cant design a thermistor as an interrupt source

[rainman1]

I need a design a circuit that includes NTC as an interrupt source for MCU.
When it gets warm, the MCU pin needs to receive 3.3V.
I have to power sources for the circuit, and can use one of them - 3.3v and 5v.
the MCU pin can only sink up to 4mA.

Any ideas?

Thanks!

More:
I havent picked a right NTC yet, i'd like it to be around 10Kohm, but its not a must.

 

[blueroomelectronics]

Use a comparator.

 

[rainman1]

I though of something like that:

Where R1 is the NTC.

Edit:
What is your opnion?

To Blue.
I'd like to use resistors only, i needed to mention it in topic.

 

[rainman1]

its not good, becaues the voltage on R2 (output voltage to MCU pin) will be 3.3V even in 30 deg. celsius

 

[blueroomelectronics]

Then use a microcontroller with built in comparator. Resistors aren't enough.

 

[rainman1]

I would like to use resistors only, thanks.

 

[blueroomelectronics]

Well resistors won't cut it.

 

[rainman1]

Never mind!
Thanks anyway, i got the problem solved thanks to CHAREL!!

 

[blueroomelectronics]

Care to share the solution?

 

[rainman1]

Care to share the solution?

Vout that goes to MCU is voltage on R1.

What do you think?
Do you see any problems with this design?

 

[blueroomelectronics]

Yea, microcontrollers want a digital input. You're putting an analog voltage on a digital input pin. Not recommended.

 

[duffy]

Yes - what's "warm"? What's the NTC curve look like? Without a comparator you are goofing around with poorly defined trip points over what may be a uselessly small range for that digital input port.

 

[rainman1]

Blue.
Why is it a trouble for the MCU to receive analog input within its range limit?

duffy.
The NTC is NCP15XH103F03RC, its B is little above 3000. R0 is 10Kohm

 

[blueroomelectronics]

Because it will cause excessive currents on the input pin, what microcontroller are you using?

 

[duffy]

Current's not the problem, it's the undefined trip point voltage between Vhi and Vlow.

 

[blueroomelectronics]

There's an note on Microchips site about analog signals on digital inputs, that's why all A/D inputs default to analog.

 

[duffy]

Never heard of that one. Can you point to a link, or tell me how to find it? There's 57,600 results for "analog" on the site, none of the combinations I tried showed me the link.

I've used analog signals on digital pins many times for the quick and dirty A/D trick where you time an RC circuit. Never heard of an issue with the current.

 

[blueroomelectronics]

I'll dig through it or check the Microchip forums. Just something I stuck away in the back of my head.

 

[3v0]

Current's not the problem, it's the undefined trip point voltage between Vhi and Vlow.

In the undined region an input transistor acts analog which draws too much current. At least that is what I recall.

 

[rainman1]

In the undined region an input transistor acts analog which draws too much current. At least that is what I recall.

Could you please explain what you said?
I didnt quite understand it.
What transistor are you talking about?

*PLUS:
Why does it sink too much current over its limit?
I'm defining the input pin of MCU as tristate, so whats the problem?
(and even if i defined it as pull-up/down, still why would it draw more current?)