I have designed a 65W LED driver using a LNK420E acoording to RDK-290 reference design.
It has been designed for 85-270VAC input and 300-700mADC@85VDC output. When I tested the circuitry on modified RDR-290, I noticed no thermal problems, although it was eqipped with a small heatsink. However when I tested a prototype, serious thermal problem occurred. The prototype has got almost the same primary circuit, just L2 has been replaced by 330uH choke because of EMI at input voltages below 150VAC and transformer T1 has been replaced by two paralleled of EFD-25 size because of low profile design. LNK419E I used in RDK-290 has got an original heatsink (cca 35-40K/W estimated from its dimensions) but LNK420E in prototype has got a 14K/W heatsink (according a producer datasheet) mounted without any isolation foil.
When I searched for reasons, I found out very similar drain currents in both the designs but ON-state-drain voltages was very different: LNK419E exhibited 4V (measured) but LNK420E 9.2V (measured) (it should be cca 3V@25°C up to cca 5V@100°C according to the LNK datasheet) at the same conditions (85VAC@300-700mADC@85VDC, approx. 2.7A average drain current). I expected LNK420 would exhibit lower losses because of its lower ON-state drain resistance, however LNK420E has got as twice as high losses than LNK419E.
Maximum temperature on LNK's case at Ta=22°C and 85VAC@300-700mADC@85VDC:
LNK419E (RDR-290, 35-40K/W] 116°C stabilized,
LNK420E (prototype, 14K/W) 125°C - LNK internal thermal protection stopped further operation after approx. 20 mins.
So I'm confused and cannot discover a reason of such a large difference.
Note: I realized, the thermal shut down can be reset by turning the input voltage off although the LNK temperature didn't sunk below themal hysteresis treshold yet.
Can you advice me why it happens? A design schematics has been attached.
Many thanks and best regards