ADS124S08: Designing an isolated thermocouple measurement system with grounded thermocouples

Hi Bob.  Thanks for the detailed reply.  I'll have a look at the links.  I've studied the Basic Guide to Thermocouple Measurements extensively.  

My design has 6 of these boards in a chassis with a host microcontroller.    I am using an ISOW7841 to isolate my power and digital signals on each board, so each board is isolated from each other and the host micro.  Each of my 5 thermocouples on each board are physically located close together on the furnace, so I think aside from a short circuit to the power system, they should stay close to each other in terms of common mode voltage.  It sounds like it would be OK to float the board's isolated section, tying the digital 3.3V ground and the analog supply midpoint together, but not to earth ground.  The thermocouples will then be at some common mode voltage above earth ground but that common mode voltage will also be imposed on the isolated ADC system. 

Does that seem reasonable?

Bob Benjamin said:

Hi John,

One consideration in that a grounded thermocouple (TC) does not necessarily imply and specific grounding point.  A grounded TC just means that the junction connects to the source or point of measurement.  If the grounded TC is connected to a motor housing, for example, that housing may be connected to an earth safety ground which in that particular case would be an earth ground.  But if I connected a grounded TC to some piece of metal sitting on a table outside, and that table is isolated from earth ground, I would just be measuring a floating heat source.

A good resource is A Basic Guide to Thermocouple Measurements.

Regarding your specific questions see my responses below.

Best regards,

Bob B

Part Number: ADS124S08

Hi.  I am designing a system for reading a large number (~600) of thermocouples using multiple modules. Each module using the ADS124S08 reads out 5 thermocouples and one thermistor (for CJC).  Each of the thermocouples is grounded, and all 5 of the thermocouples transition from thermocouple wire to copper wire at an insulated isothermal junction block, so all 10 thermocouple connections are at the same temperature, and that temperature is read by the one thermistor.  The raw voltages are sent upstream to a larger system which calibrates the data and converts it to temperatures.

I've designed a system using an isolated power supply derived from an ISOW7841 isolator, and an LM27762 regulator/charge pump to prevent a short circuit from 277V power in the furnace to one of the thermocouples from destroying more than one of the modules, in the worst case.  An LM 27762 charge pump/regulator is used to derive a bipolar supply so that the grounded thernocouple voltages fall in the middle of the power supplies.

I have a few questions about the design, especially about the grounding.  I'll attach a simplified schematic of the ADC section with one of the thermocouple circuits shown.

1) Should I solidly ground the middle point of the +-2.5V supplies, and the digital ground?  Or should I use a high-impedance AC/DC ground between the signals and the earth ground, bearing in mind that the earth ground is tied solidly to my thermocouples. [Bob] If your power supply and measurement system is truly isolated then there should not be an issue.  However from your schematic I only see one small portion of your system so I don't see how the measurement blocks come together and how the measurement data is passed through to a central processing unit.  As far as the ground connections, there is an absolute maximum rating in Table 7.1 of the ADS124S08 datasheet that specifies that AVSS to DGND must be within the range of -2.8V to 0.3V. As you are deriving the +/-2.5V from the DVDD 3.3V, you should connect the DGND to the AGND or mid-point of the generated bipolar supply.

2) I am thinking I should put clamp diodes on the signals after they go through the fuse and the 2770 ohm resistor to clamp any short circuit currents to the rails to protect the ADC inputs. [Bob] I would suggest using TVS diodes here to clamp to a specific voltage.  The protection circuit must also limit the input current through the input pins to +/-10mA as per absolute maximum ratings.  You may want to review the Precision Labs information in this EOS presentation:

https://training.ti.com/node/1139109?context=1139109

 

3) Is there any point in using an isolated power system if the signals are grounded anyway?  I was thinking that it would reduce the potential for ground loops, but I don't know if that's true or not. [Bob] If the power system is truly relative to a specific voltage potential and all of the grounds are the same (which is seldom the case) then you wouldn't need to isolate.  However, ground loops can always be a potential issue so this is a consideration.

Depending on how your voltage sources are generated the potential could be considerably different relative to earth ground between a control unit and the modules.  I have seen 40V and more between bench power supplies outputs when powered off from the same AC source.  So earth ground could be much different in potential than board ground.   Isolation is also a good safety choice as well in protecting the human element.

Any other observations about the design would be welcome!

ADC_schem.pdf