Analysis of Tesla's automotive thermal management system and electric drive system

01

First Generation Roadster

Before reviewing the design of Tesla's inverter, we have to mention AC Propulsion, a company that has left a strong mark in the history of electric vehicle development. Founder Al Cocconi participated in the development of the first mass-produced electric car, the GM EV1, but after GM "killed" the EV1, Cocconi left for Southern California to found AC Propulsion, designing and building a small number of prototype electric cars, the T-Zero. The car is only for one person, with only a small door, making it difficult to get in and out. The power part is powered by lead-acid batteries in series, and the upper and lower bridges of each bridge arm in the inverter are connected in parallel by 4-6 IGBT single tubes, with a total of 24 or 36 IGBT single tubes.

After Tesla was founded, it obtained the technical license for the powertrain system from AC Propulsion, including the IGBT single-tube parallel technology used in the first-generation Roadster inverter. Tesla did not stop paying patent fees to AC Propulsion until it produced about 500 powertrain systems and changed the system control from analog to digital.

However, since then, multi-tube parallel connection has become the core feature of Tesla's inverter design. In addition to path dependence, there are also supply chain considerations. In the first decade of this century, there were very few mass-produced automotive-grade IGBT module products launched on the market, only Infineon HybridPACK1, etc., but they could not meet Tesla's power output requirements. Although industrial modules have high-current versions, they are not designed for automobiles after all. Reliability, traceability and size cannot meet Tesla's requirements. At that time, no manufacturer was willing to customize expensive automotive-grade power module products for Tesla.

It was the right time and the right situation. Although the current specification of IGBT single tube was still small at that time, there were many suppliers, especially the headquarters of International Recorder (IR), one of the main IGBT manufacturers, was also located in California, which made it convenient for Tesla to communicate with it and select or even customize the appropriate IGBT single tube. The specific process in this regard can be referred to the memoir of Zhihong, one of the leaders of this project at IR at that time, "A Past Story of Winning the Tesla Model S IGBT Contract". The powertrain part of Roadster is called PEM (Power Electronics Module), which occupies the front half of the trunk, located behind the battery pack and above the motor. PEM began mass production in 2008. Before version 1.5, in addition to the "Tesla Motors" logo, there was also a "PEM 185" logo, which means the output power is 185kW. For versions 2.0 and 2.5, only the logo is left, or the "Tesla Motors" logo is changed to "Roadster Sport". From the PEM disassembly below, we can see that the overall layout of each version is roughly the same, with half of the space being for high-voltage connectors, high-voltage relays and fuses, and the other half being the inverter part, with three half-bridge arms placed horizontally. However, further disassembly shows that there have been at least two versions of the inverter design.

Tesla's first-generation Roadster powertrain PEM disassembly diagram. Judging from the IGBT single-tube package used on the power board, there are at least two versions (Source: Gruber Motors, Tesla Owners US in English)

The earlier PEM 185 used IGBTs in standard TO247 packages, and each switch consisted of 14 IGBTs in parallel, which was a significant increase compared to the original AC Propulsion solution. The inverter used a total of 84 IGBTs, including at least the Infineon 75A IGBT IKW75N60T.

In the later version, Tesla used the 600V 120A AUIRGPS4067D1 customized by IR, which also uses 14 pieces in parallel. This IGBT uses the TO-247 Plus package (also known as TP-247, Super-247), which eliminates the screw holes used for fixing in the TO247 package, so a larger bare chip can be installed to increase the output current.

However, these two IGBTs use the same installation method, both of which are 90 degrees after the IGBT pins are bent (Trim and Form) and attached to the power PCB board, and the conductive collector on the back is attached to the heat sink through the insulating thermal paste coating, and then the entire IGBT power board is fixed to the heat sink with screws. The main failure mode of this installation method is IGBT short circuit caused by cracking of the insulating thermal conductive layer after long-term use, and damage to the electrolytic capacitor.

02

Model S/X

The Model S, which went into production in 2012, made significant improvements to the powertrain, and the inverter design also completely abandoned the flat-laying method of the previous generation and changed to a three-dimensional structure. The Model X, which went into production in 2015, also used the same design, so it can be called the second-generation powertrain.

