datasheetmosfetthermal
Here's a SOA graph:
What I don't fully understand is the notes area at the bottom:
*Notes:
Tc=25°C
Tj=150°C
Single pulse
Every MOSFET datasheet seems to have a similar set of notes, where Tc (case temperature) is 25°C, and Tj (die temperature) is the maximum operating temperature.
However, it seems unrealistic to keep the case at 25°C–how would I determine the safe operating area, for example, a situation where the case temperature is 60°C?
The SOA for this FET shows an additional limitation (beyond the traditional Vds, Imax, Rds, and P lines), that is not commonly found in other MOSFETs: the lines under the 10 ms and 1 ms markers. The only possible reason for this limitation that I found appears on the last page of IXYS Power MOSFET Datasheet Parameters Definition:
" These theoretical constant power curves are derived from calculation with assumption of essentially uniform junction temperature across the Power MOSFET die. This assumption is not always valid, especially for a large die MOSFETs. ...(more explanation)... Modern Power MOSFETs optimized for a switch-mode applications were found to have limited capability to operate in the right-side bottom corner of the FBSOA graph."
Note that this transistor is intended for switching, not for linear operation, so it is probably optimized for that task: it is not designed for operating for longer time in this lower-right corner of the SOA. Note also that for small times (< 10us, as occur in switching) there is essentially no such lower-right-hand-corner limitation.
For comparison: the SOA of a random MSOFET (irf7146):
How do I determine the VDS of the MOSFET in my application?
For power dissipation, given you are using an external part (series resistor) to limit the current, you just need to know the Rds(on) of the FET and the load current. Then Vds in the "on" state is just \$I \times{} R_{ds}(on)\$.
the MOSFET will never be exposed to 36v because each LED diode has a forward voltage drop and the MOSFET itself will be at the end of the chain. Am I being led astray with this assumption?
As horta points out, you also have to consider the case when the MOSFET is "off". Then there is no current through the LEDs and the Vds on the MOSFET is the whole 36 V.
About Safe Operating Area (SOA)
NXP app note AN11158 explains the reason for the limits on different parts of the SOA curve:
Because you are using an external current limiting device, you should be in part (1) of the curve, limited by resistive self-heating.
When the LEDS are off, you will be in part (4) of the curve, limited by breakdown of the body diode or other junctions in the device.
Best Answer
These are transient response curves. A single pulse of the power and duration in the table will cause the junction temperature to momentarily rise 150 - 25 = 125 deg C. If your starting temperature is higher, the allowed power is proportional. "Single" in this context means that enough time must pass before the next pulse to allow the junction to cool back to ambient, probably several minutes.
For steady state thermal calculations, it is usually easier to use the thermal resistance values.