Week 8 Challenge:Thermal Management

Abstract:  Control functionality of modern vehicles is getting more and more complex. Programming complex embedded systems involves reasoning through intricate system interactions along paths between sensors, actuators and control processors. This is a time-consuming and error-prone process. Furthermore, the resulting code generally lacks modularity and robustness. Model-based programming addresses these limitations, allowing engineers to program by specifying high-level control strategies and by assembling common-sense models of the system hardware and software.

Introduction:

Today’s software development has big challenges to master like shortened development times for the cars in total versus longer development times for the software, high safety requirements and especially the growing complexity because of the rising number of functions and the increasing interaction between the functions. To master these challenges car producers and suppliers conduct a paradigm change in the software development from hand-coded to model-based development. 2 A model-based development process is specifically attractive in embedded domains like Automotive Software due to the fact that development in these domains is driven by two strong forces:

• On the one side the evolutionary development of automotive systems, dealing with the iterated integration of new functions into a substantial amount of existing functionality from pervious system versions;
• On the other side platform-independent development, substantially reducing the amount of reengineering/ maintenance caused by fast changing hardware generations. As a result, a model-based approach is pursued to enable a shift of focus of the development process on the early phases, supporting a function based rather than a code-based engineering of automotive systems.

Earlier, many embedded applications were developed using assembly level programming. However, they did not provide portability. This disadvantage was overcome by the advent of various high-level languages like C, Pascal, and COBOL. However, it was the C language that got extensive acceptance for embedded systems, and it continues to do so. The C code written is more reliable, scalable, and portable; and in fact, much easier to understand. Embedded C Programming is the soul of the processor functioning inside each and every embedded system we come across in our daily life, such as mobile phones, washing machines, and digital cameras. Each processor is associated with embedded software. The first and foremost thing is the embedded software that decides to function of the embedded system. Embedded C language is most frequently used to program the microcontroller

What is C Language?

C language was developed by Dennis Ritchie in 1969. It is a collection of one or more functions, and every function is a collection of statements performing a specific task.
C language is a middle-level language as it supports high-level applications and low-level applications. Before going into the details of embedded C programming, we should know about RAM memory organization.

The main features of the C language include the following.

• C language is software designed with different keywords, data types, variables, constants, etc.
• Embedded C is a generic term given to a programming language written in C, which is associated with a particular hardware architecture.
• Embedded C is an extension to the C language with some additional header files. These header files may change from controller to controller.

What is an Embedded C Programming ?

In every embedded system based projects, Embedded C programming plays a key role to make the microcontroller run & perform the preferred actions. At present, we normally utilize several electronic devices like mobile phones, washing machines, security systems, refrigerators, digital cameras, etc. The controlling of these embedded devices can be done with the help of an embedded C program. For example in a digital camera, if we press a camera button to capture a photo then the microcontroller will execute the required function to click the image as well as to store it. Embedded C programming builds with a set of functions where every function is a set of statements that are utilized to execute some particular tasks. Both the embedded C and C languages are the same and implemented through some fundamental elements like a variable, character set, keywords, data types, declaration of variables, expressions, statements. All these elements play a key role while writing an embedded C program.

The embedded system designers must know about the hardware architecture to write programs. These programs play a prominent role in monitoring and controlling external devices. They also directly operate and use the internal architecture of the microcontroller, such as interrupt handling, timers, serial communication, and other available features.

1.  Differentiate between the embedded C programming and model based system approaches. 

2. a.) Draw a high-level schematic of a remote controller toy car system. Explain the working principle of the toy car system?

 The first remote controllers were developed in the early 1990s, and the first remotes were connected with wires to devices. Nowadays remotes use infrared control and thus are capable of controlling several things at a time. The remotes are not only used for entertainment, but also for industries, military requirements, and recreation. Infrared remote controls were developed in the late 1970s. These remote controls are based on the principle of using infrared light as the medium of communication. And use photo receptors and different light frequencies for different functions. These remotes also use invisible light beams to send signals to electronic devices. Radio-controlled or remote-controlled toys, popularly called RC toys, are self-powered and can be controlled from a distance using a remote that works with radio waves.

