Shutdown current draw higher than it should be?

[Things]

I've designed a little PCB that basically combines a boost converter, single cell LiPo charger, and an attiny85, designed to run off a 400mAh LiPo cell.

I've just removed the polyfuse and stuck my DMM in it's place, and put it on the mA setting (and probes in the mA socket etc), and it's power draw in shutdown (boost converter disabled) is fluctuating between about 1.5mA and 2.5mA, which is much, much more than it should be.

The boost converter's quiescent current draw is specified as 20uA max. The battery charger specifies around 10uA max. The only other thing connected to the battery directly is a resistive divider that totals around 2.2Mohms. Yet I'm getting on the order of 2mA? 

I'm wondering whether this is just the DMM messing with me, or whether something is actually wrong with the circuit?
https://dl.dropboxusercontent.com/u/203420/Collar/Sheet1.pdf

(Note that I've manually added a 1.8Mohm resistor on pin 7 of the attiny where it connects to the battery, as I was getting current backfeeding through the ESD diodes, and even completely disconnected it for the test).

Edit: Just logged it with the DMM, started by discharging the caps and ran it for 240 seconds. Y axis is uA and X axis is seconds .. really strange. No idea what's going on here 



Help?

[lapm]

Well attiny draws current depending whats it doing. its not constant current. Also boost coverter can introduce current flacuations...

[Things]

The attiny isn't connected to the battery. It's powered from the boost converter. This measurement is while the boost converter is disabled. Should also mention for those not familiar with the part, the boost converter has load disconnect too.

[bktemp]

Vbat is directly connected to PB2/ADC1 pin. Every IO pin has an internal diode to Vcc. The circuit is powering itself over this diode.

[Things]

Vbat is directly connected to PB2/ADC1 pin. Every IO pin has an internal diode to Vcc. The circuit is powering itself over this diode.

I mentioned under the schematic link that I had disconnected that for the test.

[bktemp]

Did you verify the voltage at Vcc is 0V? 2mA sounds like the supply current of the AVR.

[Things]

Did you verify the voltage at Vcc is 0V? 2mA sounds like the supply current of the AVR.

Reads about 103mV. Have a feeling this could be more an issue with my DMM, though it seems pretty repeatable. Hmm.

Edit: Just stuck a couple 1R resistors in parallel as a shunt, and I'm still getting a reading of around 1.6mA and slowly climbing .. why it's so high in the first place is bizzare enough, but slowly climbing too?? 

[Things]

I just built up another board and it's drawing a steady 3.6uA .. looks like something is wrong with the other one. Bah, what a tail chase