From the gap between Volkswagen, Ford and Tesla, we can see the difficulties and breaking points of the intelligent electrical architecture

Remarks: Skateboard chassis / domain controller + real name, company, position

preface

If vehicle intelligence is the future, then intelligent electrical architecture must be its foundation.

In the previous article, we also said that Tesla's three distributed area controllers have basically implemented intelligent electrical architecture, but you figured out how Tesla implements intelligent electrical architecture, can you guide Volkswagen and Ford to achieve intelligent electrical architecture? The answer is clearly no. As Teacher Zhang Xiaoyu said, the knowledge of turning eggs into chickens can guide chickens, but how can it help ducks?

This article will start from the comparison of the electrical architecture of Tesla, Volkswagen and Ford 3 new SUVs, to see why Tesla can lead the industry for 6 years to achieve intelligent electrical architecture, and how the historical baggage of traditional car companies hinders the landing of new technologies, from the system design of the vehicle, to the cognition of the cost model and the technical model, and finally to what technical difficulties to face when landing, and how to achieve the landing of the intelligent electrical architecture.

This article is involved in passenger cars and commercial vehicle applications, the article will also popularize some basic vehicle electrical principles, traditional solution device costs, semiconductor solutions and costs, wiring harness basic knowledge, load basic characteristics, HSD chip basics, MOSFET basics, the relationship between current and cost, related electrical and electronic design difficulties, chip parameter selection, etc., is a brick and stone, hope to work with industry partners to promote the landing of intelligent electrical architecture.

body

Today's automotive industry is undergoing drastic electrification and intelligent changes, which will inevitably lead to an increase in the number of vehicle ECUs and an increase in the complexity of the electrical architecture. What are the considerations?

Just so, there is a foreign website 3IS based on Tesla Model Y, compared volkswagen ID.4 and Ford Mach E electronic and electrical architecture, these three cars are pure electric SUVs, the mass production time is similar; and Volkswagen and Ford as a traditional old large car company, and Tesla comparison is very representative.

From the ECU and network type node comparison table in the figure below, we can see that the number of ECUs of Model Y, ID.4, and Mach E is 26, 52, and 51, respectively, and Tesla's integration is significantly higher, mainly because Tesla integrates the functions of many small ECUs into the area controller. As we said earlier, for example, Tesla's Model 3 FBCM is responsible for power distribution, but also responsible for some left headlight control, air conditioning control, thermal management and other functions, spanning the traditional body, cockpit, chassis and power domain.

As we said before, traditional OEMs have a lot of historical baggage, according to past experience, based on the existing mature modules for multiplexing can significantly shorten the vehicle development cycle and reduce development costs, and ensure the reliability of the vehicle. However, if the steps are too big, and large-scale integration is carried out as soon as it comes up, it will involve the whole body.

Therefore, for traditional car companies, any change needs to be very cautious, because there are many constraints, it is very difficult to change, the cycle is very long, the scope is very wide, the risk is very large, and the cost is also very high.

Everyone should remember that when ID.3 was first listed, there were many software bugs, and tens of thousands of cars were parked in the Volkswagen factory waiting for upgrades. The public is still like this, and other homes can be imagined. So we can see that Ford and Volkswagen, despite claiming to be new all-electric architectures, still reuse a lot of small ECUs.

But one thing is undeniable is that Volkswagen has made a lot of updates in the process of upgrading the electrical and electronic architecture, if you are interested, you can look at the comparison between the Audi e-tron and the ID.4 electrical and electronic architecture, you can see the huge changes.

Comparison of the number of modules and network nodes of the three vehicles (Source: 3IS)

As can be seen from the number of Lin buses in the figure above, the number of ID.4 and Mach E is almost twice that of Tesla, which also shows that they reuse a lot of small ECUs based on Lin communication, which is a very typical traditional design method.

