AD8495 cold junction for DAQ with digital filters

Hi CharlesC 

Thank you for reaching out and for consulting your setup. May I request several of your design requirements: like Supply operating voltage, Temperature measurement range, Schematics, and end applications?

Your setup is doing fine and below are my comments:

1) I would have divided the 5V in 4+1V. Having the 1V potential at Vref, the 5V potential at V+ and the 0V potential at V-. to be able to measure negative température (output at 0°C would be at 1V potential)
     

Arsol:   The AD849x series is configurable with a single and dual supply configuration. For Single supply (+2.7V to 36V) it is recommended to use 5V and Dual supply (+/-2.7V to +/-18V). For measuring negative temperatures and using a Single Supply configuration, level shifts the output by introducing a positive Voltage at the VREF pin (>+Vs). Normally, the Output Voltage is negative and less than the forced Vref value.  

2) the In- would be 1Meg to 0V potential to have the open detection and be able to use any thermocouple termination.

3) where I'm not sure is whether I need the capacitors as I can use a digital low pass filter of 1 Hz and remove everything else. (I don't need a higher DAQ frequency for temperature).

Arsol: In case, the digital filtering process will not work, you can use the recommended low pass filter design. This is useful when the intended application is situated in a noisy environment and the long thermocouple leads can act as antenna that will pick up unwanted signals.

4) Reading would be between Output (DAQS S+) and Vref (DAQS S-) whish would give 0V(+noise) at 0°C.

Arsol: To give better suggestions, may I ask for your schematics, please? If you have errors in temperature measurements. Remove the thermocouple wire from the socket to isolate the setup. Then, you can configure the AD8495 to act as a thermometer. By shorting to ground both inputs, Vref and -Vs. Lastly, short the Sense and Out pins.

5) On fabrication side, I intend to solder (regular tin?) a usual femal type k socket to type K wires. solder these wires to the chip pin and to the 1Meg. and all the remaining wires would be copper.

Arsol: you can use any type of solder, for the thermocouple socket place these as close to the -IN (Pin1) and +IN (Pin8) of the Chip. Keeping the reference junction close to the AD8495. Hence, will eliminate temperature errors. any type of socket you can use. Below is just an option from Omega Engineering, thermocouple J and K sockets.

Hi CharlesC 

Thank you for reaching out and for consulting your setup. May I request several of your design requirements: like Supply operating voltage, Temperature measurement range, Schematics, and end applications?

Your setup is doing fine and below are my comments:

1) I would have divided the 5V in 4+1V. Having the 1V potential at Vref, the 5V potential at V+ and the 0V potential at V-. to be able to measure negative température (output at 0°C would be at 1V potential)
     

Arsol:   The AD849x series is configurable with a single and dual supply configuration. For Single supply (+2.7V to 36V) it is recommended to use 5V and Dual supply (+/-2.7V to +/-18V). For measuring negative temperatures and using a Single Supply configuration, level shifts the output by introducing a positive Voltage at the VREF pin (>+Vs). Normally, the Output Voltage is negative and less than the forced Vref value.  

2) the In- would be 1Meg to 0V potential to have the open detection and be able to use any thermocouple termination.

3) where I'm not sure is whether I need the capacitors as I can use a digital low pass filter of 1 Hz and remove everything else. (I don't need a higher DAQ frequency for temperature).

Arsol: In case, the digital filtering process will not work, you can use the recommended low pass filter design. This is useful when the intended application is situated in a noisy environment and the long thermocouple leads can act as antenna that will pick up unwanted signals.

4) Reading would be between Output (DAQS S+) and Vref (DAQS S-) whish would give 0V(+noise) at 0°C.

Arsol: To give better suggestions, may I ask for your schematics, please? If you have errors in temperature measurements. Remove the thermocouple wire from the socket to isolate the setup. Then, you can configure the AD8495 to act as a thermometer. By shorting to ground both inputs, Vref and -Vs. Lastly, short the Sense and Out pins.

5) On fabrication side, I intend to solder (regular tin?) a usual femal type k socket to type K wires. solder these wires to the chip pin and to the 1Meg. and all the remaining wires would be copper.

Arsol: you can use any type of solder, for the thermocouple socket place these as close to the -IN (Pin1) and +IN (Pin8) of the Chip. Keeping the reference junction close to the AD8495. Hence, will eliminate temperature errors. any type of socket you can use. Below is just an option from Omega Engineering, thermocouple J and K sockets.

this is an adaptor that will be always between 0-30C.

measurement goes between -40 to 200C for thermocouples with whish we record temperature of tested part every seconds

Supply voltage comes from a DC battery powered data acqusitioin system with no ground. it's a 5V excitation voltage with 70mA max. 

i'm not sure for the resistance values to use

Hi CharlesC 

I forgot to mention that you have to keep the Reference Junction (this is the area where the thermocouple socket is connected to the 8495 pins) within 0~50℃.

On the schematics you plan to source 1V to REF; this will be done through the 100 ohms resistor. I suggest adding up a buffer between the 100ohm to the REF pin; using either an OP777 or AD8613 Amp.

I’m seeing this black connection as the low potential? Please modify your resistor divider circuit values and be able to produce 1V through the buffer. Then, from the buffer to the REF pin.

Another is the 1Mohm resistor should connect the green connection to the low potential.

regards,

Arsol

Hi CharlesC 

In addition to the filter suggestion, Resistors should be kept to about 100Ω if possible to preserve accuracy. So generally the trade-off is how large a capacitor you're willing to use.

Shielding can also do a good job reducing EM pickup and is worth trying if it's easier for you or if the filter is not enough.

Thanks a lot Arsol , i'm starting soon
I will use: OP777 or AD8613 for ref
I will connect 1mΩ to low potential (black) (pin#2)
For the hardware filtering with capacitors/resistance that you suggest. I wasn't going to make any as I don't anticipate EM and I have low pass filering (hadware and software) in the mesurment hardware after the AD8495 , but you seem to be pushing for it. Do you think it's not sufficient? 
DAQ measurment is between pin#3 and #4 in my assembly

Regardless of the length, a thermocouple can cause an antenna effect; this issue can generally be solved with a low-pass filter on the inputs (+IN and -IN). Another suggestion is if the filters are not enough to reduce the EMI and Noise; you can try shielding these input signals.  

 Another item to take note of is a ground loop. The thermocouple should only be grounded at one point. For example, if the thermocouple has a grounded tip, but you also connected one lead to the ground at your PCB, it would form a ground loop. This can also be an issue with non-isolated thermocouples if they are in contact with a material that has a ground connection.

Best regards