1. When we push the control, the transmitter sends a specific number of electrical pulses corresponding to that action through the air. The transmitter has its own power source, usually in the form of a 9-volt battery. Without the battery, the transmitter will not be able to send the radiowaves to the receiver.
2. Once the RC toy receives the radio waves, the motors kick into life to cause a specific action to occur. The power source sends power to all working parts, including the motor. The transmitter enables control through radio waves and the receiver activates the motors. When we press a button on the transmitter to make the RC toy go forward or backward, a pair of electrical contacts touch. Receiver identifies signals, sends it to circuit.
    Circuit board translates the number of electrical pulses (signals) into action. Full-function controllers have six controls and these works through following the pulse sequences:
3. 1. Forward: 16 pulses
   2. Reverse: 40 pulses
   3. Forward left: 28 pulses
   4. Forward right: 34 pulses
   5. Reverse left: 52 pulses
   6. Reverse right: 46 pulses

BLOCK DIAGRAM:

To build a RC car we need to make sure that these blocks are available. The above blocks are divided into two sections Remote and Car for the purpose of understanding. Starting from Keypad which controls the movement of the Car whereas Encoder and decoders are meant to encode and decode the movement signals for secured transmission. Also transmitter sends the movement signals through wireless medium and receiver fetches the signals for the destination.

CIRCUIT DIAGRAM OF REMOTE CONTROL: (Transmitter)

The above Remote control Circuit consists of three important components Keypad, Encoder IC HT12E and a RF Transmitter RF433 module.

KEYPAD: The Keypad for our remote is made up of four individual buttons which is wired to the data pins of the data pins AD0 to AD3 of the encoder IC. When these buttons are pressed it passes the movement signals to the encoder,

ENCODER HT12E: This encoder serves the purpose of encoding the movement signals fed by the individual buttons. The Data pins AD0 to AD3 is active low hence pressing the button closes the circuit and a logic 1 or high signal is fed to these pins. It also contains Address Pins A0 to A7 which allow us to try out different address combinations for secured transmission of data but make sure you use the same address combination in the receiving end (decoder). In the above circuit we did’t try out any address combinations hence all the pins are grounded. The Dout Pin gives the encoded output data.

RF433 TX: This is a simple RF433 TX module which operates at frequency at 433MHz. The encoded data from the Encoder is fed into DIN pin of the TX and a simple antenna is attached to its 4th pin.

OSCILLATOR FREQUENCY: The Oscillator frequency of HT12E is 3.25Khz (refer the oscillation frequency vs supply voltage graph in the datasheet). This frequency is fixed using the resistor R1 ( 1.1M ) . Modifying this resistor value will alter the output frequency.

CIRCUIT DIAGRAM OF REMOTE CONTROL: (Receiver side)

RF433 RX: This simple RF433 RX module which operates at 433MHz. The received encoded data signals are received through the antenna and the signal is obtained from DATA pin to feed the decoder.

HT12D DECODER: The encoded signals from the RX module is fed into the DIN pin of the decoder. The signal is then decoded by the decoder; remember to use the same address combination used in the encoder if any otherwise the movement signals will be interpreted incorrectly. The decoder also consists of VT pin which serves as a identification if any RF link is established, LED was connected to this pin for identification.

OSCILLATION FREQUENCY: For successful reception of incoming signal and decode them, the HT12D oscillation frequency should be 50 times the encoder oscillation frequency. In our case HT12E oscillation frequency is 3.25 KHz, hence our decoder oscillation frequency should be 162.5 KHz. Fixing the value of R1 as 62K does this job (refer the oscillation frequency vs supply voltage graph in the datasheet).

L293D: This IC serves as a Bidirectional motor driver. It is highly impossible to drive heavy loads such as motors using decoder IC hence we have used a dedicated L293D IC for this purpose.

MOTORS: Two simple DC motors are used for the movement of RC Car. A motor for moving the Car forward and backwards whereas the other one is used to steer the car left and right.

  b.) What are the differences between a remote control toy car and an actual electric vehicle?

 2. c. State its limitations

Limitation of EV:

• Electric cars have a shorter range than gas-powered cars.
• Recharging the battery takes time.
• They are usually more expensive than gas-powered cars.
• It can sometimes be difficult to find a charging station.
• There aren't as many model options.
• Over heating ECU's can cause fire.

 Limitation of RC Toy CAR:

 The batteries are often at the root of many RC problems. Not running at all, running very slowly, or even stopping suddenly can be battery-related.

