100W LED flashlight constant current driver board based on LM3429

------------------------------------ This constant current power supply board uses TI's dedicated DC input LED control The device LM3429 is used to completely calculate the parameters of the magnetic device and control part. The above circuit was simulated using Tina-ti. The original plan used three 129 chip inductors in series, but later it was discovered that the PQ2016 using PC40 cores had more advantages in area and cost, so it was replaced with a large inductor, excluding the two turns on the PCB, and the remaining turns The numbers need to be wound by themselves. Solutions such as PCB transformers can also be used in mass production. Among them, the inductor uses three strands of enameled wire with an outer diameter of 0.8mm and winds 5 turns (excluding the two turns on the PCB). The two sides are padded by about 0.19mm, or the center column is ground away by 0.38mm. Finally, the calculation part was made into a calculation book using mathcad, which was added to the attachment for reference.

I just discovered that there is a serious deviation in the calculation. Specifically, the inductance calculation formula is missing a square. As a result, the inductor is three times larger than the normal volume. Normally, it should be designed according to the inductance of 3.3uH, 10A. I will modify it after getting the board for debugging. PCB.

Updated on March 8. The board has been welded and debugged. It can generally work normally. There are a few problems.

1. At 100W full load, the switch tube temperature is too high and a radiator needs to be added. 2. The inductance value is calculated incorrectly, so the PQ2016 magnetic core is not suitable here. It is better to use the finished chip 129 inductor (in addition, do not use it at this frequency) Iron powder core inductor, such as one-piece iron powder core inductor) 3. During debugging, we encountered the problem of period increase. The frequency is about ten times lower than the normal 600kHz. We can only rely on week-by-week current limit for switching and PWM dimming. Low-frequency oscillation also occurred. It was suspected to be a problem with the disassembled chip itself. It improved after re-soldering. 4. The peripheral circuit of LM317 is drawn wrongly

PCB has been updated and will be re-sampled later

Okay, it works