Brent Brown wrote:
> Some points in response to no particular post but the question in general:-
>
> 1. The internal protection diodes have always been given only an "absolute 
> maximum" rating. That means that when you pass *any* current through these 
> diodes at all you are taking the part outside of it's reccomended operating 
> conditions - chip not guaranteed to function normally. Keeping within the absolute 
> maximum rating under these conditions guarantees only that the chip will not be 
> destroyed and will function normally again afterward. Typically this value used to be 
> +/-10mA, although interestingly I just noticed on the data sheet for a PIC16F886 it is 
> now +/-20mA (both input clamp current and output clamp current).
>
> 2. In real world applications where input protection is required against over/under 
> voltage you must consider what current, if any, will flow in the PIC's input protection 
> diodes. If you pass *any* current through these diodes you must realise that you are 
> immediately outside the bounds of what the data sheet describes and you are on 
> your own. It is a grey area. Having said all that people can and do use these diodes, 
> but, I repeat, results are not guaranteed. As a rule of thumb it *might* be 
> reasonable to design such that under worse case operating conditions the current 
> would be no more than say 10% of the absolute maximum rating. Then you would 
> want to test this to give yourself some kind of confidence the PIC in question 
> behaved acceptably under these conditions (no latch up, program glitches, etc).
>
> 3. The absolute maximum specification of (Quote from PICF886 data sheet) says  
> "Voltage on all other pins with respect to VSS ....-0.3V to (VDD + 0.3V)". This is 
> worth worrying about. It used to be -0.6V to Vdd + 0.6V on some if not all earlier 
> PIC's. It is much, much, harder to design for +/-0.3V than 0.6V. I don't know why the

I am calling my Microchip FAE tomorrow and get more of a grip on this. 
Peter has possibly opened our eyes about
this. If what Peter says is true, the utility of PICs just dropped about 
90%, because it will be costly to keep them safe.

I have been depending on a TVS to clamp at about 7V on a 5V-operated 
PIC. The PIC clamp diodes had to absorb
at least 1V of energy (difference between what the TVS clips and the VDD 
level). We have tested carefully without finding
a single problem with multiple PICs (PIC12F629/so8).

--Bob    
>  
> apparent change of this spec.
>
> 4. The BAT54S is a nice little series connected dual schottky diode that is great for 
> voltage clamping. It's small and cheap. But to prevent any current going into the PIC 
> input at the all external protection diode must not exceed a Vf of 0.3V. For the 
> BAT54S that means no more than about 0.5mA @ 25 deg C. That's not much to 
> play with given it's a ~200mA part.
>
> 5. Other devices like MOV's, VDR's, TVS's, all have relatively large slope on the 
> clamping voltage so not much use for this kind of precision clamping. Good for use 
> closer to the real world signal but still need more protection before the PIC input.
>
> 6. A zener diode can limit +ve voltage pretty well, but -ve is a problem at -0.6V.
>
> 7. External protection circuits are inevitable, they do take up PCB real estate, add to 
> parts and assembly costs, etc. There are many tricks and compromises but no 
> simple circuit that works that in every scenario.
>
> 8. It would be very, very, nice if the PIC internal protection diodes were fully 
> characterised and useable. If they have no spec how can we use them within spec? 
> As has been mentioned the CMOS 4000B series chip have nice specifications on 
> these diodes and that makes the chips a delight to design with. Need to interface 
> 5V logic output to 3V logic input? Fine - add a series resistor. External protection 
> circuitry for PIC inputs would be far, far easier to design if there was a useable 
> current handling ability of the input protection diodes.
>
> 9. In reality, for now at least, input protection circuits will continue to be designed 
> that inevitably result in some +ve/-ve current flowing through the PIC input 
> protection diodes under certain conditions. The goal of the designer must be keep 
> this current as low as practicable.
>
>   

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist