LTC2986 Testbench software, result seems wrong

LTC2986 temperature measurement IC

I’m using the LTC2986 Testbench software to design my schematic.
Results are below:

The first result shows a sense resistor, and three, 3-wire, RTDs.

Notice that the sense resistor at Ch1 is connected to ground.

The second result shows a cold-junction sensor diode, a thermocouple, a sense resistor, and an RTD.

Notice that the diode at Ch1 and thermocouple at Ch3 are connected to ground.

Notice that the sense resistor at Ch5 is not connected to ground.


Why is the sense resistor in the first screen grab connected to ground and the sense resistor in the second screen grab is not?

The two sense resistors are connected differentially. If one side is grounded doesn’t that make it single-ended?

In the second screen grab the cold-junction diode sensor is connected differentially, but one side is grounded.
The thermocouple is connected differentially, but one side is grounded.

The datasheet shows these ground connections, so they must be correct.

In particular, I don’t understand why one sense resistor is grounded and the other is not.

I need some help to understand.


Fair question. The matter seems related to whether or not a given input pin is configured to act as a sink for excitation currents, as well as a measurement input. Table 32 in the datasheet summarizes 10 different supported configurations for RTD connections. The screen caps shown would appear to be indicating the opposite of the chart (third and 4th lines down) corresponding to your two scenarios, with respect to the use of an external ground connection. It’d be a nice thing to clarify, but planting a footprint for a zero-ohm jumper to GND that one can populate (or not) would seem like a plausible workaround. My instinct would be to trust the datasheet rather than the software. Take a look at the config code that the tool spits out and compare it to the register descriptions in the datasheet–that might offer some insights.

All voltage measurements are inherently differential, similar to the way that one needs to define two points in space in order to measure the distance between them. The distinction between “single ended” and “differential” in this case would be whether the internal ADC measures an input relative to the COM pin (single ended) or between two Ch xx pins (differential).

Use of “differential” connections limits the influence of currents flowing through one’s ground network on resulting measurements, at the expense of using more input channels. A number of the different supported sensor types appear to be supported either way, so whether the software chooses to use one approach or the other in a given circumstance is probably a matter of assumption or some presumption on the part of whoever developed the tool.