The second generation Tesla powertrain consists of two types: Large Drive Unit (LDU) and Small Drive Unit (SDU). The former is mainly used for the Model S/X single motor version and the rear wheel drive in the dual motor high performance four wheel drive version. The latter is mainly used for the front and rear drive of the dual motor ordinary version and the front drive of the dual motor high performance version.

Powertrain differences between Model S/X, Model 3/Y, and Model S/X Plaid (Source: Tesla)

As the name implies, LDU is larger in size, cylindrical in shape, and has a higher output power, while SDU is the opposite. Although the two powertrains appear in the same model, LDU was developed earlier than SDU and was withdrawn from the market earlier, mainly due to cost and power density considerations.

Comparison between LDU and SDU (Source: StealthEV)

03

LDU

The inverter in the LDU is a prism structure, with each phase or half-bridge occupying one face of the prism. The top and bottom of the prism are the high-voltage DC input and high-voltage AC output parts, respectively. There are three small triangular PCBs on the DC input side, which are the driver PCBs for each phase.

LDU uses the same IKW75N60T in TO247 package as PEM, but the amount is larger. Each switch is 16 IGBTs in parallel, with a total of 96 IGBTs. Although the IGBTs in LDU still need to be bent, the connection method with the busbar copper bar and the power PCB board is greatly optimized, and the area of ​​the power PCB board used is greatly reduced. Because of this, half of the IGBTs in each half-bridge part (the two middle rows) can be fixed with the busbar copper bar, while the other half (the two outer rows) need to be fixed with a set of two clamps.

Regarding the design of the inverter in the LDU, I still have a few questions to sort out. First, why does Tesla continue to use the IKW75N60T with a lower current instead of the newer AUIRGPS4067D1 with a higher current? Second, the LDU has two versions, the green PCB and the red PCB. Is there a difference between the two?

(Top) LDU with the inverter casing just removed (Middle) Detail of the inverter, taken from the DC and AC sides (Bottom) Detail of the half-bridge section, showing 8 IGBTs per row, and 8 x 2 rows of IGBTs hidden under the busbars and the long power PCB (Source: Damien Maguire, Turbo Electric)

04

SDU

SDU also uses a three-dimensional structure in the inverter, but the design method is very different from PEM and LDU, making the overall structure more compact and the power density reaching 30kW/L and 33.3kW/kg respectively.

Photo and exploded view of the Model S/X SDU (Source: Babak Fahimi UT Dallas)

First, the IGBT single tube uses AUIRGPS4067D1, 6 pieces are connected in parallel, and the total usage is 36 pieces. Although the cost of a single IGBT increases, the total cost is lower because the usage is reduced. However, according to the communication with Tesla engineers, the number of parallel IGBTs is small, the requirements for chip consistency are higher, and the actual design difficulty increases. Therefore, Tesla has added special specification binning requirements for IGBT single tubes, which has brought considerable challenges to the back-end processes of IGBT manufacturing and supply chain management.

Secondly, the layout and heat dissipation of IGBT single tubes have undergone major changes. The IGBT single tubes in the upper and lower bridge arms of each half-bridge are fixed back to back on the radiator through low-temperature welding, and are further strengthened with clamps to form a sandwich-like structure. Compared with LDU, not only the half-bridges form a three-dimensional structure, but the upper and lower bridge arms within the half-bridge are also three-dimensional structures, which fully utilizes the space, and low-temperature welding makes heat dissipation better. Now some semiconductor suppliers' double-sided water-cooled heat dissipation modules also use similar heat dissipation designs to improve power density. Thirdly, the connection of IGBT single tubes is also very different from the past. SDU no longer requires the power board to connect the IGBT single tube, but uses an inverted plug-in method to connect to the driver board. Therefore, it is no longer necessary to bend the IGBT single tube pins, which reduces the installation cost and avoids various troubles that may be caused by this (after bending the pins, the IGBT may fail sporadically, and it is difficult to determine the cause, which often leads to mutual accusations between the IGBT supplier and the system manufacturer). Then, by appropriately adjusting the length of the three pins G/D/S of the single tube, it is properly connected to the driver board and the busbar copper bar. Therefore, the pin design and manufacturing of IGBT have also become important.