XTR105

Hello Ganesh,

Thank you for your schematic and the results.

JayantD had originally brought up an issue where the output of the XTR105 was varying enough to produce up to 30degrees of temperature error when back-calculating the temperature.  For a 0-200 degree 4-20mA output application, this would correspond to: ((0.02 - 0.004)/200)*30 = 2.4mA of output variation at any time.  The data you've provided shows a changing error with temperature, but does not describe the same symptoms that JayantD described.  Are you experiencing the same variation as Jayant on the output of the XTR?

Otherwise, I've plotted the error of your circuit output below.  Although it is not as parabolic like a standard RTD error, it does have a similar shape making me question the correctness of the Rlin1, Rlin2, and Rg resistor choices.

In a 3-wire RTD application with a 0-200C desired temperature range, the correct values for Rz, Rlin1, Rlin2, and Rg are:
Rz:  100
Rg:  158
Rlin1: 16200
Rlin2:  18700

Your schematic indicates the following values:
Rz:  100
Rg:  500Ohm potentiometer
Rlin1: 24000
Rlin2:  22000

Is this design meant for a 0-200degree RTD temperature range?  If so, please try your measurements again with the values above.  If not, please provide the following information:
Minimum Temperature
Maximum Temperature
Measured Value of the Rg potentiometer (Tuned in value is not on the schematic)

I will double check your component selection based on the information above.  Also, why use a potentiometer for the Rg resistor?  Were you not able to find a close enough standard value or are you trying to trim the error down with the potentiometer?  If it is the latter then beware that the value of the Rlin2 resistor depends on the value of Rg and if you change Rg you will have to make a corresponding change to Rlin2 or you will experience unwanted non linearity errors.   

Please provide some insight into the questions raised in this post and we will get this worked out.

Thanks and Regards,
Collin Wells
Precision Linear

Hello Jayant,

Thank you very much for the updated information and system specifications. 

Ganesh's most recent measurements seem consistent with what I would expect for the XTR105 if using rounded values for resistances.

Using the same XTR_CALC.xls tool that you've used already I entered in your system specifications and received the same values as you for the RG, RLIN1, and RLIN2.  Then I modified the calculated values for RLIN1 and RLIN2 replacing them with the values you've selected of 22k and 24k respectively.  It should be noted that I have still not received confirmation from Ganesh or yourself regarding the value of Rg that you have selected for your system, so I used 117Ohms as recommended by the calculator.  Then I entered in tolerances of 0.1% for each of the four resistors and added the typical errors for RLIN, IREF1, IREF2, and GAIN based on the datasheet typical specifications. 

The calculator predicted a total output error of up to 0.18% FSR which equates to a temperature error of (0.0018 * 150) = 0.27C.  By changing the resistor tolerance to 1%, the calculator predicts a total output error up to 2.5% FSR which equates to (0.025 * 150) = 3.75C.  Therefore without knowing more about your system I believe that the recent data Ganesh took is valid and appropriate. 

We've confirmed that the output error is in alignment with the calculator/spec using a controlled source so let's move on to the noise issue that was originally described.  You've mentioned that you're using an 8m long shielded cable with the shield connected to the CHASSIS GND.  Is the cable between the XTR105 and the main system receiver or is it between the RTD and the XTR105?  If the cable is not for the RTD, how far is the RTD from the XTR105 and are it's connections shielded?  If so, what potential is this shield connected to?  Have you tried measuring the RTD resistance with a lab multimeter while in the active alternator to see if the data from the RTD is constant.  Also please probe the RTD signals with a oscilloscope to see if you detect the noise there or not.   

Could you provide additional details about the cable you're using?  Are the signals twisted together inside of the cable?  Are the signals individually shielded in the cable or does the shield contain all of the wires?  How much of the sensor/cable is outside of the shield (in distance)?  Have you tried connecting the SHIELD to other potentials in the circuit besides the chassis GND connection?  Is there improvement when the shield is connected to the GND connection used by the 4-20mA receiver or when connected to the IOUT signal itself.   

I'm trying to get a better understanding of where the noise is entering the system and if it is present at the inputs to the XTR105 or if it is getting coupled into the 4-20mA output of the XTR105.

Thank you,
Collin Wells
Precision Analog 

Hi Jayant,

Thank you for the additional information.

Let me make sure I understand the circuit. There are three RTDs inside of the alternator, one for each phase. All three RTD outputs are carried in one 10-conductor shielded cable. Each RTD is a three-wire RTD so there are 3X3 = 9 used conductors and one unused conductor. This cable is routed 8 meters before the three RTDs are somehow (not described) selected and the three RTD wires of the selected RTD are routed another 1 meter to the XTR105PCB.

Is this correct? If so, how are the three sets of RTD signals multiplexed into the single XTR105 PCB?

I'm not sure if you still need to probe the RTD signals. From what you've described the noise from the alternator is getting coupled into the XTR105 inputs through the 8 meter RTD cable. The low-pass input filters that you've designed have a fc = 1/(2*pi*1000Ohms*0.1uF)= 1591Hz cut off frequency which will not be very effective vs. the 1500RPM alternator. Try lowering the cutoff frequency two decades to 15.91Hz by increasing the capacitor values to 10uF. This will slow down the response time of the temperature measurements, but if it fixes the noise issue you can find the appropriate compromise between measurement response time and noise.

That said, this system is somewhat backwards of the traditional implementation of the XTR105. Most XTR105 systems place the XTR105 very close to the sensor to keep the XTR inputs clean and then the XTR105 current loop output is routed on the long cable length between the XTR105 and the control system. The XTR105 and any other sensor electornics are powered from the loop supply and data is sent back on the loop return, this is the advantage of the "2-wire" current loop transmitter.  The output current from the XTR105 should prove much more resistant to noise coupling over a long cable than the high impedance inputs of the XTR105.

If the additional filtering does not prove successful, you may want to rethink the overall layout of your XTR105 system.

Regards,
Collin Wells
Precision Linear