Surviving 1 second without battery power

[peu]

Im trying to understand how this LED flashlight driver works, so I can implement the same technique on my own design.

The driver is powered by a single 3.6v Li-ion battery and by powering it off-on in less than a second it changes modes: Hi 100% > Mid 35% > Lo 20% > Strobe > SOS

Here is the link to it:
http://www.dealextreme.com/details.dx/sku.6190
AMC7135 Datasheet

And here is the Schematic I drawn from inspecting the PCB:


It obviously detects when power goes off, but I don't have the slightest clue on how its done, if I have to think of some PIC function that does that it would be BOD, but to be honest, Im lost.

The capacitor in the schematic may work as a battery backup, but its rather small, so upon battery power lost the PIC must be configured to keep consumption to a minimum.

Any help will be appreciated!

[skimask]

It might not have to survive without power.
If you power up, read a byte from eeprom, you can know what mode you were in last time around, increment that value, then power down.
Next time you power up, you read the same byte, the mode has incremented, save the value, power down...wash lather rinse repeat.
And there's probably some slick single-button power on/off trickery going on.
Do a search and you'll find a slick single-button power on/off trickery circuit here too...

[peu]

It might not have to survive without power.
If you power up, read a byte from eeprom, you can know what mode you were in last time around, increment that value, then power down.
Next time you power up, you read the same byte, the mode has incremented, save the value, power down...wash lather rinse repeat.
And there's probably some slick single-button power on/off trickery going on.
Do a search and you'll find a slick single-button power on/off trickery circuit here too...

Thanks for your prompt reply Skimask (as always I must add )


If you power up say after 10 seconds it remembers the last mode, but if you do an ON-OFF-ON cycle in less than a second it changes modes as described. Thats the only way to change modes, fast ON-OFFs

No button trickeries, just fast ON-OFFs I have the driver at hand, and this technique is rather common in new multimode LED flashlights.

I know how to do it by using an extra diode and a PIC pin to detect power lost, but this circuit uses a more efficient way for doing that.

[skimask]

If you power up say after 10 seconds it remembers the last mode, but if you do an ON-OFF-ON cycle in less than a second it changes modes as described. Thats the only way to change modes, fast ON-OFFs

So you've got a counter...that counts...and it sits there...counting...
when the power button is pushed...
If the count < x then it saves the next mode and powers down
If the count > x then it doesn't switch modes in eeprom and just powers down

[Acetronics2]

Is it a Power on Reset

or

Is it a Brown Out Reset ...

Just test the PCON bits ... to know it. and do what's to do.

Alain

[skimask]

But if it's one of those single button power on/off type circuits, it'll always be a power on reset.

[peu]

So you've got a counter...that counts...and it sits there...counting...
when the power button is pushed...
If the count < x then it saves the next mode and powers down
If the count > x then it doesn't switch modes in eeprom and just powers down

But you must count while the power is out... there is no backup power. The mode is changed if you do an ON-OFF-ON cycle in less than a second.

I Just updated the schematic to clearly show the entire circuit is battery powered with no backup power, reload the image so the new one is shown.


Maybe Im missing something from your answer <scratching head>

[Acetronics2]

Maybe Im missing something from your answer <scratching head>

YES !!!

The capacitor holds enough power to the PIC not to reset all registers While you do the "quick on-off-on"

remember V b.o.r. is more than the min running voltage of the Pic ... and driving your current sinks do not drain much current from the PIC ...

so that gives enough time for the Pic to reset AND test the BOR Bit, change the mode, ... ... after that you re-apply the power ...

Héhé ...

[skimask]

Maybe Im missing something from your answer <scratching head>

Ok, I smell what your cooking...
Put a meter on the cap and see what it reads and how fast it decays.
Maybe it's a version of a super-cap in a really small package.
PIC's (nano-watt PIC in particular) can draw some seriously LOW current in a sleep mode.

