RoadTest: TI HDC1000EVM Sensor Evaluation Module
Evaluation Type: Evaluation Boards
Did you receive all parts the manufacturer stated would be included in the package?: True
What other parts do you consider comparable to this product?: A variety of temperature/humidity sensors are currently on the market. This products accuracy is equal or better than some other higher priced alternatives.
What were the biggest problems encountered?: Evaluation GUI would unpredictably lock up, resulting in several nights lost data.
High humidity is a concern in manufactured houses (mobile homes) in the crawlspace underneath. This project was to determine the feasibility of remotely measuring both temperature and humidity in that environment.
The Evaluation GUI was already loaded on my computer when I received the EVM, so the device was running within minutes of opening the package. It was definitely easy to operate. The next several days were spent checking the ability of the GUI to reliably log data continuously, and compare the output to an NIST traceable temperature humidity sensor. At times, the GUI would lock up, resulting in partial or total loss of data. For the EVM or the sensor itself, this is not a game breaker. The EVM and GUI are meant to give an overview of the capabilities of the sensor, and they do that well. I had an issue loading the software onto my laptop, but that was a hardware issue (I've had other installed software fail to run). Luckily, the GUI is just a slightly sophisticated serial monitor front end. Any terminal program will effectively connect with the EVM. The output compared favourably with the NIST traceable sensor. The humidity fell within the stated accuracy of both devices. The EVM was consistently 0.8C higher than the calibrated sensor. It is possible that this is due to the other hardware on the EVM, however I didn't find self heating of the sensor until I increased the rate of sampling to under 0.2 seconds. In my profession experience, this difference is normal and expected. I will eventually determine the proper offset and implement it in software. The following chart is an example of nearly 2500 1 second continuous reading.
After getting used to the GUI, I separated the sensor from the EVM. I made a connector to utilize Cat5 cable, and connected the sensor at the end of about 7m (23 ft) of it. I deployed the sensor underneath the house in the crawlspace, approximately 20 feet from the end and 13 feet from either side. I attached the cable to the EVM, logged the connection with Teraterm, and logged just over 12 hours of data at 1 second intervals. About 1/3rd of the readings were errors with no reading, with a large number of readings that were impossibly high. As seen below, the times with no readings logged as zero, with the erroneous readings peaking over 100% humidity.
Finally I programmed an ATmega328p using the arduino library provided by ftruzzi . Connecting to the sensor that way, I added my own resistors to SDA and SCL which yielded better results. Again, I used Teraterm to log the data, using 15 second readings. The chart below shows a drastically decreased error rate.
The spikes on the temperature readings were singular erroneous readings. The EVM bears no blame for the number of bad readings, since it wasn't designed around a 7m cable length.
The HDC1000 is an excellent sensor for environmental monitoring. Further refining of my test setup is necessary as I was unable to compare temperature and humidity for the inside and outside in real time. A future setup will log the environment under the house and the local environment. Eventually, a system may be set up to automatically ventilate the crawlspace to improve temperature and humidity in that area. An additional use for this sensor is the monitoring of temperature and humidity in new timberframe housing. During the first couple years, it is important to control both to ensure an even drying of the timber. I would like to thank Element14 for this opportunity, and the other members of this Roadtest for their feedback and information.