7 Layout Techniques

Remotes measure very small voltages using very low currents; therefore, noise at the device inputs must be minimized. Sources of errors include Electromagnetic Interference (EMI) or inductive coupling, incorrect calibration, and PCB trace resistance. To reduce errors please consider the following conditions:

1. Place the remote device as close to the remote junction sensor as possible
2. Route the D+ and D– traces next to each other and shield them from adjacent signals through the use of ground guard traces. If a multilayer PCB is used, bury these traces between the ground or V+ planes to shield them from extrinsic noise sources. TI recommends 5-mil (0.127mm) PCB traces.
3. Minimize additional thermocouple junctions caused by copper-to-solder connections. If these junctions are used, make the same number and approximate locations of copper-to-solder connections in both the D+ and D– connections to cancel any thermocouple effects.
4. Use a 0.1μF local bypass capacitor directly between the V+ and GND of the remote temperature sensor. For optimum measurement performance, minimize filter capacitance between D+ and D– to 1000pF or less. This capacitance includes any cable capacitance between the remote BJT and the temperature sensor.
5. If the connection between the remote temperature sensor and the remote sensor is wired and is less than eight inches (20.32 cm) long, use a twisted-wire pair connection. For lengths greater than eight inches, use a twisted, shielded pair with the shield grounded as close to the remote IC device as possible. Leave the remote sensor connection end of the shield wire open to avoid ground loops and 60Hz pickup.
6. Thoroughly clean and remove all flux residue in and around the pins of the device to avoid temperature offset readings as a result of leakage paths between D+ and GND, or between D+ and V+.

Figure 7-1 Recommendation for Layering the D+ and D- Trace

With a multichannel EVM, data was collected to determine the ratio between the trace length and the accuracy of the temperature sensor. The EVM shown in Figure 7-2 was placed in an oil bath with a controlled temperature environment at 25C. Data was collected with the online GUI and reported in the document. Shielded cabling was used for the remote wiring. As shown through this experiment, there was not an evident relationship between the length of the wire and the accuracy of the remote channel. Noise can still interfere through the D+ and D- and all recommendations listed above should take place.

Figure 7-2 Test Setup With Multichannel Remote EVM

Figure 7-3 Error Across Different Remote Lengths