How to Use Differential Amplifier to Measure 0-50°C Human Body Temp Difference (5cm)

i am looking for a circuit to measure difference in temperature (+ or -) on the human body.
the distance is between two thermal sensors is about 5 cm, and the reading range is between 0 -50 Celsius.
i have a circuit but i need to make it more sensible. i was thinking of adding differential amplifier.
any one has got an idea of how to improve that by adding amplifier
i am struggle with the circuit
many thank in advanced

In the attached circuit the differential amplifier should be attached where the METER is drawn.

i am stuck with the circuit because i don't know how to connect the temperature sensor because it has 3 pins, and i don't know which one is going to the human body
i will be very grateful if anyone can redraw the circuit for me and i don't know whether the circuit will work or not.
this what i want to acheive but i don't know whether the circuit will work or not
http://www.autospeed.com/cms/article.html?&title=Differential-Temp-Meter-Part-2&A=111357

In the datasheet of the LM335 is stated which pin has which function depending on your package (TO-92, SO-8, ...). Neither pin goes to the body but the Sensor itself needs to be attached where you want to measure temperature (the third pin is for adjusting the sensor output). Regarding its large size (big thermal mass or heat capacitance) the LM335 is IMHO not very suitable for measuring small temperature differences of an object near room temperature. Maybe it would be wiser to use a Pt100/Pt1000 with a very small footprint, put it directly onto the skin and isolate it against the ambient temperature. If you wire them in a Wheatstone bridge you would get a sensitivity of approximately 1mV/K (Pt1000 resp. 0.1mV/K using two Pt100). This voltage can be easily amplified if you are paying attention to noise and clibrate the bridge.

If you wanted to use the attached circuit:
- the wires to the sensors (dashed boxes) should be same length
- the gain of the amplifier is calculated by A = (R2 + R1) / R1 thus the sensitivity of the circuit is approximately (R2 + R1) / R2 * 1mv/K
- The lowpass filter at the output of the amplifier can be used to suppress noise, and its corner frequency is calculated by f = 1 / (2 * pi * R3 * C1). Its useful value depends on the expected speed of temperature changes you want to monitor. If they are very slow it would be necessary to use an active filter, otherwise the value for the capacitor would become very large. But the circuit works without this filter as well.
- the potentiometers R4 and R5 are used to set the zero point
- your voltage output has a positive value if the temperature at the second sensor (Pt1000-2) is higher than at the first one
- you need positive and negative supply voltage for the opamp as well

hi ENRICO KOECK
i am very grateful of what you have done
in order to get the difference between two sensors i have to calibrate one of them to zero then connect the other to have the difference
is there any method to get the temperature difference without calibrating one of them?
i have done some calculation and would like to know your opinion of their values
please see the attached circuit

I'm sorry, I can't open your attached file because I don't have MultiSim installed. Maybe you can post a screenshot. Well if you need the sensors only over a small temperature range (as you stated) it is maybe sufficient to adjust the potentiometers only once or once in a while to level out the resistance difference caused by production tolerance. The prequesit would be that the cables for the sensors are exactly the same length otherwise you get errors caused by the resitance of the cables and the temperature dependence of it. If you want the system to calibrate itself you need some kind of microcontroller in my opinion. And you have to watch out for the self-heating of the Pt1000 which is about 0.5K/mW (P = U^2/R). So you should consider putting resistors in series with the Pt1000s (witch lowers the sensitivity) or use a current source.

this is the circuit

this what i want

To your schematic: If you only use a positive power supply you have to change the amplifier to a differential amplifier with the second voltage set to V+/2 (in this particular case 9V/2 = 4.5V). As well if you have a difference of zero between the two sensors your output would be 4.5V. This makes it very hard to detect small voltage differences because thoose are added to the virtual zero of 4.5V. Second the voltage over the PT100's is way too high. It would mean that a current of approximately 4.5mA flows through them wich results in a power dissipation of roughly 20mW. This would result in a temperature increase of about 10K in the Pt100. So it isn't a sensor any longer but a heater.
But I don't understand your second picture. Do you mean the temperatures should be added? Or if you just connect one probe it measures the difference to what? Or displays absolute values?

hi
regarding second picture i would like to connect both sensors at the same time to measure the difference in temperature between those two point which mean subtract them from each other.
when i have 0 that mean both sides of the vertebrae have the same temperature.
when i have difference in the temperature that mean i have trapped nerve on that point.

So you are talking about temperature differences in the range of 1/100s of K? Those changes are pretty hard to detect with this kind of setup. You would need pretty high quality components and an instrumentation amplifier circuit which is way more complex. So either take something already on the market or let it be designed by someone with experience in the field. Otherwise there are many books on designing amplifier circuits and sophisticated measuring equipment.
I belive you want to rebuild something like this http://www.theinstrumentcompany.com/?

i appreciate your advice
which is totally right, it is better to buy ready made one as you said
thanks for your help