This evening I received some parts I ordered for an attempt to reduce the idle power consumption of the MS20 PIR detectors. The picture above shows the current draw of the "stock" unit versus the modification I just finished. By changing three components, I was able reduce the idle power consumption to less than a fifteenth the original value.
*** EDIT ***
Here's a link to the modification procedure for anyone who doesn't want to read through this entire thread hunting for it:
Low Power Mods
Components changed were:
1.) U1 changed from an HA17324 to a TLC27L4CN $0.93
2.) Zener diode D4 changed to an LM385Z-2.5NS $0.72
3.) Resistor R8 changed from 4.7K to 100K $0.05 ?
I didn't buy the resistor
Because the zener D4 needed some minimum current to behave, the original circuit used a 3.3-volt zener with a 4.7K pull-up resistor, providing 500 uA of bias through the zener. This was on the edge of the lower limit of working current, as indicated by the fact that the working voltage was 2.5 volts. I attempted to increase R8 with the existing zener, but the regulated voltage immediately fell off to nearly nothing.
Looking around in the catalogs, I found the LM385Z-2.5 micropower zener -- it only needed 20 uA of current to do its thing, so that meant I could reduce R8 to (2.5 V / 0.02 mA) = 125K. To be just a hair on the safe side, I used 100K for R8 to bias D4 at 25 uA.
According to the datasheets, typical current draw of the TLC27L4CN is 40 uA, so looking at the schematic, the currents added up like this:
40 uA - TLC27L4CN
25 uA - R6
25 uA - R8
10 uA - R9 (measured)
24 uA - R14
------------------------- 124 uA Expected total idle current
So, as you can see in the picture above, I'm quite pleased with the 110 uA actually measured after these three modifications were done.
You could lose another 20 uA or maybe 30 uA by increasing the values of R6 and R7, and maybe R4, but we're probably already to the point of diminishing returns.
Hope this is helpful. Now I'm thinking... maybe a 9-volt battery?