This article will start from the core power devices used in Tesla's historical model powertrains, and briefly describe the evolution history of the inverter part.
01
The first generation of Roadster
Before reviewing the design of Tesla's inverter, it is necessary to mention AC Propulsion, a company that has left a strong mark in the history of electric vehicle development.
Founder Al Cocconi was involved in the development of the first mass-produced electric vehicle, the Universal EV1, but after GM "killed" ev1, Cocconi left Nanga to create AC Propulsion, designing and building a prototype electric vehicle T-Zero.
The car is for one person only, and the door is only a small one, making it difficult to get in and out. The power part is powered by a lead-acid battery and connected in series, and the upper and lower bridges of each bridge arm in the inverter are paralleled by 20 IGBT single tubes, using a total of 120 IGBT single tubes, with a total area of 7200 square millimeters of bare discs. If you take into account the yield of the pre-process and post-process, a T-Zero needs to use about the IGBT that can be manufactured by a whole 6-inch wafer at that time.
The author took a test ride on the T-Zero at AC Propulsion, which can only be taken by one person, and it is inconvenient to get in and out.
Source: 01 Core Smell
After Tesla was founded, it obtained a technical license for powertrain systems from AC Propulsion, including the IGBT single-tube parallel technology used in the first-generation Roadster inverter. It wasn't until Tesla produced about 500 powertrain systems and changed the system controls from analog to digital that it stopped paying the AC Propulsion for patents.
But since then, multi-tube paralleling has become a core feature of Tesla's inverter design. In addition to the reasons for path dependence, there are also supply chain considerations. In the first decade of this century, there were very few mass-produced vehicle-grade IGBT module products launched on the market, only Infineon HybridPACK1 and so on, but they could not meet Tesla's power output requirements.
Although the industrial module has a high-current version, but after all, it is not designed for the car, reliability, traceability and dimensions can not meet Tesla's requirements, and no manufacturers were willing to customize expensive vehicle-grade power module products for Tesla at that time.
At that time, although the current specification of IGBT single tube was still small, there were many suppliers, especially the headquarters of International Rectifier (IR), one of the main manufacturers of IGBT, was also located in California, which was convenient for Tesla to communicate with it and select or even customize the appropriate IGBT single tube.
The Roadster's powertrain section, called the PEM (Power Electronics Module), occupies the front half of the trunk, behind the battery pack, above the motor. PEM began mass production in 2008, and before version 1.5, in addition to the logo of "Tesla Motors", there was also the logo of "PEM 185", which means that the output power is 185kW.
Versions 2.0 and 2.5 left only the logo, or changed the logo of "Tesla Motors" to "Roadster Sport". From the disassembly of the PEM below, it can be seen that the overall internal layout of each version is roughly the same, half of the space is high-voltage connectors, high-voltage relays and fuses, etc., the other half is the inverter part, and the three and a half bridge arms are placed horizontally. But further disassembly shows that the inverter design has been available in at least two versions.
The earlier PEM 185 featured an IGBT monotube package in a standard TO247 package, each switch consisting of 14 IGBT single tubes in parallel, a total of 84 IGBT single tubes were used, and the models used included at least Infineon's 75A IGBT IKW75N60T.
In later versions, Tesla switched to IR's custom 600V 120A AUIRGPS4067D1, also with 14 pieces in parallel. The IGBT is available in a TO-247 Plus package (also known as TP-247, Super-247), eliminating the screw vias used for fixing in the TO247 package, so that larger size dies can be loaded and the output current can be increased.
However, these two IGBTs use the same installation method, both IGBT bending pins (Trim and Form) after 90 degrees attached to the power PCB board, the back of the conductive collector (Collector) through the insulating thermal paste coating on the heat sink, and then with screws to fix the entire IGBT power plate on the heat sink. The main failure modes of this installation method are IGBT short circuits caused by cracking of the insulating thermal conductive layer after long-term use, and damage to electrolytic capacitors.
02
Model S/X
The 2012 mass-produced Model S made significant improvements to the powertrain, and the inverter design completely abandoned the tiled method of the previous generation and replaced it with a three-dimensional structure. The Model X, which was mass-produced in 2015, also uses the same design, so it can be called a second-generation powertrain.
