LM317 Ripple Rejection

I've developed a circuit for my model aircraft, which requires 3.3V @ 100mA. My aircraft uses a 3S LiPoly, which ranges from 9V - 12.6V. When the motor is off, the 3.3V supply is good; very low ripple. However, when it spins up, the battery voltage ripples, and the supply voltage to my MCU occasionally dips below ~2.5V, and this occasionally resets it, and at the very least it causes problems with the rest of the analog circuitry.

I've attached a diagram of my system.

What can I do to reduce this problem?

-
How long is the supply dip? Scope shot? –  markrages Oct 3 '11 at 22:04
Schematic is hard to read; can you do a monochrome version? –  Brian Carlton Oct 3 '11 at 22:12
@markrages On the order of milliseconds. –  Thomas O Oct 4 '11 at 6:47
@Brian I guess I'm so used to working with high contrast schematics :/. Next time I'll upload a lighter version. –  Thomas O Oct 4 '11 at 6:55

The 100uF cap at the input of the regulator should supply power for short dips, less than 1 ms in duration. 100 uF droops at 1V / ms at 100 mA current draw.

If your dip is longer than a few ms, you can try a bigger cap.

Also, you can gain a couple volts of headroom by using a low dropout regulator instead of the LM317, and a Schottky instead of the 1N400x diode. That might be enough, but it would be nicer to solve the problem altogether.

100 mA is a pretty high current draw. Look at your power budget. Can you arrange to have the high-current parts shut off at brownout, so the microcontroller isn't in danger of resetting?

If you stick with the LM317, you can improve the regulation by bypassing the ADJ terminal:

-
Unfortunately, the power budget mostly consists of a dsPIC33F @ 75mA. The other ~25mA is analog circuitry, which is directly connected to the power rail, no SHDN pin or otherwise. I'll try increasing the capacitor size and seeing if that fixes the problem. –  Thomas O Oct 4 '11 at 6:57

Make sure that the input voltage to the regulator isn't crashing when the motor draws current. If the regulator loses headroom, your output voltage will fall out of regulation.

It is recommended to have a 100nF ceramic capacitor directly at the input of the regulator when it's far from a good filter. I believe 25cm of wire qualifies as 'far', and the 1N4001 diode you have after the wire does a good job of separating the filter from the regulator too.

You may also need a larger capacitor after the regulator to help deal with the transient current being drawn by the motor. Try $10 \mu F$ to $22 \mu F$.

-
The min voltage seen is about 9V, even at high speeds. The motor is connected to the battery directly. The regulator does not power the motor, it provides a stable power supply to an MCU. –  Thomas O Oct 3 '11 at 21:54

Looks OK.
You have vast amounts of headroom here - dropout should not be a problem.
Motor spikes may be getting through.

Increasing cap at Vin should help.
Very large won't hurt (1000 uF?) Keep close to regulator.

Larger output cap not so useful but won't hurt.

Cap across 150R to ground a good idea but should not be major factor here
- say 1 to 10 uF.

Ensure EMI filter is not being bypassed by ground connections.
Probably OK.

Direct EMI pickup from motor possible but not likely.

-