Design review: Temperature / Humidity sensor

[krook]

Evaluate alternatives to the DHT22 currently in use to optimize for reliability and robustness.

The sensor currently used is the DHT22 temperature/humidity module, which is an Adafruit rebranded version of the AM2303 module from Aosong Electronics. This particular sensor presents some issues regarding manufacturability, scalability, and design appropriateness to upgrade it from prototype to product.

First, this is a through-hole component. In general, this isn’t a problem, but it is preferable to eliminate as many processes as possible from the assembly. A through-hole component should only be selected if there is not a suitable surface mount alternative, or if it offers a particular mechanical advantage.

Second, the sensor itself has limited availability and the original manufacturer is of unknown reputability. Generally, unless the product is highly specialized and can only be acquired from a single source, risks regarding the manufacturer should be mitigated.

Next, the interface used to communicate with the sensor is a non-standard single-wire asynchronous bus protocol. Generally, it is advisable to prioritize standard communications protocols that would be easy to port to a new sensor in the future if necessary. Further, asynchronous single-wire protocols like this generally have fairly strict timing requirements to properly encode data, and this can be a risk in harsh environments where oscillators may lose precision.

Finally, the maximum operating temperatures raise some issues since this device will need to be directly exposed to the harsh environment in order to properly measure temperature. The device is rated to 125C, which is significantly less than the temperatures the device may be exposed to in the worst case. Most of the electronics will need to be heavily insulated to protect them from these environmental hazards, but the temperature sensor, in particular, is one that is required to be directly exposed to the atmosphere, otherwise, it cannot reasonably acquire accurate environmental data.

With these considerations in mind, it is not easy to find a humidity sensor that is rated to the extreme temperatures that fire ground equipment is typically expected to endure. In this case, the minimum target should be operation at 140C for 10 minutes, but ideally, it would meet 260C for 5 minutes.

There are some alternative solutions that should be discussed which could solve all of the noted concerns.

First, it is possible that atmospheric humidity is not a critical piece of sensory information. This is something that should be critically evaluated before further engineering resources are spent trying to incorporate a specialty high-temperature sensor. If this sensor is deemed critical, the next step would be to contact reputable sensor manufacturers for recommendations on how to incorporate the sensors more safely.

It is also possible that the humidity sensor could be entirely separated from the environmental temperature sensing. In this way, the device could be better insulated (although not perfectly), so that humidity can be measured while not exposing it to the harshest environment. The temperature could then be captured by a generic K-type thermocouple with a high precision ADC and cold junction temperature sensor on the PCB.

Such a solution might be designed around the CC2D23S-2500 humidity sensor from Amphenol. This sensor uses an I2C interface for communication, has reasonably high availability, and comes in a surface-mount package. If the device is also insulated partially from the environment, this may solve all of the noted concerns.

In this solution, the temperature could then be measured with generic devices such as any K-type thermocouple, a high precision ADC such as the MCP3421, and a cold junction temperature sensor such as the MCP9800, both from Microchip.