Dear LED Forum,
I´m working on an LM3409HV circuit that always drives 12 LEDs at around 35V at 330 mA, switching at a theoretical 330 kHz and an input voltage of 48VDC. The trouble is that the pulses on the switch node show a wide-narrow-wide behavior. Another inexplicable behavior is the average voltage across the current sense resistor, which is only 178 mV despite the IADJ pin being open circuit. (There is a 100nF MLCC on that pin to GND). The IADJ pin voltage is a proper 1.249V.
To get rid of the wide-narrow-wide I´ve tried:
- Adjusting the inductor value
- Adjusting the ROFF resistor value
- Optimizing the layout,
- Adding more input (VIN to GND) capacitance (small MLCC values and large Al-E value)
- Adding 1uF MLCC in parallel to the LEDs (the original design had no output cap)
None of this has had even the slightest effect. I´m attaching a scope shot showing the switch node in Ch.1 and the COFF pin voltage in Ch.2. This plot was taken using the spring tip attachment for both waveforms, so you can see that the switch node is well-behaved in terms of ringing/overshoot. What´s driving me nuts is that the COFF pin seems to work perfectly the first time - but when it falls again something appears to confuse the control loop, causing a minimum on-time pulse. I´ve looked at the current sense voltage, though it´s difficult to sense it properly, and this waveform along with everything else seem to be the results of the COFF pin commanding that short pulse. I can´t figure out what the cause is.
I haven´t tried putting filter cap or R-C snubber across the sense resistor - could this help?
Any ideas appreciated!
Many thanks,
chris