Could you not use the standard bown out reset controller as described in one
of the Micrchip app notes? I forget which but should be relatively easy to
find.  It consisted only of a transistor and a few resistors for the
simplest configuration.  I don't know how much power the PIC will consume
when held in reset but I guess not much at all.

Regards

Mike Rigby-Jones
mrjones@nortelnetworks.com


> ----------
> From:         Marc[SMTP:marc@AARGH.FRANKEN.DE]
> Sent:         26 February 1999 12:27
> To:   PICLIST@MITVMA.MIT.EDU
> Subject:      16F84 power consumption during RESET // Brownout detector
> with sub-ms  /RESET time?
>
> Hi.
>
> I have a question about the 16F84 power consumption. What I need in my
> project
> is to remember SRAM contents over short power failures. Storing the
> information
> to EEPROM is not possible, because it is already full with other data.
>
> To accomplish that I did decouple the 5V input with a series diode, and
> added
> a capacitor on the PIC side. The (failure prone) 5V input is sensed with
> RB0/INT.
> A falling edge triggers the int, and the PIC 16F84 will go to sleep. This
> is
> to make the capacitor charge last as long as possible.
>
> The capacitor that I chose for size and cost reasons allows data retention
> of about
> 30 seconds, which is plenty enough for the problem. However, the voltage
> falls
> below the minimum allowed for the PIC to function properly (brown-out).
> There is
> a window, about 8 seconds after power failure to about 15 seconds, in
> which the
> 16F84 will not restart properly. Before and after it works just fine.
>
> I wonder what's the best way to avoid that. "Best" in my case is "minimum
> components".
> Either I can connect MCLR to 5V input, which puts the 16F84 into RESET
> during
> power failure. I can't find any power consumption figure in the data sheet
> for
> that situation. But I guess, since the oscillator is supposed to start
> during
> RESET, it will consume as much as during normal operation. Is that
> correct? Do
> you have any information about this?
>
> The other solution is to build a 5V rise detector that issues a reset
> pulse to
> the 16F84. It must be very short in duration, though. The device is
> expected
> to respond quickly. Can anyone of you suggest a VCC rise reset generator
> that
> is available in SOT23 and does _not_ enlengthen the /RESET pulse into the
> ms
> range (like the DS1233 for example does)? Possibly only as short as VCC
> takes
> to rise above 4.5V? Or is a discrete circuit more appropriate to the
> problem,
> as no reset is required on 5V =fall=, only on =rise=?
>