Testing:  Using rev - of board.  All capacitors suggested values.

Testing with the same LSM303C, and ATTiny441.  all operating at 115,200.  Rev 635 of firmware.

Test 1: 5V from shrink tubed USB serial dongle with 20cm leds.
	Orientation, mostly up, tilted down on regulator side.
	3.0V Regulator (MCP1824)
	4.92V at device
	3.02V after regulator
	Standard of Deviation: 5.81	5.81	6.61	5.39	5.56	12.84	0.95

Test 2: Same as test 1, except upside down.
	Standard of Deviation: 5.85	3.48	7.05	5.57	5.64	12.53	0.92

Test 3: Mains at 3.3V from USB dongle thing. Tilted toward AVR.
	3.0V Regulator (MCP1824)
	3.37 at device
	3.02 on regulator
	Standard of Deviation: 5.29	10.25	6.99	4.96	5.73	12.25	0.83

Test 4: Same as test 3, just upside down.
	Standard of deviation: 4.48	3.67	6.53	5.87	5.75	12.83	0.87

Test 5: 3.3V from shrink wrap, 2.5V reg.
	2.5V Regulator(TR2EF25)
	3.31 at device.
	2.49 on regulator
	Standard of deviation: 3.82	4.51	6.08	5.28	6.51	12.12	0.85

Test 6: Same as Test 5, except upside down
	Standard of deviation: 4.06	3.71	5.59	5.62	5.19	12.53	0.89

Test 7: 3.3V from shrink wrap, 2.8V reg.
	2.78V on regulator
	3.3 at device
	Standard of deviation 10.13	6.32	11.24	5.37	5.72	11.80	0.91

Test 8: Same as test 7, except upside down.
	Standard of deviation 3.63	4.12	5.65	5.86	5.19	12.84	0.99
     (Data from this run was lost)



You know what... 3.0V is good /enough/ and gives us the widest AVR range.



Now, let's try tests with LED.  5V input, 3.0V regulated. Upside down.
Test 9:	LEDs off: 			3.91	5.11	7.94	5.65	5.98	12.71	1.12
Test 10:LEDs 64/128/200:	7.07	7.02	6.22	5.31	5.57	12.25	1.77
Test 11:LEDs 64/128/200:	5.35	3.47	5.89	4.95	5.26	12.00	0.90
Test 12:LEDs ff/ff/ff:		5.15	3.87	5.58	5.49	5.18	12.28	1.19
Test 13: Just turned off:   5.23	9.31	11.17	5.30	4.77	12.33	1.85  (You can see temp sailing down)

Looks like temperature differentials are what is killing us, not the LED disrupting the power.