PIC18 power okay, then slowly falls, then dies...chip resets...why?

I am still fighting with a PIC 18F97J60 to get the ethernet working. It seems to have finally come to life the other night, with a couple LEDs attached to a few available portB I/O lines and the LEDs on the ethernet jack flickering for the first time.

However...after starting up and running for a few seconds, both the LEDs on the ethernet jack and my portB LEDs dim down in a ramp-like linear fashion, then finally go out. The processor appears to reset and the cycle starts over...and over... The 3.3V regulator is getting very hot and appears to maybe be shutting down to prevent thermal overload.

What in the world could be causing this behavior? The PIC is getting programmed and seems to run fine. The only possibility I can think of is that because my regulator is low dropout model (Texas Instruments tlv2217-33), and it recommends using a low ESR cap. The cap I'm using isn't anything special. I didn't notice this when I designed it, because it was a last minute swap for a previous regulator that wasn't low-dropout. How critical is the output capacitance ESR to these thing's proper operation?

And if a low ESR cap on the regulator's output is critical, what about the fact that in a typical Microchip reference design (which I used mainly for the PIC's power lines) there are anywhere from five to twenty 0.1µF caps sprinkled around the board at each place power enters the chip? Certainly those ceramic SMD caps aren't low ESR either (which is where the output of the regulator feeds through). Won't they also all contribute to this problem if that is indeed what is happening? Do I just need to find a different regulator?

Thanks for any help
-Josh

Might have solved my own problem...even though the input range for this regulator says its good up to 16V, apparently, they still have the potential to overheat even when operating within their specified input parameters. (I was running it at 12V 500mA). I switched to a 5V, 2A supply, and it's running both cool and continuously!

If there is/was an ESR sensitivity issue, maybe I picked a voltage where my random cap is within range of tolerances for the regulator to continue working.

I think you'll find that dropping 13V at ~100mA is outside the specified operating parameter for the wattage it can handle without a good heat sink.

Mike.

You will notice that with regulators as you feed higher voltage, they have to dissipate this somewhere, usually a meaty heatsink will cure the cause.
My board with the 97J60 runs from a 19V Laptop PSU, 3.5A, it is feeding the actual HTPC as well as the control board, but I use a 12V 1A LDO, 5V 1A LDO and a 3.3V 300mA LDO (from Microchip) respectively, so I have the 12V running the 5V and the 5V running the 3.3V, works extremely well!
And, although they are heatsinked, the original bench version of the board was running a 12V 1A DC wall supply into the 5V reg, the the 5V into the MCHP 3.3V with NO heatsinks and it ran a bit toasty, but not anywhere near to thermal shutdown levels!

All regs were 1A LDO type apart from the 3.3V which was 300mA

I have noticed on the PICDEM2.NET board the 3.3V reg runs quite hot on there, but never blips at all.

As for capacitors, just oodle decoupling caps everywhere, especially on regs, I have never had an issue with cheap caps on a LDO or standard regulator before.

Wilksey