[peu]

I think I understand what you guys mean, this would be the metacode

set BODEN config bit to 1

set BOD bit to 1
set POR bit to 1
read level from eeprom / set level

mainloop
check for BOD/POR bits at PCON register
if BOD=0 then wait for POR=0 and change to next level and write it to eeprom.
goto mainloop
end

Is this correct?

I have some questions:

What would happen if power is lost while writing to eeprom?
Can this be handled with PBP without using interrupts?

Regarding the cap, I don't have a oscilloscope to see the discharge curve, it seems to last around 1seg to fall below 2v but this measurement was done by the eyemeter.

[Acetronics2]

Hi, Peu

NO, you did not understand ...

1) You enable BODEN, nothing else required ...

2) AT program start, you first read the BOR bit

3) If BOR Bit = 0 THEN (and only IF ) you increase the EEPROM content : That means power is back, and did not reach ZERO !!! ( no Pb to write , then ...)

4) If BOR Bit = 1 power has gone to ZERO ( and BOR bit has been reset ) ... just read the EEPROM content .

5) you SET BOR bit

6) you run your "lighting" program ...

Think simple !!!

Alain

[peu]

KISS principle... I know... Im getting older&rusty (just crossed the 40 mark).

Will give your post explanation a try in real code and let you know. Thanks!

[peu]

Hi, Peu

NO, you did not understand ...

1) You enable BODEN, nothing else required ...

2) AT program start, you first read the BOR bit

3) If BOR Bit = 1 THEN (and only IF ) you increase the EEPROM content : That means power is back, and did not reach ZERO !!! ( no Pb to write , then ...)

4) If BOR Bit = 0 power has gone to ZERO ( and BOR bit has been cleared ) ... just read the EEPROM content .

5) you run your "lighting" program ...

Think simple !!!

Alain

Im kinda lost again:

From TABLE 9-6: INITIALIZATION CONDITION FOR SPECIAL REGISTERS
PCON is initialized with %00000010 (POR=1 BOD=0)

From 2.2.2.6 PCON Register
REGISTER 2-6: PCON — POWER CONTROL REGISTER (ADDRESS: 8Eh)

bit 0 BOD: Brown-out Detect STATUS bit
1 = No Brown-out Detect occurred
0 = A Brown-out Detect occurred (must be set in software after a Brown-out Detect occurs)

From these datasheet entries I dont understand when BOD goes to 1

[Acetronics2]

Hi,

Do not worry !!!

1 = No Brown-out Detect occurred
0 = A Brown-out Detect occurred (must be set in software after a Brown-out Detect occurs)

sooo ... just use the Bit level corresponding to what happend ! ... invert BOR levels I indicated for 3) and 4) and you'll have it ...

I've correct the thread , NOW ... I had not used that feature for a while ...

<< Im getting older&rusty (just crossed the 40 mark). >>

... and me the 51 Mark !!!

Alain

[peu]

Hi, Peu

NO, you did not understand ...

1) You enable BODEN, nothing else required ...

2) AT program start, you first read the BOR bit

3) If BOR Bit = 0 THEN (and only IF ) you increase the EEPROM content : That means power is back, and did not reach ZERO !!! ( no Pb to write , then ...)

4) If BOR Bit = 1 power has gone to ZERO ( and BOR bit has been reset ) ... just read the EEPROM content .

5) you SET BOR bit

6) you run your "lighting" program ...

Think simple !!!

Alain

I still have a doubt: when does the BOD bit goes from the initialization value ZERO (table 9-6) to ONE (No Brown-out Detect occurred) in a cold start scenario?

At cold start the circuit should read the stored eeprom value not write a new one

[peu]

Im reading different PIC datasheets so shed some light about the BOD sequence, but most of them share the same wording, Im really lost.

