Cheap protection...

[Bernd Felsche]

Buchholz, Steven writes:

>> Cost is a significant factor - the uC is worth only a few bucks, so
>> I don't want to spend hundreds protecting it! Opto-couplers are a
>> bit OTT (IMHO) and large.
>> 
>> So far, a simple voltage divider with forward signal diode, and
>> filter capacitor feeding into a (hex) Schmitt trigger (74C14) is how
>> far I've got. But the voltage can vary from about 8V to perhaps in
>> excess of 16V, but I want those levels to be "true" always. The
>> Schmitt will sense 3.5 to 5V as true when running at 5V.
>> 
>> Should I include a (4.7V) Zener on the input to the trigger?

>... I forget the actual numbers, but not too long ago I was surprised to
>read that blowing the horn puts spikes that can be somewhere in the 200v
>region on the power lines ...

I hope it's not that bad when you get further up the wire because most 
OEM ECU's would have trouble coping with much more than 80V. Maybe
short spikes - they can be absorbed by capcitors.

>I don't think the problem is the cost of replacing the CPU ... it is more
>the hassle of having to fix something that's broken.  

That as much as anything - though designs I've seen (which isn't
that great a number) have "sacrificial" components such as
small-signal diodes, etc which will blow if somebody does the
unthinkable and shorts a plug lead to a sensor input. Or something
equally dumb.

>I've seen optocouplers that come in small SMT packages ... I'll bet the cost
>isn't that much.  AAMOF, I picked up a really cool demo kit from
>http://www.lumex.com that included optocouplers and a whole selection of
>lamps, LEDs & displays.  

SMT is a bit of a rude word in my vocabulary. I'm not equipped for
it at all. My hands are too shaky for normal soldering as is.
The *count* of the parts is important, as is the cost.

>It isn't clear to me that you need to have a voltage divider to protect the
>input ... a series resistor, perhaps with diodes to the rails should be
>sufficient.  Some parts already have the diodes built in.  I'd also put

I don't think the 74C14 Schmitt trigger would like 12V input if it's
running from a 5V supply rail. The Schmitt trigger connects to the
inputs of the uC, which shares the regulated 5V supply.

So I divide the external voltages on 2.2 and 1.2 MOhm resistors.
That means I can cope with a nominal 9.9 to 14.2V range within the
trigger levels (plus the signal diode drop).

>ferrite bead filters on the signal inputs.  You may want to look around to
>see if someone doesn't offer an input device that has built-in protection.  

The protection capacitor doubles as a low-pass filter. The signals
I'm sensing originate from switch closures and relay contacts in the
main. So I can happily discard everything below a couple of kHz.
The beads are a good idea, especially if the device were for
something more significant than a "smart" courtesy light delay!

The devices I'm planning to use already have "protection", but I
like the belt and braces approach. The Schmitt trigger is the
second-level sacrifice. Cheap to replace and easy to test.

BTW: I discovered that leaving my GSM phone in the car sets of the
car alarm. I was already swearing at the alarm when I noticed the
phone clipped adjacent to the main wiring panel. Doh!

So at least I have one way of testing anything I build against 
severe EMI.

-- 
Real Name: Bernd Felsche
    Email: nospam.bernie at perth.DIALix.com.au
     http://www.perth.dialix.com.au/~bernie - Private HP
----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org