In the last two articles, we have analyzed the intelligent power distribution scheme and electrical architecture of Tesla Model 3 in detail, and interested partners can go to see "Why did Tesla "kill" fuses and relays? And "What kind of electrical architecture is required for autonomous commercial vehicles?" ”。 From the table below, we can just understand the impact of intelligent electrical architecture on vehicles from another side.

We can see that ID.4 and Mach E are still traditional designs in this regard, with 3 distribution boxes, 1 front compartment, 2 cockpits, and a large number of relays and fuses, while the Model Y follows the model 3 design, all using semiconductor solutions for substitution, and the number of traditional relays and fuses is 0.

Comparison of the number of power distribution modules, fuses and relays in three vehicles (Source: 3IS)

This design difference has led volkswagen and Ford to adopt the concept of domain control architecture, but the electrical and electronic architecture of the three is still quite different.

Tesla's architecture is closer to the zone control architecture, which can be confirmed by the progress from the model S's internal harness length of up to 3 km to the Model 3's only 1.5 km, because the regional architecture has obvious value for the savings of the harness. Aptiv has also calculated that using the regional architecture can reduce the cost of the wiring harness by 25%, while Visteon believes that the regional architecture can save 50% or more of the harness length.

Zone Controller Savings on Wiring Harnesses (Source: Visteon)

In addition, Visteon also specifically elaborated on the value of regional intelligent power distribution, including:

1. Dual power supply graded power supply;

2. Promote the electronicization of power distribution technology and cancel the traditional fuse;

3. Central distribution box virtualization, protection feature optimization;

4. Intelligent power management, fault prediction based on current and voltage diagnosis;

5. Other benefits of fuse and load optimization.

The value of regional smart power distribution (Source: Visteon)

In fact, we have carried out too many in-depth analysis of these things in the last two articles, you can not say that Volkswagen and Ford do not understand the regional architecture, or have not analyzed the value of regional intelligent power distribution, right? The Model 3 rolled off the production line in September 2017, and Volkswagen and Ford were mass-produced in 2020 and 2021, with three or four years in between, but Tesla is still the only OEM in the world to adopt regional intelligent power distribution, which is enough to explain the problem.

In addition, the conclusion given by 3IS is also interesting - 3IS said: "It is difficult to simply say whose architecture is the best, depending on the purpose and constraints." Traditional OEMs using legacy technology can reduce R&D costs, although this is not the best. Tesla has no choice but to start from scratch, so it can take a completely different path, and it doesn't have any constraints. ”

Architectural Comparison Conclusion (Source: 3IS)

In fact, Tesla is also constrained for old models. For the Model S launched in 2012 and the Model X launched in 2015, Tesla will not be able to use this technology in all new cars until 2022, starting from the first time the Model 3 uses the intelligent power distribution scheme, which has been separated by 5 years, so the difficulty of upgrading the old model can be seen, which is still on the basis of Tesla's mature regional intelligent distribution architecture.

Tesla will switch electronic fuses across the board in 2022 (Source: Teslatap)

In addition, through the analysis of Tesla's old model Model X (2015-2020) and Model S (2016-2020), we can see that even if the traditional distribution box solution, Tesla's design is different from the traditional OEM.

Tesla Model S/X front compartment distribution box (Source: Teslatap)

Tesla Model S/X cockpit distribution box (Source: Network)

As we can see from the above figure, the number of Plug-in relays used by Tesla vehicles is very small, only 5 (the traditional car is close to 20), and there are only fuses on the cockpit distribution box, no relays, which lays the groundwork for Tesla's innovative architecture of regional intelligent power distribution on the Model 3; compared with the ID.4 and Mach E, which were mass-produced 5 years later, respectively, 7 and 22.

So, why is the gap between Volkswagen, Ford and Tesla in the electrical architecture so large? Next, we will conduct a detailed analysis of the difficulties in the implementation of intelligent electrical architecture from the perspective of system, cost, cognition and technology.

