 'H' bridge with low rds(on) mosfets and without secondary side rectification losses
 

PsycoBob[Q2]

Enlightened
Joined
Apr 9, 2002
Messages
518
Re: New constant-current, constant-voltage LED dri

A nice plus side of Li-ion cells the the ones Pila market is that the buck/boost circuitry doesn't need to be held to the old standard of AA battery tubes.

Don't most MOSFET's require a fairly high voltage to ge the gat to work reliably? I remember having problems finding one that would work below 4v. As always, we want better (and brighter) lights, and I a 3w Li-ion is on my want-list.
 

RobD

Newly Enlightened
Joined
Jul 6, 2003
Messages
40
Re: New constant-current, constant-voltage LED dri

Yes and no. Mosfets are trickier than they appear. I like running 'N' types as high side switches using charge pumps to get the gate voltage up to where the conduct fully. Check out the Micrel MIC5014. Saves a lot of parts but it isn't that fastyest on the block. The newer 'P' types are getting better rds(on) everyday.
My thinking is that I'll build a bigger flashlight to stuff the parts in if I can get the constant current and efficiency I need.
Right now I'm doing a double Li-Ion rechargeable light where it runs from one battery and then I switch to the other battery when the first one drops to 2.7 volts. The charger is built in and I'm using the Linear LTC4052 as a charge regulator with excellent results.
 
Top