Suggestion about led driver

Hi to all, i want to regulate constant current in a single led like this (Luxeon LXML-PB01-0030):

formatting link

I'm looking for a led driver with these features:

  1. No PWM output, only constant current in led
  2. Easily drived by a microcontroller (better in PWM)
  3. With a way to measure led current (so micro can automatically tune that)
  4. Cheaper and compact

Any help? Thanks in advance

Reply to
lionelgreenstreet
Loading thread data ...

lionelgreenstreet schrieb:

Hello,

if you want no PWM at the driver output, you need a linear analog driver. If you want to control the driver by a PWM signal from the microcontroller, you have to convert the PWM signal to a analog voltage with a suitable lowpass filter.

Bye

Reply to
Uwe Hercksen

Yeah, pretty easy to do with discrete parts, maybe a dual op-amp - one to amplify the voltage across a shunt resistor and the second to control the output driver-- then you could get current limiting (even if the micro PWM floats or goes to Vdd) and a reasonable signal to measure the current using an on-chip ADC if you want to monitor it.

With a 5V nominal supply and maybe a 2.7 ohm 1W sense resistor the worst case driver dissipation is around half a watt, so a TO-263 type MOSFET should be fine.

Reply to
Spehro Pefhany

Thanks for your answer... do you know any integrated circuit that can be used? I need an output linear current (no output PWM current) dimmable up to

0.8A. Thanks
Reply to
lionelgreenstreet

What power supplies do you have available? ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

3.3V/5V/12V

3.3V is better (is just one led)

Reply to
lionelgreenstreet

But the forward drop is very close to the supply value.

Perhaps use a power MOSFET, set up with an OpAmp as a controlled current source? ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

I've recently used this current source for a blue LED driver, where I had an unregulated supply and had to control the current reasonably well with minimum head room (Li-Ion battery). The LED current is set by RLED (with .6V or whatever across it). I'd have to think a little about how to tune it dynamically.

VCC VCC + + | | .-. V ->

| | - 10K | | | '-' | | ||-+ + ||

Reply to
krw

I just looked at the datasheet and a comment you made later on that you'd prefer to work off of a 3.3V supply. The sheet says that LXML-PB01 parts are spec'd for 2.55 min, 3.08 typ, and 3.51 max when running at 350mA (Table 3.) At 700mA, this is a typical of 3.30V. All with the thermal pad at 25C. (As temp goes higher, the voltage drops.) Elsewhere, you also write that you want this to work up to 800mA. The upshot here is that even if you assume you get exactly typical responses here, with 3.3V typical at 750mA, you can expect ZERO voltage headroom from a 3.3V operating voltage rail. That provides nothing to work with for the circuit.

First off, that kind of tight headroom (assuming you can relax your specs somewhere) suggests a current mirror to me. That would operate open loop without some feedback method, but I'm not convinced you need that kind of accuracy, yet. n any case, I don't know what you really want to control. If it is light output, you might be better off with a detector than controlling current, itself. LEDs are notorious for optical output drift over time AND temperature and observing the current won't really get at that. And if you can accept that kind of appoximation, it's quite possible that open loop operation of a current mirror is good enough, too.

Anyway, I can't imagine you doing any better than about 3V across the LED with a 3V supply and pretty much saturated current mirror. And at 3V, you are NOT going to get 800mA into that LED. Not going to happen. Maybe 300-350mA. That's it.

With 5V, you are probably safe. Roughly speaking, from the specs, I get a model that looks like: V = 2.86 + 0.63*I. This means at 800mA, V=3.36V. That's "typical." If we use the worst case spec, which is 430mV higher, it's V=3.79V. Call it 3.8V. This gives 1.2V headroom, which is okay. It does suggest that you will be dissipating as much as 1.75W in the driver, so keep that in mind, too. (This uses the Vmin for the LED, just to be safer about it.)

Anyway, here's a thought that is cheap to do and cheap to bench test. The BD136 is readily available (in stock) and not too expensive. They will need a little bit of heat sinking at 1.75W each, though. They can only go to about 1W (1.25W if ambient is 25C), without it.

Assuming 1.2V headroom, I'd use 200mV for negative feedback (Rneg, I'll mention that later.) That leaves 1.0V for Vce of Q2. The Vbe near that area looks like 0.9V. This means the Vce=1V keeps it pretty close, but maybe just out of, saturation. Beta is about 60 at 800mA with Vce=2V, so the base drive would be about 13mA. But Vce=1V and looking at the saturation curves they provide, which look dire near 1A and beta=20, I'm betting the beta will be closer to 30 than

  1. So call it 800/30 = 27mA, to be safe.

Q1 can be set up by Rx to require a lot less Ic. Roughly, Rx can be set by looking at charts. Q2's Vbe is expected to be at least 0.9V. Leaving room for 1.0V, this means that Rx*Ic plus Q1's Vbe at that Ic must be the same. Rx=(1.0-Vbe)/Ic for Q1. However, we only have about 2.3V room for Rset (assume 3.2V at the filtered output of the PWM, 200mV for Rneg, and maybe 700mV for Q1's Vbe.) Let's set Q1's Ic to about the same as the base drive for Q2, namely 27mA. At

27mA for Ic, I see a typical of 0.68V for Vbe. If I allow for the possibility of a higher Vbe, say another 50mV worth, then this makes Rx=10 ohms. (You may which to tinker with it, when built.) This means Rset=2.3V/(2*27mA) or 42 ohms. Make it 39. With Q3's emitter current now about 55mA, plus another 800mA from the LED, this is about 855mA. A 0.22 ohm resistor works for that designed 200mV drop there on Rneg. R3 provides an always-conductive path. Q4 and R4 let you short out the LED for fast turn-off, if you need it. Otherwise drop out both parts. I picked an arbitrary R1+C1 for an assumed pulse period of 10kHz and ripple that looked small enough. It's arbitrary, though. R5 may be unnecessary, but I figured pulling that node off in case Q3 isn't sinking at all. Drop it out, if you want to.

+---------+--------- (+)LED

|-----,

|
|
|
|
|
+-----------+---------+--------- (-)LED
|
|
|
\

/ R4

\ 1000

/
|
|
|

--------------------------------------------------------------------'

Simplified version.

Just a thought.

Jon

Reply to
Jon Kirwan

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.