@tlfong01 thanks for your reply. I think your provided LED driver is quite good, but I do not feel comfortable with the maximum rating of 350mA. I have to be aware of the absolute worst case pulse forward current of up to 1A the LEDs can draw. I am currently prototyping the last schematic i send to verify it, but on behalf of LED drivers i would rather go with something like this: datasheet.lcsc.com/szlcsc/…

@OYPS Yes, I very much agree with you about the 350mA limitation. On the other hand, I hesitate to support you on the P Marian circuit for a couple of reasons, including the following: (1) The driving transistor is power NPN BJT, which is very power inefficient, comparing with N-channel power MOSFET.

(2) The circuit design looks a bit amateur or casual, eg, using MOSFET 2N7000 to control bipolar circuits, looks weird. (3) I am pretty sure that this seemingly not too professional circuit can be improved.

Coming back to the buck concurrent source chip you suggested:

@OYPS I am very glad that you have found this LED driver. I skimmed the datasheet and I am very impressed. All in all, for 1W/350mA, 3W/1A power LED applications, I would give your LM3405 five stars, and my ZXLD1350 only 4 stars.

Both your and my recommendation have all the constant current source LED driver features I want, as highlighted in the following picture:

My quick and dirty conclusion is that even my recommended ZXLD1350 can do 1A max, I still think your recommended LM3405 is the best, or almost perfect for your application.