1. Vehicle system angle

As we said in the previous chapter, many designs on the car are actually related to the whole body, because the car is a very complex integrated system, a car has tens of thousands of parts, and many systems are interrelated. For example, let's take Tesla's "high voltage does not power down" strategy, which involves a lot of specific design:

1. The high-voltage power battery will remain connected after the Tesla Model 3 is parked, and the high-voltage battery will discharge at a discharge rate of about 1% per day;

2. The "static" working current of Model 3 is 2.6A, while the static current of the traditional high-voltage vehicle is about 15mA~20mA to ensure that the battery does not lose power and can start normally next time (low-voltage no power cannot be on the high voltage, because BMS and VCU are used in 12V battery electricity);

3. Such a large current will cause the lead-acid battery to quickly run out, it is estimated to be a day, so Tesla has pioneered the design of BMS integration of small DC-DC, providing 12V power for the whole vehicle after parking to prevent the battery from losing power;

4. This design was originally designed to support all Online services, such as Sentinel mode;

5. This design in turn pushed Tesla to directly cancel the high-voltage pre-charging circuit, which is estimated to be the world's first;

6. Tesla uses the unheard of low-voltage battery DC-DC inverter for high-voltage precharging, this design does not support frequent high-voltage power-up, so the corresponding design is Tesla's high-voltage after the first power-on, generally not powered down.

Speaking of more around, I made a brain map, let's take a look:

Tesla high voltage does not power down design (source: Zuo Chenggang)

It can be seen from here that the implementation of a function requires the cooperation of the entire system design. Here is only the high-voltage part, it involves the pre-charging scheme, BMS, DC-DC and other modules and strategies of the new design; in fact, low-voltage power distribution and control Tesla also has the corresponding design to support.

Therefore, if you do not stand in the perspective of the vehicle system to conduct in-depth analysis of the implementation of a function, the steps are too large, and when you come up, you will engage in a lot of new functions, do a lot of integrated design, it will lead to a whole body, changing A will lead to problems with B, and only when B is moved will C not work. There are many constraints, it is very difficult to change, the cycle is very long, the scope is very wide, the risk will be very large, and the cost is also very high.

Tens of thousands of ID.3s parked in the Volkswagen factory waiting for the upgrade are still vivid, and even Tesla has to spend years upgrading old models, so traditional OEMs have to be cautious, and the design has become the "best choice" that has to be used.

Cost perspective

From the perspective of traditional vehicle design, cost is the first, intelligent electrical architecture everyone knows very well, but THE OEM looks at the cost, the project is definitely yellow, and even the impulse to discuss with you down is gone.

The author has conducted cost analysis for the electrical architecture of commercial vehicles and several major OEMs, the electric parts of the vehicle (excluding wiring harnesses), after upgrading to the intelligent electrical architecture, the cost has at least doubled, even for commercial vehicles that are not so sensitive to cost, this is absolutely unacceptable, not to mention the cost-sensitive passenger cars.

The following picture we shared in the previous article, I posted it again to you to see, for 12V systems, the semiconductor solution is more expensive than the current 30A relay, not to mention the largest proportion of fuses in the distribution box, the vehicle all uses semiconductor solutions, the cost increase is very large, especially the high current loop.

Below we roughly list the cost comparison of various solutions, you can feel it.

HSD/Relay Current - Cost and Replacement Speed (Source: Infineon)

12V system solution cost comparison (Source: Zuo Chenggang)

Comparison of the number of vehicle fuses and relays (Source: Zuo Chenggang)

For example, for a 10A circuit, the use of fuses is a dime, and the chip is seven or eight dollars. However, the cost of the chip is not linear with the increase of the current level, such as a 30A fuse or a dime, and the chip will have to be more than twenty dollars. No matter how large the current is, there is no HSD, only the MOS scheme can be used. The whole vehicle so many fuses, especially for the first-class distribution box, the large current is particularly much, the cost comparison of the smart distribution box simply can't bear to look directly, if you are Tier 1 are embarrassed to raise the cost with the OEM.

