Using generic boost converters as constant current sources

Hello Larry,

You are on the correct track with this approach and many boost configuration LED drivers in fact use this approach. But yes, there are some pitfalls.

1) Most boost converters use a common bandgap voltage for the regulation point of the error amp. Typically around 1.25V. If you use a 1.25 ohm conventional resistor this will result in 1.25W dissipation in the current sense alone at 1 amp and this is quite a bit for a single SMT resistor; though possibly tolerable if the 1.25 ohm resistor is several higher value units placed in parallel to give the net 1.25 ohm equivalent.

2) It the 1.25W dissipation in the current sense resistor is unattactive you can use a lower sense resistor value and gain up the sense voltage with a simple non-inverting op amp stage between the sense point and the feedback input. This will reduce dissipation, allow flexible choice of sense resistor value but increase complexity somewhat. A SOT23-5 single supply op-amp like the LMV431 should work fine.

3) Be sure to go though the boost converter calculations to determine the necessary peak current for this application. Just some rough math... 25V times 1A output is 25 Watts output power.  From a 5V source this is 5A of average current.  Neglecting losses, for a conventional 5V to 25V boost, the duty cycle of the switch will be  80%. So you need to make sure that the maximum duty cycle (aka duty factor) of the switch and control section will accommodate around 85% guaranteed "Minimum maximum."   So your peak switch current will be about 5/0.8 * 1.4 (for 40% inductor ripple current) = 8.75 Amps peak. (Not a problem for a modern discrete mosfet.).  You might look at the LM3481 as the controller;  the mosfet switch will also have a low value resistor for peak switch current sense.

4) Be aware of the issue that occurs with a boost configuration when an open LED string occurs. If the stage is generating 25W of output power and then the LED string opens; the stored energy in the power inductor will need to go somewhere and if you aren't careful it just all goes into the output filter capacitor and the output voltage goes way way up and the mosfet or other elements get damaged from over-voltage stress. So there should be a voltage limiting feedback loop as well as the current limiting feedback loop, and possibly some dissipative clamp.

5) Also be aware that conventional boost converters don't provide any sort of load current limit in the event of a short circuit on the load. Suppose the high side of the LED string gets accidently shorted to ground. There is a direct path from the 5V input through the inductor and catch diode directly to the fault path to ground. This fault path can quickly damage the diode unless the input supply quickly folds back into current limit.  Whether the system survives this short depends on the behavior of the system supply.

If the design can tolerate the added expense and physical size of a transformer, then a flyback configuration might be a better choice. This topology will give you inherent short circuit protection and has some other advantages. 

Alan Martin