> > 2 - Redesign the circuit to guarantee a maximum rise time.

> Unfortunately, option (2) is a bit bogus: without a rise time spec
> you'll never know if your circuit is fast enough.

I couldn't say that I disagree with your comment but when you
think about it, a risetime spec for a reset circuit is rather bogus
itself.

If you consider an ideal situation, there is no need for time to come
into the situation other than sometime after the oscillator has
started and the pre-reset state is established. Other than that, the
requirement is purely voltage threshold based.

Add in some reality and you have the situation where a slowly
rising Vcc reaches the threshold point. At this point the micro is
most susceptible to noise. Any noise that exceeds the hysteresis
threshold will try to put the micro back into the reset state and if
that transition occurs within a minimum time (reset pulse width),
puts the micro into an undefined state. Also the act of the micro
coming out of reset makes it draw substantially more current, both
internally and with whatever is attached to the pins, again reducing
the effective hysteresis threshold.
To counteract this you can either ensure that the voltage is higher
by the time the current draw increases or lower the impedance of
the supply.

If you consider why a supply would rise slowly, it's generally
because it has a high impedance in there somewhere. So if you
give the micro a generous, low impedance source close by, the
effects of the slow risetimes are reduced locally. A capacitor to do
this has to be quite large because we are talking about lowering
the supply impedance over a much longer time period than the
10nF decoupling issues often discussed.

If you put say 47uF right next to the micro, you'll increase the
startup reliability for slow Vcc risetimes even though it is counter-
intuitive to add capacitance. However, as Bob says, you don't have
a figure to work towards as a spec so you can only improve and
not "fix" the problem with this approach.

So if I had a product where a slow risetime was an exception and
testing revealed that it could occur, I would be happy with this as a
fix. If a slow risetime was expected as the norm or a common
occurance, I would want to do something a bit more affirmative. In
that case I would look at something like a regulator with a
shutdown facility and use a zener to turn it on.

Steve.



======================================================
Steve Baldwin                Electronic Product Design
TLA Microsystems Ltd         Microcontroller Specialists
PO Box 15-680, New Lynn      http://www.tla.co.nz
Auckland, New Zealand        ph  +64 9 820-2221
email: steveb@tla.co.nz      fax +64 9 820-1929
======================================================

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads