Constant current IR LED drive

[TopGunPk]

Hi.

I am trying to make a Constant-Current drive for IR LEDs.
The idea is to be able to drive IR LEDs at a constant current of, say 300mA.
The frequency is 38 kHz, 50% duty cycle.

My simulation shows the results to be OK ( i tried in Multisim 8.0). Upon building it on a perf-board, it does not work. Sometimes it works, sometimes it burns the LED.

Attached is teh Schematic as i made in Multisim for simulation.
I am guessing the Slew rate of the opamp is somehow creating the problem. The opamp i am using is an LM2904.

Now i am testing this thing at a base of 1 kHz as initial test, as i am assuming the Slew rate to be the limiting factor here.

I will appreciate if you can point me in the right direction for this to work.

Shan

[fcb]

Can your LED handle 150mA average?  That seems pretty high.

[TopGunPk]

Thanks FCB.

Well, normal IR LEDs are rated at 100mA, and can handle peak currents of 300mA.

Since IR signals are a burst of information, modulated at 38 kHz (commonly) carrier, that too at 50% duty cycle, so yes it can handle that.

[TopGunPk]

Should i use a Difference amplifier accross the Sense resistor first, and then add that to the loop of the opamp controlling the drive transistor ?

[RobK_NL]

normal IR LEDs are rated at 100mA, and can handle peak currents of 300mA.

Where do you get that from? Please give a link to the datasheet of a "normal" IR LED that shows this to be true.

The ones I checked show maximum peak currents of 200mA @ t_p=100µs and 50% dutycycle. And that is the absolute maximum rating at 25°C.

[TopGunPk]

Vishay VSLB3940

Please see attached datasheet.

[Benta]

You're misreading the data sheet. Your design is less than marginal.
The way to increase output power is to use more than one IRLED.

[TopGunPk]

Well, i intend to use 4x of these LEDs, all being driven by their own constant current source...
these 4x CCS will be fed from one signal...

this signal will be a envelope modded by a 38 kHz carrier

Basically, generated from a microcontroller pin

[TopGunPk]

ALso, many commercial IR remote controls, like TV remotes, tehy apply peak power, because its pulsed... so it does not destroy the LEDs... and achieves a longer range.

Multiple LEDs will give more coverage area / angle, not necessarily more range.

Or am i missing something ?

[RobK_NL]

I have yet to see a commercial IR remote with a constant current source.

[TopGunPk]

You are right, teh commercial remotes are built to a price... dirt cheap. .they dont generally have CCS, another reason is they can live off with a simple series pass current-limiting element (resistor) because most have only 1 LED.

I used this arrangement with 4 LEDs in parallel, each powered from same source, each having its own series pass current limit resistor, BUT it has problems, specially when i fire up less than 4.... OR if one blows out, then the others blow out sooner.

Hence i am trying to design a CCS controlled IR LED circuit, another reason for it is to have a variable input power supply (5V to 24V DC).

Do you thin this is the right direction i am heading in ?

I am using a simple opamp based closed loop circuit that will balance the inputs (inverting input has teh V-drop on a sense resistor against teh curent flowig through it, and non-inverting input is the reference voltage that will be the signal from teh micro).

In steady state, i am not seeing any problems.. its the signal (38 kHz carrier) that makes the circuit not perform.

[tszaboo]

Yes, this LED can handle 300mA. And yes, this is ridiculously complicated for driving an LED. Just connect it to a NUD4001 or a bunch of constant current drivers, and done.

[Benta]

This LED can handle 100 mA continuous and 1 A for 100 us at a duty rate of 10%. 300 mA at a duty rate of 50% is more than that, do the calculation (use RMS values).

[fcb]

Topgunpk, you are obviously over-driving the LED (hence the early deaths). This sort of LED run at 100mA will get hot - effective light output will reduce by perhaps 30% at higher temperature. Have to tried to monitor junction temperature (difficult to do, but you can drill a hole through the epoxy and but a small thermocouple through).

If you feel 'advanced' then you can monitor the forward voltage at a known (low current), perform a temperature calibration and then calculate the actual junction temperature as you come off your pulse train.  This is a problem we had with a client project where we wanted to estimate lifetime for a pulsing LED - we even developed a piece of kit specifically to analyse diodes (DPA, Diode Parameter Analyser).

Don't underestimate the humble resistor to drive an LED, the variation in forward voltage with temperature for the LED you are using is around -5%, far lower than the temperature related drop-off in light intensity and going the opposite way (i.e. the drop in Vforward will compensate a little bit for the drop in light intensity at higher temps).

If you must use a constant current source (what power source are you driving these from?), then I would suggest building a classic constant current regulator (one transistor, one resistor and a perhaps two diodes/resistor to set the base voltage - modulate the base drive resistor), it's fairly stable and you won't have to get involved in loop stability/ringing/overshoot like you do with the opamp solution, etc...

As it sounds like you need wider area coverage (hence 4 LED's?) I would probably use more LED's driven at lower current and with narrower angle lenses.  Alternatively if you have control over the receiver electronics (so you can scope the receiver output) then perhaps reduce the duty cycle - 50% is not necessary to get through to some of the receivers I've tested in the past for a project.

[TopGunPk]

@NANDBlog
Thanks. I just got the NUD4001 part, and tried with it.

I have connected 1 IR LED to the NUD4001 part, and driving at 5V, i am able to feed a captured IR signal on it and get the exact output, as measured through an IR Receiver. The signal also works on my appliance.

However, if i try to power from 24V  -OR-  even 12V, the IR LED comes ON continuously. I am did the calculations as indicated on the NUD4001 datasheet's page 4, and i am seeing the Drop voltage to be too high, resulting in higher Power dissipation as what the device can handle safely. But i do not understand why it is putting the IR LED constantly ON, even though i am driving the transistor on NUD4001 with my ON/OFF Carrier.

[ogden]

@NANDBlog
Thanks. I just got the NUD4001 part, and tried with it.

I have connected 1 IR LED to the NUD4001 part, and driving at 5V, i am able to feed a captured IR signal on it and get the exact output, as measured through an IR Receiver. The signal also works on my appliance.

However, if i try to power from 24V  -OR-  even 12V, the IR LED comes ON continuously. I am did the calculations as indicated on the NUD4001 datasheet's page 4, and i am seeing the Drop voltage to be too high, resulting in higher Power dissipation as what the device can handle safely. But i do not understand why it is putting the IR LED constantly ON, even though i am driving the transistor on NUD4001 with my ON/OFF Carrier.

Hard to tell w/o seeing your circuit schematics

[tszaboo]

@NANDBlog
Thanks. I just got the NUD4001 part, and tried with it.

I have connected 1 IR LED to the NUD4001 part, and driving at 5V, i am able to feed a captured IR signal on it and get the exact output, as measured through an IR Receiver. The signal also works on my appliance.

However, if i try to power from 24V  -OR-  even 12V, the IR LED comes ON continuously. I am did the calculations as indicated on the NUD4001 datasheet's page 4, and i am seeing the Drop voltage to be too high, resulting in higher Power dissipation as what the device can handle safely. But i do not understand why it is putting the IR LED constantly ON, even though i am driving the transistor on NUD4001 with my ON/OFF Carrier.

I'm not surprised, @300mA, 24V you have some 6W in the LED driver, while it only can handle about 1.5W.
You need a DC-DC converter for that, or just drop in some big resistor to dissipate the heat. And decide on the system voltage.

[ogden]

You need a DC-DC converter for that, or just drop in some big resistor to dissipate the heat

.. or connect enough IR diodes in series so IC does no exceed max specified heat dissipation.