The total components of the second-generation Tesla power are the Lage Drive Unit (LDU) and the Small Drive Unit (SDU). The former is mainly used in the model S/X single-motor version, and rear-wheel drive in the dual-motor high-performance four-wheel drive version. The latter is mainly used for the front and rear drives of the ordinary version of the dual motor, and the front drive in the high-performance version of the dual motor.
Powertrain differences for Model S/X, Model 3/Y, and Model S/X Plaid
Source: Tesla
As the name suggests, the LDU is larger, cylindrical, and the output power is also larger, while the SDU is the opposite. Although the two powertrains appeared in the same model, the LDU was developed earlier than the SDU and exited the market earlier, mainly due to cost and power density considerations.
Comparison of LDUs and SDUs
Source：StealthEV
The inverter in the LDU has a triangular structure, and each phase, or half-bridge part, occupies one face of the triangular prism. The top and bottom ends of the prism are the HVDC input section and the HV AC output section, respectively. There are also three small triangular PCBs on the DC input side, which are the driving PCB boards for each phase.
The LDU uses the same TO247 package of IKW75N60T as PEM, but with a large amount of use, each switch is 16 IGBT single tubes in parallel, sharing 84 IGBTs. Although IGBTs in LDU still need bending pins, their connection with the busbar copper strip and power PCB board is greatly optimized, and the power PCB board area used is reduced a lot. Because of this, half of the IGBTs (the middle two rows) in each half-bridge section can be fixed with busbar copper rows, while the other half (the outer two rows) needs to be fixed with two sets of fixtures.
Regarding the design of the inverter in the LDU, the author still has several problems to clarify. One is why Tesla continues to use the IKW75N60T with a smaller current instead of the newer, more current AUIRGPS4067D1? Second, there are two versions of LDU, green PCB and red PCB, is there a difference between the two?
Source：Damien Maguire，Turbo Electric
(Top) The LDU that has just been disassembled from the inverter housing
(Middle) Inverter detail diagram, taken from the DC side and the AC side, respectively
(Bottom) Detailed view of the half-bridge section shows 8 IGBT single tubes per row, with another 8 x 2 row IGBT single tubes hidden under the busbar copper row and the long strip of power PCB board
SDU also adopts 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 reaches 30kW/L and 33.3kW/kg respectively.
First of all, the IGBT single tube selected AUIRGPS4067D1, 6 pieces in parallel, the total dosage of 36 pieces. Although the cost of monolithic IGBT increased, the total cost was lower due to the decrease in dosage. 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 is increased. Therefore, Tesla has added special binning requirements to the IGBT single tube, which has brought no small challenge to the back-end process of IGBT manufacturing and supply chain management.
Secondly, the layout and heat dissipation of IGBT single tubes have changed significantly. Through a double-ended clamp, the IGBT single tube in each half-bridge upper and lower axle arms is fixed back-to-back to the radiator, forming a sandwich-like structure. Compared with the LDU, not only the three-dimensional structure is formed between the half-bridges, but also the upper and lower bridge arms within the half-bridge are also three-dimensional structures, making full use of the space. Now some semiconductor suppliers' double-sided water-cooled cooling modules also use similar heat dissipation designs to improve power density.
Third, the connection of IGBT single tubes is also very different from the past. The SDU does not need the power board to be connected to the IGBT single tube, but uses the inverted way to connect to the drive board. Therefore, it is no longer necessary to bend the IGBT single-tube pin, which reduces the installation cost and avoids various troubles that may be caused by it (after bending the IGBT, the IGBT may fail sporadically, it is difficult to judge the cause, which often leads to mutual blame between the IGBT supplier and the system manufacturer). The length of the three pins of the single tube G/D/S is adjusted appropriately so that it is properly connected to the drive plate and the busbar copper row. Therefore, the design and manufacture of IGBT's pins has also become important.
The advent of SDUs has led Tesla to have stricter mechanical, electrical, and manufacturability requirements for IGBT devices. The author is also fortunate to participate in the customization of IGBT single tubes as a supplier in cooperation with a number of Tesla core R&D personnel, and thus responsible for the development of the next generation of Tesla custom IGBT devices. Since then, Tesla has begun to work more closely with the head manufacturers of power semiconductors, deeply involved in the definition and design of core power devices, and finally launched an epoch-making third-generation powertrain.
