Simple ADC qestion/doubt

[TKS]

What would happen if i just feed the pic vref + with 12volts and its vref - with 10volts?

could it cope with that?

Tks

 

[Nigel Goodwin]

You would probably destroy the chip - it's a REALLY stupid idea! - read the maximum specs in the datasheet!. It's also completely pointless, it's easy to scale it to those values anyway!.

 

[TKS]

The point is Nigel i need to measure voltage difference over some mosfets
aside if its possible or not the difference will be max. 1volt

the source is 12 / 14Volts DC takeaway 1 volt do you untherstand the problem? i´m playing with a opamp but cant make anything out of it..
(already searching for docs / examples)


Tks

 

[Nigel Goodwin]

It's easy to use an opamp or two to shift the voltage to reference to zero, and amplify it as well - giving a 0-5V output for a 10V-12V input.

 

[Pommie]

A couple of 9V zeners and two 2k resistors should solve your problem.

Mike.

 

[Nigel Goodwin]

A couple of 9V zeners and two 2k resistors should solve your problem.

And are you going to explain how?, or just leave it at that?.

 

[Pommie]

9V zener from test voltage to PIC pin plus resistor to GND. Pin will by test voltage - 9V.

Mike.
Edit, Added attachment. Was a little rushed earlier, just watching the first episode of the new 24.

 

[Nigel Goodwin]

Be OK if zeners worked anything like they are supposed to - but as it is, it would be pretty inaccurate!.

But why TWO zeners?.

 

[Pommie]

Be OK if zeners worked anything like they are supposed to - but as it is, it would be pretty inaccurate!.

But why TWO zeners?.

Zeners are normally pretty accurate if you keep the current through them at a reasonable level.

Why TWO, the OP said he wanted to measure the value across a MOSFET and so I assumed he needed to read 2 voltages.

Mike.
Edit, changed the statement on current as in this case it will vary by a factor of 3. It should still be fairly accurate.

 

[TKS]

How accurate would it be?

Where is the inacurance??

Tks

 

[Nigel Goodwin]

How accurate would it be?

Where is the inacurance??

The actual voltage lost across the zener will vary (quite a bit) with the current through it, and the temperature.

 

[Pommie]

The biggest inaccuracy would be the tolerance. Typically 5%. This would give the output as,
12 -> 12-9.1-0.45 = 2.45
12 -> 12-9.1+0.45 = 3.35
Paying a bit more and getting 1% zeners would give,
12 -> 12-9.1-0.09 = 2.81
12 -> 12-9.1+0.09 = 2.99

The above error would be fixed and so could be compensated for in software. No good for production but a possibility if it's a one off. This could be automatic if you know a fixed state. I.E. if when no current is flowing the voltage is 12V then the Zener error can be read indirectly by the ADC.

Error due to current would again depend on the zener chosen. With a BZX family zener then increasing the current from 1mA to 20mA results in a 0.1V output change. So using a 1K resistor in the above scenario would keep the zener within it's comfort zone and result in a voltage variation of about 0.015V.

Temperature error would be a max of 7mV per degree C. At 0C the output will have dropped 0.14V and at 40C will have risen 0.14V. Typically this would be half this value.

Overall, not very accurate, but it depends on the application.

Mike.