ADC input protection ??

[mindstorm88]

Hi guys , I would like to know how do you protect your PIC adc inputs against reverse polarity ??? Ì'm feeding battery volrtage to my PIC16F877A and would like to protect it against operator error !!!!

Thanks

[lsimaster]

The best protection scheme is a series resistor and a pair of schottky diodes (such as BAT54S) one to each power rail. The resistor should be sized to not burn out the diodes and no affect the sample accuracy of the A/D. The Microchip data sheets give info on the accuracy issue.

The schottky keeps the voltage drop below the processors internal protection diode forward drops and therfore prevents latchup and destruction.

We have used this approach for many years with complete success. Typically we use a 10K, 1206 resistor although sometimes go smaller depending on how we are using the A/D, which comes down to the sample rate. The same circuit is also effective on digital inputs that are exposed to the external world.

[logical]

[lsimaster]

The zener has a problem because it is more expensive and, most important, the zener tolerance is poor. Look at a zener break over curve with tolerance factored in and you will see what I mean. The zener provides some protection just not as good.

The schottky diode pair clamps the input within (Vcc + VS) and (Vss - VS). VS has a very tight tolerance. Using the schottky insures that VS is always less that the internal clamp diode forward drop as those are regular silicon junction diodes. The difference is about .3V vs .7V. I designed integrated circuits for a lot of years so am very familiar with CMOS input structures.

We use the BAT54S because it is very inexpensive in tape and reel and the SOT23 size doesn't waste a lot of board space.

[mkuang]

[lsimaster]

This is not true. Zener diodes have very poor tolerance compared to a forward biased diode junction.

Furthermore they are extremely leaky so will completely distort the signal to be measured. The leakage current across the input resistor will make the circuit useless and the leakage gets extremely bad as you approach the zener point. The zener point is not a point but rather a place on a current curve. You guys need to learn a lot more about semiconductor physics.

Small schottky diodes have very low reverse leakage so this is only a problem in circuits of extreme accuracy. Furthermore the stacked structure makes it so one leakage mostly cancels the other.

The 1N914 is not a schottky but rather a very old junction diode designed by the original Fairchild Semiconductor. The modern equivalent is the 1N4148.

[mindstorm88]

Thanks guy !!!! i'm gonna go with Lsimaster solution , i've just read an application note about it, i'm convinced !!!!!

[logical]

[Guest]

Sorry, I didn't mean the 1n914...but I did use a pair of discrete Schottkys, I forget what the part number is though as I would have to look it up in my schematic.

You could just buffer the input to the ADC with an opamp power from a single 5V supply. Then there is no chance of your input going negative. They sell unity gain followers in discrete packages.

[asmboy]

re zeners -
To be SAFE you need to use a typical 6.2V zener to make sure you are not losing too much single to the SLOPPY Observed breakover voltages.

Real word zeners are FAR from sharp at breakover.

I've been using ONLY the BAT54s or BAV99 fast - diode method described already - dude face it - that's how the PROS do it.

[cchappyboy]

Hi Guys,
I got a question here. For BAT54S its VS is 40V. Does that mean it will work if the input signal amplitude is over 40V ? How can it protect the port ? Can this A/D port afford more than 5V ? Do not laugh at my silly question. I appreciate any response.

Thanks

[Guest]

If your ADC is powered from a 5V supply then you connect the cathode of one Schottky to 5V. The anode of this Schottky, the cathode of the second Schottky, will connect to the ADC input. The anode of the second Schottky connects to ground.

THe point is to clamp the input to either ground or 5V. If you put 40V on the anode of the first Schottky then your circuit will clamp between ground and 40V which isn't much protection at all. I personally wouldn't drive the ADC without some sort of buffer just for isolation purposes.

[SherpaDoug]

I have only been designing instrumentation for a living for 27 years, so maybe I am not a "pro"yet ;-) But I often use a series resistor to a pair of clamp diodes to the rails PLUS A SECOND SERIES RESISTOR from that node to the A/D input pin.

The clamp diodes, 1N4148 or simillar, clamp the overvoltage to about 0.7V beyond the rails. The second series resistor and the PIC internal clamp diodes handle the small excess voltage that is left. In the case of ESD, where you may have a pulse of 10,000V on the input and 10A or more being eaten by the first clamp, the second clamping stage is needed to protect the PIC. If at all possible a capacitor on the first clamping node also helps with ESD.
_________________
The search for better is endless. Instead simply find very good and get the job done.

[mindstorm88]

So SherpaDoug is 1kOhms, diodes , 1kOhms OK ??? will it affect measurement a bit ???

Thanks

[SherpaDoug]