• Missing batteries. Check the RC and transmitter.
• Installed wrong. Make sure the batteries are in the right slots, facing the right direction, and completely seated in their little slots.
• Old batteries. Try some fresh batteries. Or if you're using a cheap brand of batteries, try a different brand. If using a battery pack, make sure it is fully and properly charged.
• Corrosion. If the vehicle has been sitting for a long time or the battery compartment has been exposed to moisture and air there could be some corrosion. In addition to replacing the batteries, clean the battery contacts.

3. a.) Explain the cooling technologies of a power converter?  

Abstract:

Power electronics devices such as MOSFETs, GTOs, IGBTs, IGCTs etc. are now widely used to efficiently deliver electrical power in home electronics, industrial drives, telecommunication, transport, electric grid and numerous other applications. Lets discusses cooling technologies that have evolved in step to remove increasing levels of heat dissipation and manage junction temperatures to achieve goals for efficiency, cost, and reliability. Cooling technologies rely on heat spreading and convection. In applications that use natural or forced air cooling, water heat pipes provide efficient heat spreading for size and weight reduction. Previous concepts are reviewed and an improved heat sink concept with staggered fin density is described for more isothermal cooling. Where gravity can drive liquid flow, thermosiphons provide efficient heat transport to remote fin volumes that can be oriented for natural and/or forced air cooling. Liquid cold plates (LCPs) offer the means to cool high heat loads and heat fluxes including double sided cooling for the highest density packaging. LCPs can be used both in single phase cooling systems with aqueous or oil based coolants and in two-phase cooling systems with dielectric fluids and refrigerants. Previous concepts are reviewed and new concepts including an air-cooled heat sink, a thermosiphon heat sink, a vortex flow LCP and a shear flow direct contact cooling concept are described.

Introduction:

Power electronics devices and systems are vital in the efficient generation, transmission and distribution, conversion, and a huge variety of end uses of electric power. More and more applications are adopting power electronics technologies to improve energy efficiency, reliability, and control and it is anticipated that in the future all electrical power will flow through a power semiconductor device at least once.

Silicon remains the workhorse material for power semiconductors and to avoid device failure due to thermal runaway, effective cooling is critical. Figure below shows the maximum safe junction temperatures for silicon devices. The maximum safe junction temperatures in SiC could exceed 300 oC .  So that even in high ambient temperatures, sufficient cooling may be provided by smaller and lower cost heat sinks resulting in improved volumetric power density.

All electronic devices generate heat due to their unavoidable internal losses and inefficiencies. The higher the efficiency rating of the device, the less internal heat is generated within it. If we could achieve 100% efficiency, and technology is getting ever closer to that elusive goal, no heat would be generated within the device and, therefore, no cooling would be required. Until then, the generated heat must be dissipated to maximize the end product's reliability and prevent its premature failure.

Figure above shows Photo of a convection-cooled (no fan) AC-DC switchmode power supply. Notice the black heat sinks, some of which have fins, while others are vertically mounted black sheet metal, which also serve as heat sinks to the high power devices used in the power supply circuits. This model (ZWD-PAF) is capable of providing up to 225-watts of output power with an ambient air temperature ranging from -10°C to +50°C.

There are three primary techniques of transferring or dissipating heat from power and other electronic devices: These are conduction, convection, and radiant. In all cases, the heat is being transferred from the hot electronic device to another medium that is at a lower temperature. All cooling techniques include some components of conduction, convection, and radiant cooling in their process. Heat is constantly seeking to move from its source to an object, or through a medium that is cooler.         

Conduction cooling: This is defined as the transfer of heat from one hot part to another cooler part by direct contact. For example, many DC-DC converters have a flat surface (baseplate) that is designed to mount directly to an external heatsink or cold plate that will conduct the heat away from the power device by direct contact, thereby cooling it. Most power supplies use internal heatsink that are in contact with the power devices, via a thermal conductive paste or pad, to conduct away the heat. The heatsink, in turn, depend upon convection cooling to transfer their conducted heat to the cooler surrounding air. 

Convection cooling: This involves the transfer of heat from a power device by the action of the natural air flow (air is actually a low-density fluid) surrounding and contacting the device. Many power devices are rated for natural convection cooling as long as the air surrounding the unit remains within a limited temperature range that is cooler than the device. The advantage of convection cooling is that no fans are required (Figure 1). Fans, due to their mechanical components, tend to reduce the mean-time-between-failures (MTBF) ratings of power supplies. Convection cooling works best when the products have a natural circulating source of air. For enclosed applications, to insure a natural exchange of air, a number of vents should be provided at various locations in the case or enclosure.

