I have an adjustable constant current LED driver design based on the TPS54620 DC-DC switcher. I've prototyped the circuit, and I'm finding it to be unstable at low output currents and when the input voltage approaches the upper end of the 5.4-8.4v input range.
The driver is designed to supply a programmable current ranging from 100mA to 2A to the LED. It's powered by two Li-ion batteries creating an input voltage range of 5.4-8.4v.
- The LMP7718 U2A amplifies the current sense voltage across resistor R2. This amplified voltage feeds the FB pin on the TPS54620
- The MCP4261 IC is a dual 5K digital potentiometer that allows me to program the gain of the LMP7718A U2A. This gives me programmable current capability.
- The MCP4261 also allows me to program the compensation resistor on the TPS54620 to accommodate different output currents.
The problem I'm encountering is that the driver operates on the edge of stability, and the output often oscillates. For instance, with the output current set to 100mA (LMP7718 Gain = 160), the driver will oscillate when the input voltage exceeds 6.8V. At higher output currents, the driver is more stable, but still experiences some oscillation briefly after a power-on.
For the compensation network, the digital potentiometer is set to 100 ohms in series with the 47nF C11 capacitor. I chose this low value for the resistance to reduce the crossover frequency, but this does not seem to be enough for stability.
Is there anything I can change with the compensation network to stabilize the design? The load is an unchanging LED, so the response time isn't a big concern here. Although I do want to avoid overshoots which could damage the LED.
Thanks in Advance
