Reducing power consumption during sleep

[gjs_rsdi]

Hi

PIC12F1840
CCS PCM C Compiler, Version 5.062

I am working on a alarm system using the above PIC. It is usually in sleep mode except wake-up in case of the external interrupt triggered by the sensor.
I don't have pull-up resistor on the pin as the interrupt is L to H.

The program works as expected, but:
A regular square 9V alkaline battery get to end of life after 130 hours. Not feasible

I am testing now with 3*AA alkaline batteries so I eliminated the voltage regulator as I know is also taking power even with very low power consumption.
During wake-up (for 10 seconds) I am enabling the ADC to test the battery,
no conversion during sleep.

My questions are:
1. Disabling the MCLR~ internal pull-up resistor is a bad thing?
2. Making all the pins input before sleep will reduce power consumption?
3. What else I can do to reduce the PIC power consumption during sleep?

From my testing and reading by the internet, the sensor connected to the PIC consumes 5 to 10 uA in idle. I have to state that my capability to test
accurately the current is very low.

Best wishes
Joe

[dluu13]

Somebody correct me if I'm wrong but I believe setting pins as input will actually take more power. Instead unused pins should be set to output low.

[newguy]

[temtronic]

Check the Microchip website for a really old(20+ years) Application Note about 'power consumption. Maybe AN669 ? Wish I still had real databooks.. it's a few pages long has a LOT of data about clock speeds,powerup times, and coin cell battery life. It's really a 'must read' for all 'ultra low power' applications.
If there's an 'L' version of that PIC...check the datasheet ! could be X10 better or more for reduced power. Clock speed...slower is better though the Ap Note tests showed go slow, then fast(to do math ),then slow saved energy.
I've got a Radio Shack LCD clock module here still running on 2 AA alkalines, better part of 30 years and accurate....

Ok it took a bit of googling....but


AN606B is the apnote I've been thinking about...

Even though it's old...it's a 'must read'.
I assume there's newer info butthis is the one a memebr reading in a DATABOOK, about 2" thick.......

Jay

[PCM programmer]

[Ttelmah]

Key thing not mentioned.
If he is using a 9v battery, he will be using a regulator to drop to the
5v needed by the PIC.
In general most basic regulators will draw hundreds of times the current
needed by the PIC....
If you want low power, you need to be using ultra low power regulators.
A TPS783, versus a 78L05, reduces regulator consumption massively.

A 7805, has a quiescent of about 5 to 8mA.
A 78L05 about the same.
A 78LC05 drops this to 1.1uA.
A TPS430 takes this down to under 500nA.

In each case a linear regulator is also inefficient. I mA drawn by the target
implies slightly more than 1mA into the regulator. For efficiency a
switching regulator is needed, but again an ultra low power type.
A switcher like the TPS62743 (there is also a Motorola, and a Maxim
unit with even less consumption).

Generally the 9v battery will be rated at perhaps 400mAh. If the regulator
is drawing a couple of mA, it'll run down in only a couple of hundred hours,
whatever is done to the PIC....

[PCM programmer]

[gjs_rsdi]

Thank you everyone that posted.

As I mentioned in my first post, the boards run now on 3*AA alkaline batteries, it gives a maximum of 4.79 volt when I started testing endurance.
The reason for the change from 9V is the current of the voltage regulator itself.
After 24 hours the 3*AA batteries are giving 4.55 volt.

RA0 is output low when in sleep, when no sleep output H/L to a LED via 560 ohm resistor.
RA1 is analog input via 510 ohm resistor and have 100nF capacitor to the VSS
RA2 is input from HC-SR501 PIR. Data sheet says works from 4.5 volt VCC and up, from my testing works at 3.9 volt also. I am testing now how low the VCC can be.
RA3 is input MCLR with the internal pull up.
RA4 is output low in sleep, output high if not in sleep.
RA5 is output low, not used

I made the current consumption test as PCM Programmer suggested, I have now a multi-meter that goes down to 200uA.
Sleep mode:
Without sensor 103uA
With sensor 169uA so sensor in idle consumes 66uA.
Active mode: 9.5mA for 10 seconds

The AA battery is rated at 800mAH when discharged at 100mA to 1.1 volt per cell.
So at sleep the 3*AA batteries will last around 200 days down to 3.3 volt
If the sensor will not work at lower VCC than 3.9, to this level the battery is rated 200mAH, the batteries will last for 49 days.

I will be back with the final results, I suppose will take long time.
I will buy (if available here) the TPS430 that Ttelmah proposed, will try also with LP2950 and see the results as I would like to work with the 9 volt battery because the small size and price, they cost here 25 cents (if the endurance will be around 60 days)

I want to thank everyone who contributed to this topic
Best wishes
Joe

[PCM programmer]

[gjs_rsdi]

Hi PCM Programmer.

I wrote that 103uA is acceptable because the time that the battery will last.

After reading the part of the data sheet you told me I understand that with all the peripherals disabled I will have a current of 0.5uA during sleep. But I have BOR enabled 8uA and the weak pull-up resistor on the MCLR pin typical 140uA.
I can't have MCLR enabled without the resistor I think.

Thank you PCM Programmer
Best wishes
Joe

[Ttelmah]

The weak pullup, draws nothing, if the pin is not pulled down, It only draws
it's current when current is drawn on the pin. You should not be measuring
103uA.

Look at the data sheet:

[gjs_rsdi]

Hi Ttelmah

I will make again some tests and see what can I do.
Will be back with the results.

Best wishes
Joe

[alan]

[gjs_rsdi]

Hi Alan

As I know the analog input don't draw much current. Also the analog circuit is not working during sleep. As I mention in my previous post, I will make an additional test to see how much current I have in sleep mode. Can't be 3mA as I am measuring 103uA

Best wishes
Joe

[Ttelmah]

Tell us about the sensor. Part number?.