Figure above shows Photo of a convection-cooled (no fan) AC-DC switchmode power supply. Notice the black heat sinks, some of which have fins, while others are vertically mounted black sheet metal, which also serve as heat sinks to the high power devices used in the power supply circuits. This model (ZWD-PAF) is capable of providing up to 225-watts of output power with an ambient air temperature ranging from -10°C to +50°C.

Many heat- generating electronic devices, including DC-DC converters, microprocessors, hybrid circuits, etc., require the use of heatsink to facilitate convection cooling and to assist in transferring the heat away from the devices to the cooler air. Heatsink utilize both conduction cooling (i.e., the device must make good thermal contact with the heatsink) and convection cooling. Heatsinks are designed to transfer the heat from the device to the ambient air, primarily by substantially increasing the surface area that comes in contact with the air. Because the surfaces of electronic devices and heatsink are not perfect, some type of thermally conductive interface material is necessary to fill the tiny voids. Although this material must be thermally conductive, in some instances it also needs to be an electrical insulator; for example, a thin silicon pad. In most other applications, a thin layer of thermal grease can be used as the interface material.

Another type of convection cooling requires forced-air flow, via fans that direct the air across the power devices, with or without heatsink. Many power supplies come with a built-in fan to provide this forced-air type of convection cooling. Other types of power devices specify the amount of air flow that must pass through or around the unit, in cubic or linear feet-per-minute (CFM or LFM), in order for the device to provide its maximum rated output power.

Radiant cooling: This is the transfer of heat by means of electromagnetic radiation (energy waves) that flows from a hot object (e.g., power device) to a cooler object. True radiant heat transfer can take place in a vacuum and does not require air. For example, our Sun not only emits light waves, but, also infrared heat waves through great distances in space, which results in our Earth having daylight and various degrees of warmth.

Thermoelectric coolers: Pettier coolers, also known as thermoelectric coolers (TEC), are solid-state devices that function similar to heat pumps. TECs require an external DC power source to operate them. The power device to be cooled is mounted on one side of the TEC with an appropriate thermal grease or thermal pad. When a voltage is applied to the TEC, the heat is pulled away from the power device and pumped to the opposite side that has a heatsink and sometimes a fan to dissipate the heat. Thermoelectric coolers are effective, but have low efficiencies and are more expensive than heatsink and/or fans. However, they are used to cool laser diodes, ultra-fast microprocessors, and laboratory instruments.

Liquid cooled cold plates: In some applications where multiple electronic power devices are employed or where there is a concentrated heat load in a confined area, it may be necessary to consider the use of liquid cooled cold plates. As the name implies, these are usually comprised of aluminum or copper plates on which the power devices are mounted that contain parallel ducts or serpentine tubes through which a liquid coolant is pumped and then routed to an external heat exchanger where the liquid is cooled; then it is recycled through the pump in a continuous closed loop. This type of cooling is very effective, but also one of the most expensive techniques to employ. Nonetheless, liquid cooled cold plates are used extensively in many military, aerospace, RF amplifier, and medical, industrial, and telecom applications.

THERMAL RESISTANCES IMPEDE HEAT TRANSFERS

The majority of cooling techniques employed today are utilized in power, semiconductor, and microprocessor devices, many of which are being introduced on a daily basis. Each of these active devices is specified to operate safely within a certain temperature range that must not be exceeded. The maximum operating temperature is commonly based on maintaining the semiconductor's internal junction temperatures below its maximum. Since the internal junction temperatures cannot be measured directly, devices such as MOSFETs, DC-DC converters, power bricks, etc., are specified to operate safely as long as its metal case, mounting tab, or baseplate temperature is kept below the specified maximum. To achieve this, a network of thermal resistances must be considered and taken into account. Thermal resistances are analogous to electronic resistances. They represent the mechanical interfaces between the various layers of materials which impede the flow of heat from one level to the next. A diagram of these series-connected thermal resistances beginning with the internal semiconductor junctions and concluding with the ambient air is shown in Figure below.

 b.) Take an example of a system and assume the efficiency and other parameters.

