Reports
Contents
Title: | Peltier Chip as Net Radiometer Test | ||||||||
Date: | 2019-07-23 | ||||||||
Data File: | OxfordRadComp_Peltier_mV.csv OxfordRadComp_Peltier_mA.csv |
||||||||
Refers to: | 160848 | ||||||||
Dennis, Kyaw Tha, and Joe have been discussing using a Peltier chip sensor as an inexpensive net radiometer, so we tested it at Oxford tract by wiring it to a CR6 datalogger with other Rnet sensors and an AMP reference sensor CNR4 4-band radiometer sn 160848. The Peltier chip was installed on 2019-06-25 10:00 and was deployed for four weeks. From 2019-06-25 10:00 to 2019-07-12 16:45, we recorded the voltage from the Peltier chip. From 2019-07-15 15:05 to 2019-07-23 09:55, we added a 10 ohm resistor and measured current from the Peltier chip, because LEDs have a linear current response to light (but a non-linear voltage response to light).
Figure 1. 1-min voltage from Peltier chip with reference Rnet calculated from reference sensor. The Peltier chip mV reading has been linearly transformed to fit better on this graph. Overall, the Peltier voltage matches well midday, but the overnight values tend to be different, especially in the last few days of this campaign.
Figure 2. Linear regression between Peltier chip voltage and reference Rnet.
Fig 5. There is some hysteresis in the relationship between Peltier mV and Rnet (CNR4) reference (W/m2) at nighttime. Data shown is from 2019-06-25 to 2019-07-02. Data is color-coded by hour, 21:00-05:59.
Fig 6. The net radiation between the CNR4 and the RNET-LITE are similar, and the Peltier chip is still showing hysteresis relative to those two sensors. Data shown is from 2019-06-25 to 2019-07-02.
Figure 3. 1-min current from Peltier chip (with 10 ohm resistor across the H and L channels) with reference Rnet calculated from reference sensor. The Peltier chip current reading has been scaled by a factor of 1000000 to fit better on the graph. These two time-series match ok overnight, but not at all during the day.
Figure 4. Linear regression between Peltier chip current and reference Rnet. It's a mess. |
|