Russell:  What pin were you using for the input, and on what chip?

--Tom Rogers  VP-R&D  Time Tech Inc.

-------------------------------------------------

Russell McMahon wrote:

> Moral of this story:
>         Don't rely on PIC protection diodes and a series
>         resistor for input signals which swing substantially
>         outside the PIC's power rails.
>
> I implemented a "cheap and nasty" PIC RS232 receive interface
> as proposed by microEngineering labs in their BASIC compiler
> manual, using a single 22K resistor from a PC serial TX data
> pin to the PIC receive data pin (and a similar arrangement with
> a smaller resistor for PIC TX data).  As I had not used their
> BASIC compiler before I initially presumed that problems
> getting some simple code going was due to my misunderstanding
> of the code syntax in the manual (the language is simple but
> the manual is utterly diabolical - at least the compiler is
> better than its manual!)
>
> But no! - the problem lay in the interface.
> Adding 2 Schottky diodes to the PIC pin to clamp voltage
> excursions to (almost) ground and 5v supply cured the problem
> and RS232 input worked as desired. While a number of people in
> this list have proposed using the PIC's protection diodes to
> clamp voltage excursions outside the power rails it is clear
> that, in this case, this practice caused the chip to
> malfunction. This is of course rather poor design practice but
> is a known 'trick' in many other circumstances - especially
> during fault conditions. I am not certain whether the positive
> or negative excursion caused the problem - I'll go back and
> check later - at the moment the Schottky solution works OK. Vin
> high current is about (12-5)/22k = 0.3ma and Vin low is about
> 12/22K = 0.54ma. Incidentally - the PIC data sheet have Gnd and
> Vpp as maximum Vin limits with no reference, on a quick scan,
> of acceptable maximum currents when the protection diodes
> conduct.
>
> (The transmit circuit worked with no problems - as would be
> expected as the PIC is driving a high impedance. While this may
> tend to pull to one or other rail the affect will be swamped by
> the PIC's low output impedance.)
>
> In this case, to implement R232 a simple "proper" design could
> be implemented, probably, using resistors alone. 2 resistors
> and 1 transistor is the standard el-cheapo solution in such
> cases (1 R from P.C. to transistor base, 1 R in transistor
> collector, (back diode base to ground if you're squeamish). I
> will be using an opto-coupler ultimately to protect the PC and
> the target object from each other.