[time-nuts] device security (was: Clocking a PIC16F628A from a Rubidium Standard)
attila at kinali.ch
Mon Nov 28 17:38:00 EST 2011
On Mon, 28 Nov 2011 14:34:31 -0500
Chuck Harris <cfharris at erols.com> wrote:
> Yes, the paper even says so.
> The graph of failure rate is interesting. How useful is a graph
> that shows the failure rate of a general purpose film resistor as
> being from about 5E-10 to 8E-8 ?
> In any case, they are saying that 5% of a (to be pessimistic) 8E-8
> phenomenon results in a short circuit... I think that qualifies as
> very rare.
> Put two resistors in series, and it becomes even more rare.
Erm.. i have to agree with David here: when it comes to mains, always
use twice the "security" you would need at least. And you have to.
I don't know about the US, but the regulations applicable to devices
that connect directly to mains in europe are over a mether thick A4 sized
documents. And if you even miss one little thing, you will not be allowed
to sell your device. And believe me, this is being checked.
And for resistor failures and compact fluorescent bulbs, we recently
designed for a customer the electronics of a CFL. The most common
failure mode while testing and optimizing was, that the input resistor
that was supposed to limit the inrush current, fried. In half of the
cases with a short. And the colour of a burning resistor is quite nice too,
but a bit smelly. Arcing is not as nice as burning, but at least it
doesn't smell as much....
... yes, we fried quite a few resitors in the process...
Why does it take years to find the answers to
the questions one should have asked long ago?
More information about the time-nuts