I am using an LM25037 in a "dumb" half bridge converter (i.e. always operating at full duty). Pin 4 (FB) and pin 8 (CS) tied to ground, pin 3 (COMP) left open. RT1 is 22.1K, RT2 = 30.1K for a nominal frequency of 270kHz and dead time of 150ns. At room temperature everything looks fine and the frequency and dead time are as expected, as measured at the output pins (12 and 13). At elevated temperature (chip temp approximately 90 C), the frequency remains at 270kHz, but the dead time as measured at the output pins (12 and 13) dramatically increases to approximately 550ns (a 400ns increase!). This is causing the output of the half bridge converter to drop below expected limits.
It is worth noting that the LM25037 outputs go to a gate drive buffer IC and so the power dissipation on the LM25037 is pretty low...junction temperature is not very much above the outside temperature of 90 C. Layout around the LM25037 is very tight.
I have never seen this type of behavior before with this device (used in many designs over the past 4-5 years), and also this temperature behavior of the dead time does not jive with Figure 10 of the datasheet (which shows only a few nanosecond change over a temp range of -50 C to +150 C). Has anyone seen this before? Any clues as to what may be causing this dead time increase?
