Hi all,

I am in this situation: a pcb with (o.a.) a microcontroller, a remote DC 9V power supply, a DC/DC buck converter to 5V and from there to a linear regulator for 3V3. Besides that, there is a second buck converter with the input connected to the 9V DC to power a white LED module capable of handling 1A, Vf varies per subtype between 2V6 and 3V3.

The led should run at around 300-400 mA, that way it runs most economic and doesn't generate much heat.

Currently I have been doing this for years by fine tuning the second converter's output voltage very close to the Vf that makes the led draw 300-400 mA. I know it's not correct but apparently the temperature remains quite stable and so does the forwarding voltage.

But now I want to do it properly, i.e. using current sourcing. I have considered a few alternatives, but all have drawbacks. Please share your thoughts there.

1) apply a resistor between 9V and the LED. Pro: simple. Cons: very wasteful and produces much heat
2) likewise to 1) but connect to the 5V rail. Pros and cons like 1) but less severe.
3) use a current source or mirror from one or two bjts or MOSFETs. Pro: very good regulation, cons: still wasteful, stil heating
4) use an IC that does this job. Con: very hard to find one that can supply 300-500 mA and that's actually available and solderable by hand.
5) use a specific led driver, in a switching configuration. That would be nice. No power waste and no heat. Now find one that is hand solderable (preferably DIP, otherwise SOIC), one that can operate at 300-500 mA, can be bought by customers/amateurs, has a decent price and is actually available. I've tried but got lost here. Always either one of the conditions is not met.
6) another approach?

Please note the power should be switchable by a PWM signal of 120 Hz. The switching can de done by the IC itself or an additional external MOSFET, I don't care about that.

Your thoughts please!

 

There are a number of LED driver PWM ICs that will work. Here is one that will take a voltage from 4.5 to 40V and give you a constant current of 1A. (set by RS). There is very little power loss in this type of circuit.

Here is another example.

 

You can get a buck converter board with current limit for practically nothing. Set it to max voltage and 400mA current limit and you have a cheap efficient constant current LED driver.

 

BTW I forgot to mention another approach. The buck converter I am using now has a variable resistor for setting the output voltage. I could remove it and replace it with a transistor (bjt or MOSFET?) and have a current sensing circuit drive it I guess? Does anyone have experience with this? Unfortunately I am not smart enough to figure out the rest... Also the buck converter is miniature and everything SMT so very hard to track the traces.

 

You can get a buck converter board with current limit for practically nothing. Set it to max voltage and 400mA current limit and you have a cheap efficient constant current LED driver.

Interesting idea! But I fear it's just too big, I haven't seen them on a small break-out board like the usual voltage controlled converters.

 

There are a number of LED driver PWM ICs that will work. Here is one that will take a voltage from 4.5 to 40V and give you a constant current of 1A. (set by RS). There is very little power loss in this type of circuit.
View attachment 328132
Here is another example.
View attachment 328133

The al8861 seems to be comparable in dimension to soic, so that could work. The tps seems to be available in sub-soic packages only.

EDIT: not quite easy to obtain using non-business channels, just as usual.

 

Consider that any scheme using a linear device will result in power lost as heat. and then consider that a switcher regulator, unless it has a filtered DC output, will be supplying the LED power at some higher current for a shorter time. So depending on all of your goals, how much improvement will you actually gain?? Will it be worth the effort??

 

Interesting idea! But I fear it's just too big, I haven't seen them on a small break-out board like the usual voltage controlled converters.

Example

 

Example

The XL4015 current limits at 5A and I don't see how to set that to 1A.
digikey.com has about 100 different types of LED PWMs that might work. They will have a resistor to set the current limit. The link has a filter set to 1 to 4A and the input voltage set to 12 to 50V.

 

The XL4015 current limits at 5A and I don't see how to set that to 1A.
digikey.com has about 100 different types of LED PWMs that might work. They will have a resistor to set the current limit. The link has a filter set to 1 to 4A and the input voltage set to 12 to 50V.

Are you talking a out the board I linked to?

Output current:Adjustable maximum 5A

And there is a pot to adjust the current limit.

 

(6) ---> If your supply is really 9.0V, connect it directly to 3 of those LEDs in series with nothing else.

 

If your supply is really 9.0V, connect it directly to 3 of those LEDs in series with nothing else.

No. Never jam a voltage on a LED. The LEDs in post #1 are 2V6 to 3V3. LEDs are not temperature stable. I know hobby people get away with putting 3V on a 3V LED but in production it is dangerous. You will have large variations in current form one LED to another.

 

And there is a pot to adjust the current limit.

I was looking at the data sheet for the XL4015. I see the board has an op-amp, pot and more to add current limit. I did not look at its schematic.

 

Consider that any scheme using a linear device will result in power lost as heat. and then consider that a switcher regulator, unless it has a filtered DC output, will be supplying the LED power at some higher current for a shorter time. So depending on all of your goals, how much improvement will you actually gain?? Will it be worth the effort??

I think you're absolutely right. I did some testing though and found that using a buck converter considerably less power was used. Something like 5% loss against 20% so it really does matter. These buck converters have become really efficient nowadays! I am not at home at the moment, maybe I'll test again soon and note the actual figures.

 

Example

Yes I know these. Way too big unfortunately. This is more like I am using: example

 

PT4115 seems to be well available, let's see how it works out. Also ordered some 47μH coils and then we're ready to go I guess.