When designing a circuit, thermal calculations are indispensable, and particularly where power devices that handle large amounts of power are concerned, they are extremely important, not only with respect to operation lifetime but from the standpoint of safety as well.

Thermal Resistance of Packages Capable of Back-Surface Heat Dissipation .

Above figure are presented terms and definitions relating to the thermal resistance of packages that can dissipate heat from the back surface, as is the case with packages onto which heat sinks can be installed such as the TO-220FM and the TO-247, and packages a back-surface pin of which can be mounted onto a circuit board, such as the TO-252 and the TO-263

  `T_A`: Ambient (atmosphere) temperature.

`T_J` : Junction temperature.

`T_C` : Package back-surface temperature.

`T_T` : Package Marking-surface temperature.

`R_thJA` : Thermal resistance between Junction and ambient (atmosphere).

`R_thJC` : Thermal resistance between Junction and Package back-surface.

The basic structure of a package able to dissipate heat from the back surface consists of a lead frame ("Frame" in the figure), die bonding between the chip and the lead frame, the MOSFET chip ("Chip"), and the resin package ("Mold"). The "junction" in the explanation of symbols refer to the PN junction, or put more simply, the chip.

The table below is an example of the absolute maximum ratings and thermal resistances appearing on a data sheet for an Nch MOSFET in a TO-247 package. Here T_J is stipulated as an absolute maximum rating, and so in thermal calculations the need to keep the absolute maximum rating T_J (also sometimes written T_JMAX) from being exceeded is the ultimate requirement. Here R_thJA, which is of fundamental importance for the thermal resistance, and R_thJC, which is necessary for thermal calculations when the back surface of the package is mounted on the board or when a heat sink is mounted in close contact, are shown.

Relationship Between Thermal Resistance and Power Dissipation .

In the above absolute maximum ratings table, the power dissipation P_D is indicated. P_D and the thermal resistance R_thJC are related by the following equation.

`P_D= frac {T_J-T_C}{R_thJC}`

`= (150-25)/0.26 ≅480W`

For T_J the absolute maximum rating of 150°C is substituted, and for T_C, 25°C, which is the condition for the power dissipation, is used. For R_thJC, the maximum value of 0.26°C/W is substituted. As indicated in note *4, R_thJC is the value at T_C=25°C, and so matches the condition stipulated for P_D. The calculation result is approximately 480 W,　and is the value of P_D appearing in the table.

To make this equation a little easier to understand, the difference between T_C and the absolute maximum rating T_J is the allowed value for heat generated by the chip itself. For the above conditions this is 150 – 25 = 125°C, and the chip is allowed to generate heat up to 125°C. Heat generation is the thermal resistance times the power consumption, and so by dividing the allowable heat generation by the thermal resistance, the power consumption that can be allowed, that is, the power dissipation is obtained.

Here there is a matter that must be understood. The provided value of R_thJC has the condition that T_C = 25°C. Put another way, if T_C is not 25°C, R_thJC is not 0.26°C/W. When considering actual conditions of use, the condition that T_C = 25°C is hardly ever satisfied, and so the R_thJC presented on the data sheet as an example cannot be used in thermal calculations for the actual conditions of use. It may seem contradictory for a data sheet to present data that cannot actually be used, but in many cases, specific conditions are required in order to stipulate specification values, and it is understood that specified values on a data sheet, whether for thermal resistance or anything else, are values stated for particular conditions.

Hence in order to use R_thJC to determine the power dissipation under actual conditions of use, the R_thJC under those conditions must be measured and ascertained. However, this in turn requires corresponding measurement instruments, environment preparations, and the like.

As one other method, R_thJA can be used. The relationship between T_J and R_thJA can be expressed by the following equation.

`T_J= (R_thJA×P)+T_A`

P is the power loss (power consumption). The quantity within parentheses corresponds to the heat generated by the chip itself; this plus the ambient temperature T_A is equal to T_J. It should be considered that by using R_thJA, T_A can be substituted for T_C in the previous equation. Through this calculation, the power loss P or T_A is adjusted such that T_J does not exceed the absolute maximum rating of 150°C.

