Single-point calibration of the RTS node is easy to perform. One method is presented here using the water ice-point and works with both waterproof and non-waterproof temperature sensors. The RTS firmware provides both offset and gain calibration coefficients. For most purposes, a single point offset calibration is probably sufficient. To accurately perform both gain and offset calibration requires more sophisticated VBE equipment and is beyond the scope of this investigation.
Two sensors are used, one encased in epoxy and the other with no insulating material around the leads. Both sensors are standard cheap jerseys from China 2N3904 transistors with base and collector tied together.
The temperature standard consists of an insulated mug filled with packed, finely crushed ice made from distilled water and a small quantity of liquid distilled water to fill in the gaps and make sure cheap NBA jerseys the mixture is at the ice-point. Inserted in the middle is a test tube filled with pure mineral oil into which the sensors are inserted. Mineral oil is used to prevent leakage currents between cheap NFL jerseys the leads of the uninsulated sensor. If only waterproof sensors are to be calibrated, the test tube and mineral oil is not required – Dray the sensor can be inserted directly in the water/ice mixture.
The entire apparatus is then placed in a small cooler filled half-way with ice and the lid closed. This secondary container and ice was used in an attempt to create a cooler atmosphere surrounding the test tube to minimize the errors due to heat conducting down the test tube walls. It is uncertain if this actually helps, but any negative impacts seem minimal.
After inserting the sensors, the apparatus must be allowed to sit and con achieve thermal equilibrium. It takes a few minutes for the mineral oil to cool down to the temperature of the surrounding ice/water mixture and for the sensors themselves to reach equilibrium with the mineral oil.
Samples were taken using both sensors connected to both channels A Nfl and B of the RTS node to separate any sensor effects from RTS errors. Forty pre-calibration samples are plotted and summarized below.
|Channel A||Channel B|
The average error was +0.832C. A correction of -0.8125C (0x8D) was programmed into the Channel A and Channel B offset registers and the measurements repeated.
|Channel A||Channel B|
Since an average of all measurements was used to calculate the offset correction, the remaining error is likely due to the channel mismatch on the LTC2990. Based on the data above, this error is well within the +/-1.5C maximum Remote Diode Temperature Total Unadjusted Error from the LTC2990 datasheet. In practice, calibration of a specific sensor on a specific channel will give the best results.
In the example above, an offset was applied to calibrate the sensors around the ice-point of water. If the sensor is primarily used to detect freezing conditions, this is sufficient. Technically speaking, the errors due to the sensor are likely based on the ideality factor not being the 1.004 assumed by the LTC2990. Given that, a gain correction is likely a better solution than an offset correction. However, without an accurate 2nd temperature point for a two-point calibration, demonstrating this is difficult and is left as an exercise for the reader… Seriously, though, if you come up with a reliable and simple two-point calibration method, we would like to wholesale Cleveland Browns jerseys hear from you.