03
Model 3/Y
The Model 3/Y powertrain is more compact than the previous generation, especially the inverter part. One of the reasons is that compared with other companies' three-in-one electric drive systems, Tesla inverters have chosen to remove the cover plate from the previous generation and stick closely to the reducer, thus reducing the weight and volume of the inverter. But more importantly, a completely new power device was chosen in the new generation of inverters, which changed the overall design of the inverter.
While Tesla is still optimizing the design of the SDU, the core developers are already thinking about how the next generation of powertrains should be realized. In particular, the TO247 and TO247 Plus packages used in the IGBT single tube, the core devices in the previous two generations of systems and three designs, have little potential to further increase the current specification and improve performance. At the same time, although IGBT technology continues to advance, it has brought about quantitative changes rather than qualitative changes. In summary, the IGBT single tube is about to reach the performance bottleneck. In view of this, Tesla not only discusses the choice of new power chips with power semiconductor manufacturers, but also cooperates with some advanced packaging technology companies to develop new packages. The result was the TPAK (Tesla Pack) module, which revolutionized the following points.
First, Tesla pioneered the use of silicon carbide chips in mass-produced electric vehicles instead of IGBT chips. Although the module cost of TPAK SiC is high, it is in line with the trend of industrial upgrading, and large-scale field use data of silicon carbide is obtained at least 3 years earlier than competitors.
Second, the TPAK package adopts a single switch module (Single Switch Module) design between the single tube and the conventional module, which not only exceeds the output current, output power, parasitic inductance and other limitations brought by the previous single tube package, but also retains the flexibility of multi-tube parallel connection, and can choose how many TPAK modules are needed in parallel according to different inverter power output needs. And Tesla's multi-tube parallel experience accumulated over the past 10 years can continue to be used.
Third, the TPAK module uses sintering (sintering) as a connection method inside and outside. Inside the module, the chip is connected to the DBC via a silver sintering layer instead of a soldering layer. On the outside of the module, TPAK's baseplate is also sintered to the heat sink in place of the thermal paste coating. The combined action of the two not only makes the heat dissipation ability of the system go up to a higher level, but also the reliability of TPAK itself, especially the number of power cycles, has also been greatly improved. In addition, the improvement in thermal performance means that the chip of the same size can output a larger current at a defined junction temperature, or use a smaller chip to reduce the cost of the chip at the same current.
Finally, TPAK has very few parasitic parameters, so it can be used as a general-purpose module to fit not only silicon carbide chips, but also IGBT chips and gallium nitride chips. This can make it easier for suppliers to share the back-end production line to produce different TPAK modules, so as to reduce costs and increase production. At the same time, the design of the inverter is only considered in a modular package form, reusing mechanical and thermal designs, and reducing costs at the inverter system level.
As a result, 4 such TPAK SiC modules are connected in parallel to form the upper or lower bridge of the bridge arm, and the drain and source of the module are connected with the busbar copper row by laser welding, and a total of 24 TPAK modules are used to form the first generation of Model 3/Y inverters.
04
Model S/X Plaid
In the middle of last year and the end of last year, the Model S Plaid and model X Plaid began to be delivered separately, so there is not much teardown analysis on the network at present. From the information that can be collected, Model S/X Plaid continues to use the inverter design in Model 3/Y, and even the drive and control of the former inverter is also marked with the word "Model 3" on the PCB board, the only visible change is that the high-pressure part of the Platinum inverter has added a pyrotechnic actuator, which can immediately cut off the connection with the motor when the TPAK module fails due to a short circuit.
At the system level, the model S/X Plaid differs significantly from the Model 3/Y in that the Model S/X Plaid rear drive is a dual motor, driven separately by two TPAK modules. In addition, the motor used in the Model S/X Plaid has been improved, especially with carbon fiber reinforcement in the rotor section.
05
Cybertruck and the second-generation Roadster
Both models are still in the in-house development phase, so information is extremely limited. From the information revealed by Elon Musk in Twitter, the second-generation Roadster uses a faster motor than the Model S/X Plaid.