Sure, you are correct, it can also be done without any resistors. 

The use of voltage dividers, mainly a resistor in parallel to the NTC is
to help to flat the curve and reduce the peak of high resistances in
very low temperatures, so you can gain in resolution.  Without any
parallel resistor, the high resistence span is very high, you have a
curve like this:

*
*
*
 *
  *
   *
     *
         *
                     *
                                            *

with a parallel resistor, you have something like this:

*
 *
  *
    *
       *
           *
               *
                      *
                               *
                                           *

and then, with the voltage divider, you can accomodate the exactly
voltage span you need to your ADC circuit, without using any other
front-gain amplifier, or something like that.

By the way, if your software can do nice calculations rather than use a
look-ahead table, you can use this formula:

12th Degree Polynomial for NTC 44006 10k@25¡C.
Functional range from -40¡C to +150¡C.
X = Celsius Degree.
 
NTC_Ohms = a + bx + cx^2 + dx^3 + ex^4 + fx^5 + gx^6 + hx^7 + ix^8 +
jx^9 + kx^10 + lx^11 + mx^12

Legend:
-------
a = 28059.557
b = -1358.1528
c = 41.513538
d = -0.77551934
e = 0.005170511
f = -1.999025e-006
g = 2.345903e-006
h = -7.3344661e-008
i = 8.6221767e-010
j = -4.3167639e-012
k = 2.6210006e-015
l = 5.3463809e-017
m = -1.4179029e-019



Sean Breheny wrote:
> 
> Hi Wagner,
> 
> I'll have to admit ignorance as to how linear NTCs are. What I was saying
> is that it seems to me that it is totally unnecessary to introduce
> additional non-linearity by using it in a voltage divider configuration.
> One could simply use it with a transimpedance amp. Then, any nonlinearity
> that you have comes only from the NTC itself, not from the voltage divider
> equation.