So what to do?

Intelligent electrical architecture as a subversive technological innovation, in its brings innovative value at the same time, but also brought a significant increase in costs, technology to land, to apply, you need someone to pay for the cost of innovation, but the OEM does not want to directly increase the cost significantly, which will directly affect the profit, for the traditional OEM who makes money by selling cars, this is unacceptable.

In general, if the cost of technology increases between 20% and 30%, this technology is relatively easy to land, and for intelligent electrical architecture, it is necessary to consider the cost reduction of wiring harnesses and the cost reduction of research and development from the system level, but the cost of the entire system is difficult to estimate, you tell the OEM is reduced, even if the OEM is approved, the specific is not easy to calculate, which leads to the cost of this level is very sad, OEM leaders are difficult to say on this technology.

The system cost of semiconductor solutions is reduced (Source: Philippe Dupuy)

Philippe Dupuy of NXP believes that OEMs understand the value of semiconductor solutions, are also the main drivers of vehicle electrification, and have calculated that the cost savings at the system and vehicle levels are calculated. But he also acknowledges that to date, semiconductor solutions have not made significant progress.

Teacher He Fan said that you can't solve all the problems at the same time and on the same occasion, you have to change the constraints and change your thinking to solve them. This is what we often say, you can't treat your head with a headache and a foot pain, but you say that the best way to cure a headache is probably to soak your feet with a basin of hot water. For intelligent electrical architecture, the problem brought about by the increase in cost is the same, and we need to find a breaking point.

The innovative model of "hardware embedding + software payment" is a breakthrough point for the landing of intelligent electrical architecture. As we analyzed earlier, Tesla led the innovative model of "hardware embedding + software payment", taking hardware cost as part of value embedding, and the cost of hardware can be recovered through the software payment model in the later stage.

Traditional vehicle design, the requirements are clearly defined at the beginning, the whole vehicle is not updated during the life cycle, so it can use highly customized low-cost hardware, enough to use, do not need to be powerful. However, if the hardware is not strong enough, the software and hardware are difficult to decouple, the software and hardware cannot be separated, it cannot cope with the changes in new needs, the hardware cannot be redefined by software, and it is difficult to achieve "hardware embedding + software payment".

Without the software payment model, OEMs cannot have this kind of hardware embedded innovation by selling cars, and even have no ideas. Because although innovation generates new value, but this value is not currently reflected in the consumer side, or not directly reflected, then no one is willing to pay for the cost of innovation, and the pace of innovation will be slowed down.

In the previous article, we also said that the construction of intelligent electrical architecture can echo the inherent needs of the current OEM software capacity building, and at the same time, it can also increase the OEM's brand premium, extend the value chain, and also provide the possibility for the transformation of OEMs from vehicle manufacturers to service providers.

Here I think of another example given by He Fan, such as the debt problem of African countries, the thinking of Western countries is to reduce interest rates, or reduce some of them. And Chinese might say, let's discuss the problem of road construction. We have known since childhood that if you want to be rich, you must first build roads. Roads are built, the economy grows, and the debt problem is solved. Westerners are thinking about how to solve the problem directly, while Chinese considering a roundabout solution, because it cannot be solved directly!

Therefore, based on the current stage, for the cost of intelligent electrical architecture, our only way may be only roundabouts, and in the future, intelligent electrical architecture will be part of the regional architecture, is to support the future of new energy trucks and high-end unmanned driving technology infrastructure.

Iii. Cognitive perspective

In addition to cost, another obstacle to the development of new technologies or new things should be people's cognition. Just like when the mobile phone was originally integrated with a resolution of only 300,000 pixels, no one thought that this thing had any real value, whether to use the camera or the camera when taking pictures. Later, everyone knows that the value of mobile phone photography has slowly been recognized, consumers are willing to pay for a lens of 100 million pixels, and even manufacturers can use this as a selling point for publicity.

