Temperature Sensors

There are two types of temperature sensors used:

• DS18B20 temperature sensor. This tiny microchip is very accurate and durable. They come as either the raw chip or embedded in a stainless steel probe with several feet of wire attached. Accuracy is +/- 1 degree F. The maximum temperature for these sensors is 257 degrees Fahrenheit. The base chip is about $2. The chip embedded in stainless steel costs approximately $5 each. I purchased a bag of five sensors with 3 meter leads for $18 from aliexpress.com. DS18B20 sensors output a digital signal and can be connected directly to the Arduino digital pins. Each sensor has a unique address and the data pins are all connected together to form a mini data bus that is connected to a single Arduino digital pin. See https://www.sparkfun.com/products/245 for more information.

• K-Type Thermocouple. These are very rugged and can handle up to 2000 degrees Fahrenheit but are not as accurate as the DS18B20. Accuracy is within 4 degrees F (2.2C) and is influenced by the installation. Adding a new length of wire to the sensor's leads can alter the reading due to the connect, even if it is soldered. Thermocouples require an amplifier and thus cannot be connected directly to the Arduino. A small break-out board has the amplifier onboard and sits between the Arduino and the thermocouple. This project uses the MAX31850 break out board because it allows the thermocouple data to look like a DS18B20. This means the DS18B20 is interchangeable with K-Type thermocouples. See http://www.omega.com/prodinfo/thermocouples.html for more information.

DS18B20 Bare Chip

Package of DS18B20 Probes

Building Sensors

The DS18B20 is preferred over the K-Type thermocouple when the temperature permits. Thermocouples output a low voltage and can be influenced by splicing on additional wire to the existing lead.

Bronze Plugs

Most cooling circuits have bronze drain plugs. A small recess can be drilled into the drain plug and a bare DS18B20 can be inserted and secured with potting compound or high temperature epoxy. Solder the leads onto the DS18B20 before potting it into the drain plug, and ensure the leads are embedded in the epoxy or potting compound. Mixing some Cab-o-sil into the epoxy allows you to build up the compound to secure the lead wires.

Embedding the bare sensor into a bronze plug provides readings that are a few degrees lower than the fluid being measured. See below for comparisons on bonding sensors. I am working on different sensor setups, below is one of a barbed NPT fitting. One end is soldered shut and the chip is epoxied on the inside. I have some nickel-copper threaded rod on order and I plan to drill out a recess for the chip. I also am experimenting with plastic threaded rod as that would isolate the sensor from the surrounding metal which is cooler than fluid on the inside.

Stainless Steel Probes

These are rugged, low cost, and easy to bond to whatever you want to measure. They can be attached with JB-Weld or JB SteelStik however this technique is not recommended. It is much better to use the bare chip and embed it into a bronze plug as described above.

In many cases you may not care about the exact reading but how it changes over time. For example the raw water temperatures are less interesting than the difference in temperature between the raw water where it enters the system and where it exits.

K-Type Thermocouples

These come in a huge variety of forms. They are available as threaded probes and also NPT fittings. Aliexpress.com has a large variety. They are also available as a washer with a lead attached, allowing it to be secured to the engine block:

Cost

There is a huge range in cost for K-Type thermocouples. The washer style in the image above can be found on aliexpress.com for about $5, or on performance engine shops like jegs.com for $93. While the quality of the thermocouple is a factor in its accuracy, the quality of the amplifier and installation generally has a greater impact. I used a $5 thermocouple for my EGT and it has read consistently for the past year.

Comparisons of DS18B20 Installation Options

Ideally the bare chip is immersed in the fluid however this is not practical. I setup a test rig to compare temp readings from different installation techniques. My engine has 2" diameter steel pipes that take coolant to and from the heat exchanger. The descriptions below are focused on measuring the temperatures of the coolant flowing to and from the heat exchanger. While these values do not indicate the temperature of the coolant in the engine they do provide very good information on the efficiency of the heat exchanger. Fouling of the heat exchanger results in a decrease in the delta between the two sensors.

The simplest technique is to epoxy the bare chip directly to the steel pipe. Unfortunately the reading is much lower than the fluid inside.

The next technique is to use the bronze plug pictured above, drill a hole in the pipe and weld on a bung.

I had some spare 2" pipe of similar thickness and setup both techniques on the bench. I capped one end of the pipe and filled it hot water. The difference in sensor readings show the simplest technique is unacceptable. I used a bare chip immersed in the hot water as the reference point, then took readings over a few minutes to see how quickly the sensors responded.

• The bare chip took about 20 seconds to reach it highest temperature of 188 degrees.
• The epoxied chip took several minutes and the highest temperature it reached was 144 degrees.
• The bronze plug chip took about 40 seconds to reach its highest temperature of 181 degrees.

From this experiment it's clear that the chip must be insulated from the steel pipe and highly conductive metal must be used between the chip and the fluid.

I have some bronze threaded rod on order. I plan to take a 1" long piece and drill it out for the sensor. It won't be drilled all the way out and there will be about 1/8" inch of bronze remaining at the end of the tunnel. This technique eliminates the need for welding on a bung - it has a double-nut and o-ring on each side and should keep it sealed and secure. The O-rings serve dual purposes: sealing and insulating the sensor from the surrounding metal.