I made a little proggie to test what I've learned in this thread, but it does not work, maybe Im missing something stupid, but after many hours of trying, maybe what I need is a good rest:

@ device pic12f683,INTRC_OSC_NOCLKOUT , wdt_on, mclr_off, protect_off, bod_on

CMCON0 = %00000111     ' Disable comparator
VRCON  = %00000000     ' disable


'       I/Os
TRISIO = %00000000      ' no inputs      

' definitions
           
BODD var PCON.0

gpio=%00000100            'I know program cold started
pause 1000                'gpio.2 green led pin5 indicates coldstart
gpio=%00000000   

loop:

if BODD=0 then           
    gpio=%00000010        'GPIO.1 red led PIN6 shows a BOD state
    pause 1000
    gpio=%00000000
    BODD=1
endif
goto loop

then the schematic is pretty simple, a ceramic capacitor at the power lines for storing some battery, a couple of leds at the specified pins and a momentary switch that allows me to cut VDD for a moment.

It does not vork

[edit] Maybe I need to check for POR=1 and BOD=0 to know for sure that a BOD reset happened

Help!

[Acetronics2]

gpio=%00000010 'GPIO.1 red led PIN6 shows a BOD state
pause 1000


Here is the best reason ( may be others ... I must check ) for your gadget not to work ...

THE DEVICE MUST DRAW AS LITTLE CURRENT AS POSSIBLE !!!

Your led leads VCC voltage to ZERO ... so ... POR and not BOR !!!

Alain

[peu]

Rest helped clear my mind

I will test this code as soon I arrive at my office:

@ device pic12f683,INTRC_OSC_NOCLKOUT , wdt_on, mclr_off, protect_off, bod_on

CMCON0 = %00000111     ' Disable comparator
VRCON  = %00000000     ' disable


'       I/Os
TRISIO = %00000000      ' no inputs      

' definitions
 
' init values

if PCON.0=0 then
  gpio=%00000100            'Brown-out detected turn on LED on PIN5
else
  gpio=%00000010            'Brown-out NOT detected turn on LED on PIN6
endif

'main loop

loop:
goto loop

I can't imagine simpler code to test the BOD function. Do you see something wrong?

[Acetronics2]

Hi,

Was not so easy ... but finally got it !!!

all the trick is in the bits setting and testing scheduling ... Always keep in mind the processor will re-start from the beginning.

Have fun ... the ( let's be modest ! ... A ) working listing waits for your first try ...

Other trick : May be you'll have to test greater values capacitors ...

Alain

[skimask]

Hi,
Was not so easy ... but finally got it !!!
all the trick is in the bits setting and testing scheduling ... Always keep in mind the processor will re-start from the beginning.
Have fun ... the ( let's be modest ! ... A ) working listing waits for your first try ...
Other trick : May be you'll have to test greater values capacitors ...
Alain

You know...I was thinking...
In this type of project, there's nothing that says the PIC has to turn itself completely off right away.
Why can't it stay running for, say 10 seconds, after the LED is powered off.
So, the process looks like this:

Initial power on with an initial mode...
LED lit, PIC waiting...
button pressed, LED off, PIC still waiting...
If one second passes without a button press, the PIC shuts itself off.
If a button press happen inside of one second, the mode changes and the 1 second timer is reset.
Requires a bit more thought with the single button power on/off circuit, but that's easy...

[Acetronics2]

Hi, Ski

My tries show the processor runs in a Low Power mode ...

Your solution is easy with the ADC (or a comparator ! ) instead of the BOR ... But the goal was NOT TO USE an adc.

I know ... a '675 doesn't cost so much more than a '629.


Just consider it was another forum " for the fun contest" !!!

Alain

[falingtrea]

One thing to remember. The diode D1 is isolating the PIC supply from the LED. So the PIC should be able to stay in a brown out mode for a small amount of time. But your test circuit may have a problem if the indicator LED is directly connected to a PIC I/O pin. The LED may then drain the PIC power too fast and the PIC will go into POR quicker. The AMC7135 parts in the original design isolate the PIC pin enough so that it can still run off the cap charge. You need a higher impedance connection, on your test board, to the PIC to keep this from happening. Maybe you can put a buffer (like a HC04) between the indicator LED and the PIC pin and power the buffer straight off the supply.

