RTD excitation current

Hello Pasquale,

There are numerous ways to create current sources for your RTD sensor and there are numerous posts on this forum already dedicated to creating current sources/sinks.  Your assumptions regarding the two circuits are correct and the second circuit requires the precise matching of several resistors instead of the of only one.  However, from a stability standpoint the first circuit may experience issues due to cabling capacitance where the second circuit's op-amp output is isolated from the RTD cable with the 1.25k resistor, this may or may not be a concern depending on the distance of the RTD from the front-end circuitry. 

Below are a few app notes (including a recently released one from one of my colleagues), an example post on current source/sinks, and some example circuits. 

My suggestion to you would be to also look at a fully integrated solution for RTD sensors such as the ADS1247 or ADS1248.  This type of device integrates current sources, a PGA, and a sigma-delta ADC.  It would allow for a single chip solution that will produce very high accuracy without having to trim and tune several external components.

http://www.ti.com/lit/an/slyt442/slyt442.pdf 

http://www.ti.com/litv/pdf/sboa046

https://e2e.ti.com/support/amplifiers/precision_amplifiers/int-precision_amplifiers/f/16/t/140361.aspx

Regards,
Collin Wells
Precision Linear Applications

Tha REF200 offers supreme simplicity-- it can be connected to provide a precision output current of 100uA, 200uA, 400uA, and other connections as well, all without any external components!-- see the data sheet.

Thanks for the suggestions.
The REF200 could be a good solution but its temperature range is [-20:85]°C while I need [-40:85]°C, furthermore it is an old component and I would to avoid devices with risk of obsolescence.

My application requires a precision of about 1°C and the linearization is not important, since I do this digitally.
A fully integrated solution could be a good alternative but the problem is that I need to acquire 20 RTDs, while the number of inputs available on an ADS124x is not enough.

I thought to use an analog front-end circuit and a multiplexed SAR ADC.
Do you know if exist a simple analog front-end IC that contains only the excitation current circuit and the INAMP?

About the stability problem of the opamp you have mentioned, using the circuit in fig.1 the opamp see the parasitic capacitance associated at the RTD connection, thus I need to investigate on its capacitive load drive before using that solution.
The use of an external mosfet, driven by the opamp, can be useful because reduces the thermal drift of the opamp and avoids to handle heavy capaicive loads.

In the circuit shown in fig.2 the problem of heavy capacitive load is not a problem because the capacitance is in series with a R8 resistor, like a snubber.

Is this the problem ?

What is the precision reachable using the circuit in fig.2 ?

Considering I need to acquire 20 RTDs, what solution do you suggest ?

(The link: https://e2e.ti.com/support/amplifiers/precision_amplifiers/int-precision_amplifiers/f/16/t/140361.aspx doesn't work)

Thanks in advance

BR

Pasquale

Hello,

Thank you for the additional information.  Unfortunately for that many channels we do not offer a single-chip solution. 

You may want to consider looking at the ADS7953 and similar family of SAR ADCs a combination of two of these ICs will handle up to 20 channels.

We have one interesting part, the INA330 that was made for thermistors, this may or may not be able to work for what you need.  Besides that IC, we do not have a single chip analog solution that includes the excitation current and INAMP that is not also a 4-20mA 2-wire current loop driver.  I don't think this will be what you're looking for, but take a look at the XTR105, XTR115, and XTR116 for this type of circuit.  

I will leave it up to you to calculate the precision with figure 2.  It will depend on the accuracy of the current source and the accuracy of the discrete difference amplifier.  Errors in the current source will be from the inaccuracies of the 2.5V reference, offset voltage and noise errors from the amplifiers, and resistor mismatch.  Errors in the discrete difference amplifier will be due to input offset voltage errors of the amplifier and resistor mismatch. 

If you use modern high-performance amplifiers your main source of error will likely be resistor matching in the current source circuit and resistor matching in the discrete difference amp circuit.  Use good tolerance ( <1% ), low temperature coefficient resistors to help reduce errors or consider using an integrated difference amplifier instead of building your own. 

Regards,
Collin Wells
Precision Linear

Pasquale,

You may want to consider excitation of the RTD from a voltage reference through an accurate resistor. If the voltage reference is approximately 10x the voltage on the RTD, the excitation current is reasonably constant throughout the temperature range. It's obviously not constant enough, however, to use a fixed number for calculation of temperature. Still it is easy to calculate the resistance of the RTD from the voltage measured across it because it forms a simple resistor divider with the excitation resistor. Since you are already doing linearization in software, this is a very minor addition to the calculation.

The disadvantage of this approach is that it is more difficult to account for wire resistance leading to the RTD but perhaps this is not a large source of error in your application.

Regards, Bruce.

Pasquale;

If you have to excite 20 RTDs, is it possible to multiplex the current source to the RTDs? Since you are switching a current, the MUX  Ron should not be a problem. You would need to MUX the RTDs slow enough to allow settling before sampling the voltage across the RTD but that should not be a problem since the RTD thermal response time is not very fast.

You would need to excite the RTDs with a current low enough to cause little self-heating but high enough to read accurately.