The software controlled oscillator to which Matt refers is also
available on the PIC16F819 and likely on other PICs as well.  It
sounds like just what the doctor ordered.

Andrew




________________________________________________
Get your own "800" number
Voicemail, fax, email, and a lot more
http://www.ureach.com/reg/tag


---- On    , Matt Pobursky (piclist@MPS-DESIGN.COM) wrote:

> I hate to sound like a heretic on the PIClist, but have you
taken a=20
> look at the MSP430 family of MCU's?
>=20
> You can run with a 32KHz crystal for time keeping using ~5uA
then kick=20
> on the DCO (Internal programmable RC oscillator) and run the
CPU at 4-
> 8MHz (depending on battery voltage) while your peripherals
still run at=20
> 32KHz. Running at 4-8MHz @3.0V will use about 1-2mA. I
recently=20
> converted a design from a PIC16C926 to MSP430F435 and doubled
the=20
> battery life.=20
>=20
> The MSP430F1232 has similar features to the 16F628 and is
similarly=20
> priced. TI has a free assembler/C Compiler (4K code
limit)/debugger and=20
> Olimex sells inexpensive prototype boards and JTAG
programming/debug=20
> pods.
>=20
> Matt Pobursky=20
> Maximum Performance Systems=20
>=20
> On Sat, 24 May 2003 11:51:47 +0100, Philip Pemberton wrote:
> >=A0Hi, 'm currently designing a "secure ID" tag as a project.
The=20
> >=A0problem is, the only way I can get the power consumption of
the PIC=20
> >=A0down far enough (to run the tag off batteries) is to have it
run at=20
> >=A037kHz. Now, this poses a problem.
> >=A0The hash algorithm that converts the time, date and secret
key into a=20
> >=A016 digit string is incredibly slow. We're talking about
100,000=20
> >=A0instruction cycles here. At 37kHz (108.1uS), the whole hash
algorithm
> >=A0will take 10,810,810 microseconds (around ten seconds). At
4MHz, it=20
> >=A0will take around 100,000 microseconds (or one tenth of a
second).
> >=A0oes anyone know of any real-time-clock code that can handle
having=20
> >=A0its clock frequency tweaked while the program is running? I
was going
> >=A0to use Roman Black's "Bresenham Timer" due to its zero-error
claim,=20
> >=A0but if I can't switch to 4MHz and back again without losing
time,=20
> >=A0then it's useless to me.
> >=A0ould I be better off adding a Dallas DS1605 to the board,
then just=20
> >=A0let the PIC get on with generating the hash codes? This
would=20
> >=A0increase power consumption quite significantly (80uA
IIRC)...
> >=A0TW, the PIC I'm using is a 16F628.
>=20
> --
> http://www.piclist.com hint: The PICList is archived three
different
> ways.  See http://www.piclist.com/#archives for details.

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.