Effective temperature monitoring in high-voltage environments, such as automotive on-board chargers (OBC), DC/DC converters, and EV DC fast chargers, is crucial for maintaining safety and performance. These applications require isolated ICs because direct electrical connections can be hazardous and lead to inaccurate readings due to electrical noise and interference. Traditionally, negative temperature coefficient (NTC) thermistors sense temperature across isolation boundaries. This typically involves either using a high-voltage NTC with an isolated ADC or amplifier, which adds cost and complexity, or placing a standard NTC across the clearance gap from the high-voltage heat source, which results in poor temperature response time and significant measurement errors due to the thermal lag across the gap.
ISOTMP35 or ISOTMP35-Q1 solves these issues by providing a new isolated temperature sensor technology that allows to directly connect to a high-voltage heat source. This direct connection enables a much faster temperature response time of 3.1 seconds, compared to 78.3 seconds for NTC without epoxy and 47.8 seconds with epoxy. Additionally, ISOTMP35 achieves a more accurate final temperature of 72.1°C, making sure of better performance and reliability.
This application note demonstrates how ISOTMP35 enhances temperature response over other temperature sensing designs. By leveraging advanced isolation techniques and thermal management strategies, ISOTMP35 helps improve the safety and efficiency of critical applications such as automotive OBCs, DC/DC converters, and EV DC fast chargers, addressing the increasing power density and thermal challenges in modern EVs.
All trademarks are the property of their respective owners.
Temperature response time is defined as the 63% rise time from an initial temperature to a final temperature.
There are multiple ways to specify temperature response time. For a traditional temperature sensor, there are usually two methods: with stirred liquid (oil), and in air (still or moving). Temperature response time in stirred oil is done by taking sensors that are in an oil bath at one temperature, typically 25°C, then moving them very quickly to another oil bath at a different temperature. Temperature response time in air, is usually tested by plunging sensors from an ambient air environment into an oven at a specific temperature. If the sensors are plunged directly into the oven, that is moving air since the oven is moving air around on the inside. To achieve still air, the user can plunge the sensors into an enclosed box inside the oven, so the air does not circulate and remains still.
However, for an isolated temperature sensor, directional response time is also used. Directional temperature response time tests how the sensor responds when one side is exposed to heat, while the other side is not. This replicates a real use case scenario, where the ISOTMP35 can be connected to a high voltage pad or bus bar, and that high voltage area is generating heat. This test provides a repeatable and simple method to evaluate how an NTC performs while electrically isolated in the low voltage region and how the ISOTMP35 performs while directly connected to a high voltage heat source.
Figure 1-1 is an example of measuring the 63% rise time.
This example is a generic decaying exponential rise, with a 63% rise time value of 1.26, final value of 2. This curve is given by Equation 1.
The 63% time is 3s, and is determined by the denominator in the exponential. This is considered to be 1 tau on an exponential curve.
Currently, sensing temperature across an isolation boundary can be done with a non-isolated temperature sensor, such as an NTC. Isolation devices that can normally bridge an isolation boundary in a high voltage environment must meet a minimum clearance threshold depending on the high voltage level. If using a non-isolated device, it must be placed a minimum distance from the isolation boundary specified by the clearance. Clearance is the shortest distance in air between the high voltage signal pin and a low voltage signal pin. Creepage is similar to clearance, it describes the shortest distance between the high voltage and low voltage sides of a device along surfaces, such as the package or the PCB. Creepage can never be less than clearance. The minimum clearance required is determined by multiple factors, although it is primarily determined by the working voltage of the isolated device.
In the case of ISOTMP35, which is a basic isolation device, the minimum clearance is 4mm (the width of the package body). If not using an isolated temp sensor, a user must place their non-isolated sensor across the high voltage boundary at least as far away as the minimum clearance. The main drawback with this method is that while it is easy to implement, placing the sensor in the low voltage region several millimeters away from the high voltage heat source means the heat must pass across FR4 (standard PCB dielectric) which has a relatively poor thermal conductivity. This means temperature response time is reduced and the final achieved temperature is much lower than what it can be with a direct connection.
Medium | Thermal Conductivity (W/mK) |
---|---|
Galden Oil | 0.065 |
FR4 | 0.2 |
Gallium | 29 |
Graphite | 200 |
Copper | 400 |
Graphene | 1500 |
FR4 has relatively poor thermal conductivity, which means that the NTC can respond slowly to the change in temperature in the high voltage region and hence has both slow temperature response time and does not closely approach the final true temperature value. Both response time and final temperature value can be improved by using a non-conductive thermal epoxy to thermally couple the NTC with the high voltage heat source. While this can improve the thermal performance of the NTC, the performance is still not as good as direct metal-to-metal contact with the high voltage heat source.
ISOTMP35 is an analog temperature sensor that is capable of being placed across a high voltage isolation boundary. This allows the ISOTMP35 to be placed directly onto either a high voltage bus bar or onto the heat sink of a high-power FET. Because of this direct contact, heat does not need to transfer through PCB material to make it across the isolation boundary, but instead the direct contact provides immediate temperature response and a much more accurate final temperature value. This is important because a faster response time and more accurate final value can help provide better overheating protection in a variety of customer systems.
To prove the theory that the ISOTMP35 would provide superior response time to an NTC design, an oil bath was used to compare the ISOTMP35 with an NTC. The experiment was then repeated with an NTC with thermal epoxy.
The experiment is done by pouring liquid gallium inside of a crucible, which is heated by an oil bath set to 75°C. Then a PCB with a 1 square inch copper pad (2oz copper) and the device under test is submerged in that liquid gallium to observe how the device's temperature changes over time.
The thermal response setup used a Fluke 7340 oil bath, heated to 75°C (ambient temp is 25°C) with the lid off. A metal stand is placed inside the oil, but does not exceed the height of the oil so it stays fully submerged.
A graphite crucible is placed on top of the metal stand, with a 3D-printed ABS insert designed to fit the test PCB. Then the crucible is filled with liquid gallium. The working principle is that the crucible can be surrounded by the hot oil, so it can heat the liquid gallium very quickly. The top of the crucible is exposed to the air, so that the test PCB can be inserted without contacting the oil.
Liquid gallium is used because it has a thermal conductivity of 29W/mK and it can be liquid at room temperature. The oil bath uses Galden HT-200 oil, which has a thermal conductivity of 0.065W/mK, so it isn’t viable on its own for this test. Furthermore, the oil moves around while the bath is running, so it is impossible to control the submersion depth. The gallium inside the crucible is then heated to 75°C, and verified with a thermometer.