Sleep current too high.

[Mike Walne]

I'm trying to run a PIC18F1320 for months/years from a single 3V coin cell.
With no activity, I want it to go to sleep, but don't seem to be able to reduce the current to low enough levels.
I've stripped a sample board to the minimum to investigate.

Schematic looks like this:-

[temtronic]

Any other internal peripherals ? Some default to being 'on'..thus drawing power.
like...
comparators
ADC
timers
I2C
UARTs
2nd osc/clocks
portB pullups

I know...you've probably checked....but...

hth
jay

[PCM programmer]

Compile and look at your fuses at the end of the .LST file. It shows
Brownout is enabled by default.

[Mike Walne]

temtronic:-

Thanks for the hint. I will have to double check. There's a long list and they're not all obvious to a simple guy like me.

PCM programmer:-

Hadn't considered brown out at all. Looks like a good candidate, the current you're quoting is close to my measured value.
It's nearly mid-night here in the UK, will have to wait till tomorrow.
Thank you again.

Mike

[Ttelmah]

I would leave the LED pin set as output, and drive this pin low. An LED as a load, won't ensure the pin can't float into the transition region, and consumption will go up if this happens.
Glad to see you have all the other pins driven, but seems unnecessary complex, why not get rid of the resistors, and just drive the pins with the PIC?. The PIC outputs, 'static' holding a pin at a logic level draw basically nothing (femto amps).
Key is to ensure that all pins _are_ driven when the chip is asleep.

Best Wishes

[Mike Walne]

Hi Ttelmah:-

On my board I actually have most pins driving LEDs, and a couple connected to switches, all active low.
To investigate the current consumption problem I simply assembled a bare board with one LED only, to prove the board was running.
I pulled all the I/Os to 0V because it was easy to dab on 10k SMDs from each pin to a 0V track which conveniently runs underneath the DIL PIC.
I had already tried making the pins all O/P, the current in sleep mode was roughly the same for both input or output modes.
What I posted was exactly the latest version of my testing. (I was simply being honest and trying out all possibilities.)
I will take your advice and go for driving all I/Os.

Hi PCM programmer.

I changed the #fuses to include nobrownout.
The current in sleep mode has now dropped to ~3.5uA.
A long way off the nA level the data sheet lead me to expect.
I suppose I could live with it.
I can now anticipate battery life being comparable to shelf life.
Still leaves a nagging doubt there's something I've overlooked.

Will report on further progress if and when.

Mike

[temtronic]

Mike.. use a bigger battery !
While it might be a nice 'exercise' to reduce power to almost zero..most of us can't afford the time($$$$) to figure out the 'details'.
I still think some silly 'default' value for a peripheral or 'deafult 'fuse' setting is the culprit.
Boils down to how much effort you'll use for what gain.

[Ttelmah]

Going low, can be necessary in some cases. The factor of error here is so large, that it is important to work out what is wrong, since it could reflect something that will cause problems later.

Other things to consider:
I'd be explicit on all the fuses. NOFCMEN, NOIESO. Though the other oscillators should be 'off' when asleep, it is worth making sure there is nothing trying to happen here.
I'd make sure that the timer1 oscillator is off in the code.
setup_timer_1(T1_DISABLED);
If this was trying to run, it'd explain the problem.

Normally I'd say to make sure it is not the voltage regulator drawing the extra power, but you are running directly from the battery so it doesn't apply. However make sure there is not a resistive load somewhere - can you remove the PIC and measure the current without it?. Something silly like a capacitor that has gone high leakage, or a hair whisker on the board. Triple check on your pulled up pins, that they are sitting at the supply rail. If one is even 0.1v down, it implies something is drawing current here.

Best Wishes

[ckielstra]

My first hit in Google "site:ccsinfo.com/forum power consumption"
http://www.ccsinfo.com/forum/viewtopic.php?t=45460

Here the same PIC18F1320 was discussed and disabling the Fail-Safe Clock Monitor solved his problem with high current in sleep mode.

[Ttelmah]

We tend to be quite good at having similar thoughts!.

The two things I found were FSCM, and the timer1 oscillator.

Best Wishes

[Mike Walne]

I can remove the PIC, (it's socketed). Without the PIC in place current is <1nA.

With the PIC in place current is now down to ~4nA.

At this level I'm struggling to measure it.

The biggest culprits were NOBROWNOUT and NOFCMEN.

I can see from the data sheet that Timer1 defaults to running.

I've not been able to detect any improvement by either stopping Timer1 or adding NOIESO to #fuses. I'm not saying that there's no gain to be had; changing the oscillator to 32kHz now no longer produces a noticable benefit either.

I'm going to have to do a board scrub to remove flux residues if I want to go any further.

Thank you everyone. You've been so helpful

Mike

[drh]

Timer 0 in ON at reset/power up. Have you turned it OFF?
_________________
David

[Mike Walne]

Thanks David,

I had mis-read the data sheet and missed that one.

Will do.

Mike

[Ttelmah]

With the clock stopped, Timer0 stops as well.
The point about Timer1, is it has it's own oscillator, which will keep running. This _should_ default to off, but 'just in case', this was why I suggested turning it off. The fact the power doesn't change, says that it was correctly not starting the oscillator.

Best Wishes

[Mike Walne]

Hopefully, the last word from me on this thread.

On my project I have mostly outputs which drive LEDs by pulling down.

There was still excessive current in sleep mode with the outputs driven high.

The culprit was A4 which is open drain on many PICs.

A burden resistor across the LED reduced sleep current from 10's uA to low nA level.

Mike