In order to use this equation in calculations, the ambient temperature, the power loss of the MOSFET, and R_thJA must be determined. The ambient temperature and the power loss can be ascertained relatively easily. The R_thJA for this kind of package varies depending on the soldered pad area of the package back surface for mounting on the board, the thickness of the copper foil, and the material and number of layers of the board. In some cases it is possible to obtain the R_thJA for mounting under standard board conditions. 
Conclusion:

• In thermal calculations, various thermal resistances and the power loss are used to determine T_J, to confirm that ultimately the absolute maximum rating for T_J is not exceeded.
• The power dissipation is obtained by calculating backward to the power loss for which T_J does not exceed the absolute maximum rating.
• Specified values, absolute maximum ratings and the like must be considered paying attention to the stipulated conditions.

4. Refer to the below link and summarize the content in a tabular format.

https://teslamotorsclub.com/tmc/threads/tesla-thermal-management-system-explanation.88055/

#1. Main coolant radiator. Does not have a fan. When vehicle is moving, air passes through cooling the Coolant. Coolant enters from the right. This radiator can be bypassed with device #10. This is important, as when temperatures are very cold, bypassing the radiator will keep your battery and motor from cooling further rather than warming up.

#2. Coolant circulation mode selector. A device that switches between two modes: Series and Parallel. If in series, coolant passes from #1 to #3 and then from B (Battery) to #7. If is parallel, one loop passes from #1 to #7 and other loop from B to #3.

#3. 12V coolant pump. The % indicate pumps running speed. Slower pump speed uses less energy, prolongs pump life and slows the coolant flow. Full speeds are generally used during supercharging to maintain cooling to batteries.

#4. Adjustable coolant redirection valve. Sends 100% of coolant from #3 to #5, 100% form #3 to #13 or anything in between. Thermal management will determine this based on ambient temperature, battery, cabin, and motor temp.

#5. Coolant heater. Model S and X are rated for 6kW though can also run at 3kW. The coolant heater is a resistive heater that Runs on high voltage. Once activated, coolant will be warmed, and begin to flow through the battery pack. This is used to heat the Battery fast. Heat generated by #6 #8 #9# can also be used to for pack heating, through the use of those devices waste heat. Though that heating is minimal due to efficiencies of those units. A Cold pack will also cool down those devices. The Coolant Heater can be “disabled” by activating the Model S or X Range Mode. The Coolant Heater will bring the battery up to 32*F/0*C or slightly higher to allow Regen and Charging. If Range Mode is turned ON, the coolant heater will not come on until approximately -10*F or -23*C, saving a considerable amount of power if charging or regen is not needed (In general, you will save more power than regen will gain you back by disabling pack heater). Plugging in the car and starting charging will over-ride range mode setting and heat pack to needed temperature. It will draw power from Grid to heat pack while plugged in.

#6. DC-DC converter. Takes energy from high voltage pack, and steps it down to 12v (High Voltage pack is 350-404v fully charged depending on battery & trim level). This keeps the 12V battery charged and all 12V devices powered up. A more limited amount of coolant is directed through this unit, as heat generation is low, and power output of this device is variable (So only what is needed is converted, thus reducing heat build up).

#7. 12V coolant pump. This pump is required to keep the second loop of coolant flowing if #2 is in parallel mode. Acts as a backup to #3. In series mode both pumps run at equal speed. In Parallel, pumps run at speed that's required for that specific system's cooling. Some early Model S's are said to have 3 pumps, my original MS60 I was told did when it had pump 2 replaced in 2013 by mobile service. Design changes are constant. Car will still be operational with a single pump failed, however, the car may run at reduced performance and services, and supercharging may be unavailable due to excessive heat build up.

 Reference:

1. https://maker.pro/pcb/projects/remote-control-car
2. https://science.howstuffworks.com/rc-toy.htm
3. https://www.geeksforgeeks.org/what-are-the-differences-between-c-and-embedded-c/
4. https://www.liveabout.com/troubleshooting-an-rc-that-wont-run-2863133
5. https://www.powerelectronics.com/technologies/lighting-systems/article/21861451/proper-cooling-of-led-arrays-and-power-converters-improves-performance-reliability
6. https://teslamotorsclub.com/tmc/threads/tesla-thermal-management-system-explanation.88055/
7. https://www.electronics-cooling.com/2017/07/advanced-cooling-power-electronics/
8. https://www.alfalaval.com/globalassets/documents/microsites/heating-and-cooling-hub/application-sheets-manufacturing-industry/power-converter-cooling.pdf