01

Cost pattern awareness

For OEMs who are now selling cars as their old bank, the price of vehicles must be differentiated according to configuration, because the cost of materials is different. Consumers want more functions, when buying a car, they have to put money in place, if there is no money when buying a car, this configuration is not bought, and then want to install, basically can not be added.

For example, the power chassis part, you bought 1.5T, after buying the power is not OK, you and the 4S store said that they want to change 2.0T, which is outrageous, you can not change an engine, ah, but the electric car can be on the basis of hardware margin, through the software to achieve part of the power improvement.

In addition, the cognition of production and cost, when Tesla initially studied electric vehicles, the battery cost was about $600 per kilowatt-hour, and Musk thought that the future could be reduced to $80 after first-principle analysis, and later trends were also seen, battery costs with the rise of electric vehicle production, it is indeed decreasing, it is estimated that by 2029, the price of lithium-ion batteries can drop to about $60 per kilowatt-hour.

At this point, we have to mention a law that predicts price movements very accurately in the automotive field - Lycian law, that is, for every cumulative increase in production, the cost price will fall by 15%, and it will continue to decrease, and you will double and drop by 15%. The automotive industry has followed this pattern since 1900.

Based on Lytamine's law, a new product or technology can reduce its cost to half of the original after quadrupling its production. However, if you do not have this understanding, you will face high costs and "retreat from difficulties".

Cost trends based on Lyt's law (Source: Zuo Chenggang)

Earlier, we specifically analyzed the cost impact of intelligent electrical architecture, including wiring harness design, electrical design, EMC, vehicle operation and maintenance.

For example, the use of energy management algorithms can achieve intelligent energy saving and fuel saving, Bosch research shows that the generator output 100W electrical power, equivalent to 100km fuel consumption of 0.17L, in the 24V system is also 3.7A current, a little more than a 70W headlamp bulb. Therefore, the power management strategy can improve the economy of the vehicle's electrical system.

Source: BOSCH Automotive Electrical & Electronics, p. 338

However, many policymakers focus too much on direct costs and not too much on indirect costs. Therefore, the recognition of cost models will also hinder the promotion and application of many technologies.

02

Technical pattern awareness

As we said before, as an industry with a history of hundreds of years, many of its designs are inherited, and the inheritance means that there is continuity and less change. Traditional fuse relay technology has a long history, reliability is enough, the cost of use is low, comprehensive consideration, is currently the best solution after the balance of reliability and cost.

After the author has had technical exchanges with many OEMs, the first reaction of everyone is, is your technical solution without using fuses reliable? Wouldn't it be impossible to protect the burning of the thread? Has your solution been loaded? Verified? Who has used it at home?

When technical personnel in the automotive industry face new technologies, the first thing to consider is whether your design is reliable or unreliable, has anyone ever done this? Traditional technology, the fuse burned to replace a good one, and semiconductor technology is broken must replace the entire module, the cost is too high, unreliable simply can not.

Regarding the new electrical architecture, everyone's lack of awareness is mainly reflected in the following aspects:

(1) Reliability

We summarize in one sentence, that is: the lower reliability limit of the traditional electrical architecture is relatively high, but the upper limit is very low, while the intelligent electrical architecture based on the semiconductor scheme has a lower reliability limit, but the upper limit is very high!

In terms of reliability, we have a detailed analysis in the article "Kill the fuse and relay, automatic driving is safer", welcome to take a look.

Therefore, we can't say from experience that the traditional solution is already reliable, that's because you don't have a higher reliability requirement. The semiconductor solution is immature and may not be reliable at first, but we have to believe that it can be very reliable. As the saying goes, an eagle sometimes flies lower than a chicken, but a chicken can never fly as high as an eagle.