Also in your code, when you check the BOD, you also need to check the POR. This is because you will have one of two situations:

BOD cleared but POR still set - Switch click
BOD cleared and POR cleared - powered down

Your code also needs to set both bits after they are checked, otherwise they will never be '1'. The hardware only clears the bits when a situation happens. It is the program's responsibility to set these bits.

[peu]

You know...I was thinking...
In this type of project, there's nothing that says the PIC has to turn itself completely off right away.
Why can't it stay running for, say 10 seconds, after the LED is powered off.
So, the process looks like this:

Initial power on with an initial mode...
LED lit, PIC waiting...
button pressed, LED off, PIC still waiting...
If one second passes without a button press, the PIC shuts itself off.
If a button press happen inside of one second, the mode changes and the 1 second timer is reset.
Requires a bit more thought with the single button power on/off circuit, but that's easy...

This is the circuit Im using:


With this code:

@ device pic12f683,INTRC_OSC_NOCLKOUT , wdt_on, mclr_off, protect_off, bod_on

CMCON0 = %00000111     ' Disable comparator
VRCON  = %00000000     ' disable


'       I/Os
TRISIO = %00000000      ' no inputs      

' definitions
 
' init values

if PCON.0=0 then
  gpio=%00000100        'Brown-out detected turn on LED on PIN5 (red)
else
  gpio=%00000010        'Brown-out NOT detected turn on LED on PIN6 (green)
endif

'main loop

loop:
pcon.0=1
goto loop

It does not work, the minute I give it power green lits and after 5 seconds¿?¿? it goes off and red lits.

Am I missing something?

I can use a 12F675 or even a 12F683 if a simpler / more efficient solution is possible.

Regarding flashlight operation, think of a flashlight without buttons, it goes on or off by twisting the body, and the battery electrical path to the LED/uC is done using the metal body. So this means when power is gone, is gone for everything. To be even more clear, here is a CAD drawing of my latest creation:


The green you see there is the uC/driver PCB that also serves as positive contact for the battery. Ground is attached inside the flashlight head.

To check more flashlights check: www.neoca.com.ar Thanks!

[Acetronics2]

Hi, Peu

You're out ...

'@ device pic12f683,INTRC_OSC_NOCLKOUT , wdt_on, mclr_off, protect_off, bod_on

 @    __config _INTRC_OSC_NOCLKOUT & _PWRTE_ON & _WDT_ON &_BOD_ON & _MCLRE_OFF & _CP_OFF
 
 
OSCCON = %01100110
CMCON0 = %00000111     ' Disable comparator
VRCON  = %00000000     ' disable
ADCON0 = %11100000
ANSEL  = %10000000

'       I/Os
gpio   = %00000000
TRISIO = %00000000      ' no inputs 

BOD var GPIO.0
POR var GPIO.1

   
IF PCON.1 = 0 and PCON.0 = 0 THEN BOD = 1	'BROWN OUT RESET BIT
IF PCON.1 = 0 				 THEN POR = 1	'POWER ON RESET BIT

PAUSE 1000

GPIO = 0

if PCON.1 = 1 AND PCON.0 = 0 then		'PCON.0 = BOD ;PCON.1 = POR
          
    GPIO.2 = 1         'GPIO.1 red led PIN6 shows a BOD state
    PAUSE 5000
    PCON.0 = 1
    gpio = %00000000
    
endif

PCON.1 = 1
PCON.0 = 1

While 1
Wend

END

This works really fine, but with a 10 µ (MICRO) F ...and a 12F683

so , some current draw improvements are still to add ...

NOTE: All Outputs are Mosfet Buffered ( BS 170 ) ...

Alain

PS: You didn't use your brain too much, here ....

