I had a curcuit just like this, getting z-crossing pulses from mains,
clamped by a zener at one stage, an external clamp diode  before that. I
make about 500 samples of the product to test and experienced occasional
unexplained failures. I then added a small bipolar transistor before the
pic, and that solved the problem. One thing to watch out for.... on power
up, when the supply to the pic is low, a zener clamp is still above the
supply!
Regards,
    George Tyler
----- Original Message -----
From: "Alexandre Guimarces" <alexg@IIS.COM.BR>
To: <PICLIST@MITVMA.MIT.EDU>
Sent: Monday, February 25, 2002 5:48 PM
Subject: Re: [PIC] kill a PIC


> > > The internal diodes cannot protect from these spikes.
> > > ......
> > > That application note has always amased me because it is just pure bad
> > > enginnering.
> >
> > Let's see, the PIC input diodes can take about 20mA (I'm guessing).
>
>     20 ma is the correct number.
>
> > With a 100K resistor, that's 2KV input
> > With 1Meg  that's 20 KV .
> >
> > And BTW, that's continuous current.
> > For narrow pulses, the PIC could handle proportionally more current.
>
>     Not at all. That is instantaneous. The latch-up will occur very fast
as
> the semi conductor areas are very small. It is not triggered by
temperature
> rise, as fas as I know.
>
> > My point is that the resistor will die long before the PIC !
> > (a 1 Meg resistor with 20 KV across it is dissipating 400 Watts !!)
>
>     The chip will enter scr latch up before the resistor even fells the
> problem. The spikes will arc over the resistor and even if they dont they
> will not harm the resistor because they are very fast.
>
> > And of course you would have a small cap at the PIC to filter out the HF
> > junk.
> > This cap would  also greatly limit any spikes.
>
>     The cap will not help much. It will not act fast enough !
>
> > The weakest link would be the voltage rating of the resistor, you would
> need
> > a couple in series at least.
> > And yes, I personally would put a MOV or something at the junction of
the
> > resistors.
>
>     The MOV, a very fast one, would make the design more reliable. I would
> use 2 resistors in series and the MOV in the junction of the resistors.
That
> should be enough to keep the high voltage out.
>
> > Just pointing out that the App note isn't crazy when you actually do the
> > sums...
>
>     Sorry but I would never use that without the MOV's at a product of
mine.
>
> Best regards,
> Alexandre Guimaraes
>
> --
> http://www.piclist.com hint: To leave the PICList
> mailto:piclist-unsubscribe-request@mitvma.mit.edu
>
>

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu