5V control of a precision current

[Oznog]

I want to make a strobe out of a series string of 3x 3.9V, 1 amp Luxeon emitters, powered by a 12.6 car battery, which of course can go up to 14V or so. I will be flashing it with another one of my microcontrollers, probably at 5V again.

12.6V- 3*3.9V = 0.9V to control the regulation.

Since there's not a lot of voltage room here, the currents are high, and the pulses will be sharp which makes feedback tricky, I figured just an NPN and an emitter resistor would do it.

So I think shooting for 0.3 Vcesat would mean a maximum of 0.6 ohm for the emitter resistor.

I don't know if this is such a good idea, I now need a fairly tightly regulated voltage of 1.2V, depending on gain might be 3 to 10 mA. And it has to switch on and off really fast.

Any good way to do this? Or should I think about using an op amp to drive the NPN (or MOSFET) and use the voltage on the resistor as feedback to the op amp?

 

[pebe]

As you are trying to control the current through the emitters, rather than the voltage across them, how about using a transistor without an emitter resistor, and controlling its collector current by controlling its base current?

 

[Sebi]

If You can using a SENSEFET (with 5pins), it have a Kelvin source.It can produce voltage (depending from current) without resistor (and voltage lost).
BTW: If the microcontroller exist, why not use PWM for current setting?

 

[Oznog]

The LEDs are being pulsed at their absolute max current already. So ripple can't be allowed.

I've already slept on it and the obvious thing to do is a N-channel MOSFET and a TL082 op amp. The 5v pin goes into a resistive divider to get 0.6V, goes to the noninverting input, the voltage on the 0.6 ohm source resistor goes to the inverting input, and the output to the gate. I hope it's not going to oscillate.

 

[Nigel Goodwin]

The LEDs are being pulsed at their absolute max current already. So ripple can't be allowed.

What are the ratings of the LED's? - the current ratings are usually given for continuous current, if you pulse it you can pulse it at far higher currents. Remote controls commonly pulses the LED's at over an amp, for fairly low power LED's.

Basically if you only pulse it on 10% of the time you can use 10 times the continuous current.

 

[Oznog]

Luxeon III. They're far, far more powerful than any other LED. They're also well specified. 1000 is an absolute max current rating.

 

[Nigel Goodwin]

Luxeon III. They're far, far more powerful than any other LED. They're also well specified. 1000 is an absolute max current rating.

But is that a constant rating? - if so you can pulse it far higher, a normal IR LED used in a remote control is pulsed at over 1A. If the Luxeon III is supposed to be 'far more powerful' you should eb able to pulse it much more than 1A - depending on the mark/space ratio.

 

[Oznog]

Luxeons are designed differently, with a thermal slug on back. As such, there is not much difference between constant duty rating and pulse rating- long term heat buildup issues do not exist here, and that's always the factor which separates pulse and constant ratings.

The constant use rating is 700mA. Absolute max is 1 amp. Peak Pulsed Forward Current is also 1 amp. See pg 4:

**broken link removed**

 

[Nigel Goodwin]

The constant use rating is 700mA. Absolute max is 1 amp. Peak Pulsed Forward Current is also 1 amp. See pg 4:

They don't look very useful for pulsed applications then :lol:

But looking at the datasheet they seem intended for simple illumination.

 

[Roff]

This switches pretty fast in simulation - about 400ns turn-on delay, 100ns turn-off delay, and LED current rise and fall times of about 100ns I sim'ed with a battery for vref, but you can use a voltage divider from whatever gives you the stability you need. I lowered vref to 200mv so I could use a smaller MOSFET. You can substitute 2N3904 for 2N2369, but the turn-on delay of the LEDs will go up to a couple of microseconds due to storage time in the 2N3904 (2N2369 is a high-speed switch).