Failing Voltage Regulator?

[Gytis]

Hello Everyone.

To begin this post, I have drawn a schematic since it is worth more than 100 words. Please find it in the attachments! If something is blaring out as a huge mistake, please let me know!

The regulator steps 5V down to 3.3V:    http://ww1.microchip.com/downloads/en/DeviceDoc/20001826C.pdf
I had nothing loading the output except that ATXmega, and about 5uF total capacitance

I've used this 5V -> 3.3V setup for about a week, several hours every day without fault. Today after I plug in the programmer, it reads that the voltage is around 3.4-3.5V, so I thought no problem... HOWEVER when programming, it sets off a warning about the voltage being too high... so I probed with my fluke 115 MM... and it my programmer was not lying about the supply voltage being funny.
it was no longer that stable 3.314V I saw a week ago...

Obviously in circuits if something goes bad, it gets worse... so after investigating what happens upon reseting the microcontroller, i noticed that the voltage quickly becomes unstable, and jumps to 4.2V with no load... After about 1minute, the voltage becomes no longer stable, fluctuates about +50mV, +200mV after about 2 minutes... there is no periodic action (i checked with scope) it just randomly takes steps up or down...

Now onto the stranger part... in attempts to save the micro (which tolerates max 3.6V I believe ) I shorted the output of the regulator to ground... and that fixed it... for the next 5 minutes until the same thing started happening again.

I attempted to reduce load capacitance down to 2uF with the same phenomenon still occuring... I also added 3.3K resistor to load the output of the regulator even when microcontroller is reset, still with no help.

What did I do wrong/am doing wrong?

Thank you for any help!!

EDIT: Here are some snapshots of the signal from a cheap USB scope. These were all at different times. I wasn't touching the circuit or anything... just breathing.

[mikerj]

The symptoms suggest you have a bad connection on the regulators ground pin.  If this pins floats, or has a high resistance to ground then the output voltage of the regulator will increase, and it won't regulate properly.

[Gytis]

The symptoms suggest you have a bad connection on the regulators ground pin.  If this pins floats, or has a high resistance to ground then the output voltage of the regulator will increase, and it won't regulate properly.

As per your suggestion, I reheated all of the pads for the package. I also tested for resistance.. all traces seem okay... the ground pin uses a very short trace which goes to a via, which goes to the back ground plane.

Things were looking up... again, only for the first 5 minutes... now I get irregular spikes up to ~3.4V, and then settling back down to 3,3V for a few seconds... I can't see any pattern to it.

EDIT: In addition after waiting some more, it started floating around 3.4V with about 20mV tolerance for about 30 seconds, and then went back to a stable 3.313V

[bookaboo]

Does it get hot before failing? If so it may be oscillating. Id experiment with different load caps.
I once had an a PIC doing PWM driving an LM272 OPAMP to give analog out, it worked fine except when the load was removed from the OPAMP output. The voltage reg would rise to 120 degrees then shut down... and this was a TO-220 LM7805 driving 30mA top.

[Gytis]

Does it get hot before failing? If so it may be oscillating. Id experiment with different load caps.
I once had an a PIC doing PWM driving an LM272 OPAMP to give analog out, it worked fine except when the load was removed from the OPAMP output. The voltage reg would rise to 120 degrees then shut down... and this was a TO-220 LM7805 driving 30mA top.

I dont feel it get warm at all... Ill try different load caps, and see what happens.

[jlmoon]

Have you measured the current draw? Is it possible the USB supply voltage is dropping below spec?  Check both input voltage and current  to LDO and output current to load,  this might give you some indicators.

[Gytis]

Have you measured the current draw? Is it possible the USB supply voltage is dropping below spec?  Check both input voltage and current  to LDO and output current to load,  this might give you some indicators.

At this point I decided to take out the regulator since it goes well above 3.7V, and I am afraid to burn my controller. Using my power supply, the current draw is only about 30mA. I had checked the USB voltage, and it was very stable around 4.97V even when the output of LDO was messing around. I am not sure about the current going into the LDO however.

Why would this have happened?

[bookaboo]

Post your full schematic, there may be something else weird on the 3.3v line. I'd construct a test board with the suspect regulator, give it a pure resistive load and see how it behaves over time.