[falingtrea]

Change your test circuit as follows:

Put a diode between the cap and the battery disconnect point like the sample schematic at the beginning of this thread has. This will isolate the PIC power from the rest of the circuit so that the cap will power the PIC when the battery is disconnected.

Connect the LED cathodes to the I/O pins. Then connect the LED anodes to the battery disconnect point between the battery and the diode that was added above. By doing this, the LEDs won't draw power once the battery is disconnected.

[peu]

[edit] Maybe I need to check for POR=1 and BOD=0 to know for sure that a BOD reset happened

I wasn't that far with this late edit

This works really fine, but with a 10 µ (MICRO) F ...and a 12F683

so , some current draw improvements are still to add ...

NOTE: All Outputs are Mosfet Buffered ( BS 170 ) ...

Alain

PS: You didn't use your brain too much, here ....

What brain?

Well I implemented your program and it worked, but the cap value was way too high, so I started lowering till I found a ridiculously low value...

Then I examined your code, removed some chunks and commented it:

@ device pic12f683,INTRC_OSC_NOCLKOUT , wdt_on, mclr_off, protect_off, bod_on

' I use picbasic to compile!!

 '@    __config _INTRC_OSC_NOCLKOUT & _PWRTE_ON & _WDT_ON &_BOD_ON & _MCLRE_OFF & _CP_OFF
 
 
OSCCON = %01100110
CMCON0 = %00000111     ' Disable comparator
VRCON  = %00000000     ' disable
ADCON0 = %11100000
ANSEL  = %10000000

'       I/Os
gpio   = %00000000
TRISIO = %00000000      ' no inputs 

' SECTION? what is this section used for?
' BOD var GPIO.0
' POR var GPIO.1

   
' IF PCON.1 = 0 and PCON.0 = 0 THEN BOD = 1	'BROWN OUT RESET BIT
' IF PCON.1 = 0 THEN POR = 1	'POWER ON RESET BIT

' PAUSE 1000  this pause cannot be used in a flashlight environment
'  SECTION?  what is this section used for?

GPIO = 0

if PCON.1 = 1 AND PCON.0 = 0 then		'PCON.0 = BOD ;PCON.1 = POR
          
    GPIO.2 = 1          'GPIO.2 green led PIN6 shows a BOD state
    PAUSE 1000          'I don't have patience :-)
    'PCON.0 = 1         ' not needed since you set this again below
    gpio = %00000000
    
endif

'PCON.1 = 1      replaced with one line :-)
'PCON.0 = 1
PCON=%00000011

GPIO.1 = 1 'normal state red led
While 1
Wend

END

and if a picture is a 1000 words, we could say that a video is like 1000 pictures, so 1000x1000=1000000 words... WOW... I just wrote a book take a look at this video: http://www.veoh.com/videos/v6415763hQHpjMDy

As you see, the cap is reaaally small, the green led shows BOD condition, and red led is normal operation. The diode you see there is not used. Also I left all the unused pins floating

When I used large value capacitors, everytime I did a power interruption no matter if it was 1 second or 10 the BOD led turned on.

I wonder why you had to use 10uF

Also assembled a proto PCB using the 1st schematic to start testing a real world application.

[Acetronics2]

I wasn't that far with this late edit

What brain?


.

" Whose brain ??? " could have been good level humour ...

Well I implemented your program and it worked, but the cap value was way too high, so I started lowering till I found a ridiculously low value...

Then I examined your code, removed some chunks and commented it:


<< ' I use picbasic to compile!! >>

Me too !!! ... did you notice a little difference between MPLAB and Microcode studio ???


<< I wonder why you had to use 10uF >>

I don't know ... gives me a ~ .3 s power interrupt delay ... and it won't be my headache !!!
The Pic itself draws .77 mA ...

Regards
Alain

[peu]

Alain, I just sent you a Private Message.

[edit] Sorry for the missunderstanding, and thanks for your reply