INA350: How can I calculate the precision in the temperature measurement of my circuit?

Hello Martin,

We will answer the question after the weekend.

Hi Martin,

designing a precise circuit is one thing. But the true challenge is to maintain the precision after adding the protection circuits that are necessary to make the circuit withstand ESD, Surge, Burst and EMI, or by other words to make it pass the CE testings.

First, you should have an idea about the precision the circuit must have. This may forbidd certain approaches. Then, you should think about whether the customer has to calibrate the circuit each time after installing the PT100 sensor or changing the cable. And you should think about whether the customer has to additionally calibrate the circuit perdiocally after some time, even when not changing anything of the setup. The more the customer has to calibrate and recalibrate, the more long term drift of components and the less (uncalibrated) initial accuracy you can allow.

Then, you should be aware that a PT100 sensor is slightly non-linear by theory. This can be partially overcome either by the introduce of postive feedback in the first gain stage or by a software calibration, with the latter being preferred when you already use a µC.

Performing an error analysis is not so difficult. You have to consider all the offset voltages of OPAmps, their input bias currents and input offset currents which cause voltage drops across connected resistors, the manufacturing tolerances of resistors and other components and how all these parameters change with temperature (-> temperature drifts) and with time (-> long term drifts).

If you have more than one ADC input (of a precise ADC), you can also monitor the measuring current and catch and compensate changes of the measuring current by software. A good technique is the use of a ratiometric measurement where the imprecision of a certain parameter or parameters automatically cancel out.

TI has some nice front end chips for very precise PT100 measurements, by the way, which can heavily ease your design.

Kai

hi Ron Michallick,

I will be waiting for your answer when you can. Thank you

Regards.

Waltee

Hi kai klaas69,

Now I can understand that it's a lot of considerations haha ​​I really don't want the circuit to be very expensive but I saw some devices from the AD126x family and they seem to be a good option, could you help me find the best option for my application and maybe tell me if there are any guides? design with the devices that you indicate and design guides for protection circuits that work well with the device that you indicate? approximately, but In any case, I would like to be able to understand how to analyze the circuit that you showed to approximate the error in temperature, I know that it will be very useful in the line of products that we will make.

Best Regards.

Martin

Hi Martin,

I would look for a precision ADC containing an excitation current source for handling RTD sensors:

https://www.ti.com/data-converters/adc-circuit/precision-adcs/products.html#2192=Excitation%20Current%20Sources%20(iDACs)&sort=-1;asc&

The error analysis works like this:

Assume you have a measuring current of 100µA. Then, assuming a temperature of 0°C and a PT100 resistance of 100,000Ohm this will cause a voltage drop of 100µA x 100R = 10mV. If the OPAmp amplifying this signal has an input offset voltage error of 1mV, then you measure 11mV instead of 10mV. This translates to an erroneous PT100 resistance of 11mV / 100µA = 110R instead of 100R. From a PT100 table you can see that 110R corresponds to a temperature of about 26°C. So, your measuring error is 26°C.

All other error sources of the circuit can be translated in to a measuring error of temperature in this way.

Form this simple example you can see that amplifying a PT100 signal can be very challenging...

Kai

You are not going to get any kind of accuracy at all out of that circuit.  The excitation current source is far too inaccurate, and the 1.2mV offset voltage of the amplifier is unusable for RTD's.

Your best bet is to choose an inexpensive ADC that has integrated current sources.  ADS1120 is a good one.  ADS112C04 is another.

https://www.ti.com/lit/ds/symlink/ads1120.pdf

Page 56 has a perfect example.

Read the entire RTD section starting on page 52 for detailed instructions on calculating error contributions.

Hello everyone,
Thank you very much for all the contributions, with everything you told me I understood that the design was really bad for accurately measuring temperatures haha, I will do what @Bob Benjamin told me and I will create another thread in the indicated forum.

Good luck