I used MP24894GJ for my led driver design which has a cob led of 0.5A at Vf 36V.
For this current, each Vin/Rsense/L1/endim was 45V/0.33ohm/100uH/3.3V and the other parts were put just same as typical application of datasheet.
I hadn’t thought that I violated any part selection rules on datasheet in terms of mathematical calculation, but when my 45V power supply switched on, almost simultaneously, MP24894 was burned out with spark between RS pin and EN/DIM pin.
There were two same circuits for 2-channel control, and both got the same phenomenon, so I can only conclude that my design was wrong.
Please can anyone point out what’s the problem of my design and parts selection?
Hi @1010why ,
Can you elaborate on what you meant by 2-channel control? Are you using two MP23894 ICs in parallel to drive 1 string of LEDs? Can you also confirm your frequency?
Is the device blowing immediately with a visible spark and audible sound? If so, most likely there’s a severe transient at start-up. Try increasing the soft-start capacitor from 470pF to 1nF or 2.2nF.
If it’s still blowing, then is it possible you can take waveforms of IN, EN/DIM, RS, VCC at start-up? Set the trigger at a low voltage so the scope can capture the waveform before the device blows.
Also, I noticed in your layout that the IN loop looks a bit long. I recommend having the input voltage trace to be on the same layer as the IN pin of the IC instead of going through a via. Then, I recommend moving VIN4 cap to be right above D2. This way, there’s a much shorter connection from the ground of VIN4 cap and the GND pin. I also recommend moving Q2 closer to the IC via shorter traces, possibly by moving the drain connection to the D2 pad and widening the traces. The combination of input voltage going through a via and long loop can increase parasitics, which can cause voltage spikes that destroy the IC.
Thanks,
Rubas
