Analog input protection.

[yokohama]

Hi,
I've two analog inputs going to the ADC input of my microcontroller (ATmega 168).
One is range from 0 to 5V and the second is from 0V to 10V.
How can this two inputs be designed correctly in terms of protection and re-range for the 0-10V input.
The µcontroller is powered by 5V.

 

[biswaIITH]

Hi,
I've two analog inputs going to the ADC input of my microcontroller (ATmega 168).
One is range from 0 to 5V and the second is from 0V to 10V.
How can this two inputs be designed correctly in terms of protection and re-range for the 0-10V input.
The µcontroller is powered by 5V.

I guess same ADC pin is being used ...u can apply both voltages through a common voltage divider and a cap and a zener(5v) just before the pin...The revised range will be 0-2.5v (for 5v) and 0-5v (for 10v)..

 

[malli_1729]

https://www.microchip.com/forums/m633586.aspx

 

[yokohama]

@biswaIITH
Yes I can do this but only for the 0-10V input, for the 0-5V I only need some protection.

@malli_1729
Usefull link, I'll see what I can get the best from it. Thank's

- - - Updated - - -

What do you think about this :

 

[FvM]

The 5V protection looks O.K. Optimal resistor values may vary. Instead of clamping the 0-10V analog voltage to 10V, you can better clamp the divided voltage to 5V, saving the 10 V supply, provided it's not needed for other purposes.

You made a 11k/10k voltage divider which is probably not intended.

The general problem is that protection circuit has to be designed for a particular overvoltage level. No circuit can handle unlimited input voltage, you'll e.g. blow the resistors and diodes, or force the 5V supply. Smaller overvoltages can be safely handled by the ATmega substrate diodes in combination with a series resistor.

 

[D.A.(Tony)Stewart]

Protection depends on environment, such as exposure to 2kV ESD with 1ns rise time.

There are internal Schottky diodes to both rails so it may be redundant. Usually a 1~10M to ground on input in case of disconnected source.then a series current limiter 10k for surge spikes up to 10kV for 1mA drops negligible voltage for typical 20 nA input bias current. Transorbs are more robust for larger risk surges. A small input RF capacitor helps reduce unwanted stray RF carrier detection and slows down ESD impulses. again depends on environment. COmmon Mode Noise rejection is also important not shown but must be parallel path for noise balancing and twisted pairs external or better.

 

[yokohama]

You mean like this:



@SunnySkyguy; You are right, I see on the datasheet the I/O diode protection, but I prefer take care of my microcontroller by fixing the problem before.

 

[FvM]

You mean like this:

Yes, basically this way.

A problem with relying on internal clamp diodes is that forwarding biasing the diodes can affect other analog inputs, e.g. cause excessive leakage currents and respective measuring errors. Schottky diodes can avoid these effects.

The problem arises if you allow one input in overrange and want the others to measure with full accuracy.

 

[yokohama]

Thank's to all.