Finally, put a table of reliability parameters to compare:

(2) Inrush current

OEMs do electrical are more concerned about this, one is generally believed that semiconductor devices may be too sensitive, the inrush current at the moment of load start-up will lead to accidental protection, and the other is that semiconductor devices can not withstand the inrush current of the load, which may lead to damage. In previous articles, we did not conduct a detailed analysis of the load inrush current characteristics, but I will talk about it here by the way:

Inrush current characteristics of different load types of vehicles (Source: Zuo Chenggang)

Let's put a few more inrush current waveform diagrams:

Typical 24V/70W bulb start-up inrush current waveform (Source: Left Chenggang)

Typical 12V/150W DC motor start-up inrush current waveform (Source: Zuo Chenggang)

As can be seen from the above figure, the capacitive load inrush current is about 10 times, and the inductive load is between 3 and 5 times, let's look at the inrush current resistance of semiconductor devices:

HSD chip parameters for driving 2*70W/24V bulbs (Source: Infineon BTT6010-1ERA)

As can be seen from the above table, the inrush current resistance of the HSD chip (that is, the limit current limit current) is 10 times the rated current, which is enough to cope with all types of capacitive loads (except pure capacitors, large capacitors must be precharged), at the same time, the HSD chip only limits the output current to a value, and does not occur protection, turning off the load, which is an application scenario that chip designers have considered when designing the chip.

As for the inductive load for 3 to 5 times the inrush current, it is even less of a problem.

Let's take a look at the inrush current resistance of the MOS tube, and we will analyze the parameters with a power MOS used by Tesla:

MOSFET parameters used in Tesla's area controller (Source: Onsemi NVMFS5C426N-D)

This is a 40V NMOS, can be used for 12V system, the parameter is 1.3mΩ, 235A, which means that if you give enough heat dissipation, it can give you a dry current of 235A, but in fact no one dares to use it, you can not always give it liquid nitrogen cooling, we have to consider the engineering implementation. According to the author's experience, 1.3mΩ MOS gives a certain heat dissipation design, the full temperature range (-40 degrees ~ 85 degrees) dry to more than 40A should not be a big problem, the parameter table is pressed to 100 degrees, given to 29A, more conservative.

But we look at the pulse again, 100 degrees is 166A, note that the previous parameter is RJC, indicating that this is a longer transient current, similar to the inductive load impact, this parameter is also more than 5 times the rated current, the response to the inductive load 3 to 5 times the impact is no problem at all, and we look at the note 3, 1 second pulse is no problem.

Some small partners may ask again, why do I analyze HSD to talk about capacitive loads, and analyze MOS to talk about inductive loads? That's because small current-capacitive loads can be done with HSD, and because these loads are generally small, even if the impact is 10 times, it is no problem. In high-current applications, there is no corresponding HSD chip, only MOS can be used, and high-current loads are generally inductive characteristics, with long impact time, but small multiples. This is just like the application of traditional fuses, small current with sheet fast melting, anti-high multiple short pulse, to be fast protection; large current with plate type slow melting, resistant to large current long-term impact, to be reliable leather, so it seems that although it is two technical routes, but there is the same magic ah.

(3) Rated current

We have a detailed analysis of this problem in the previous article, and only put a conclusion here: under the intelligent electrical architecture, the load capacity of the distribution module can be matched according to the rated current of the load.

OEM's electrical design in the past on the current matching cognition is all based on the fuse, naturally will take the traditional fuse rated current and chip comparison, such as the original use of 20A fuse, he will require you chip also use 20A, this cognition is not right, then you need to abandon the fuse thinking, pay attention to the real load situation. The following picture is relatively simple and clear, I will put it again:

Chip load capacity comparison (Source: Zuo Chenggang)

Under normal circumstances, compared with the original fuse design, the rated current of the chip design can be smaller, or even half of the original, and correspondingly, the wire diameter design will be reduced.

Therefore, based on the chip design of the future intelligent architecture, everyone must change their cognition. When someone says that the current is 20A, you have to ask him if he has previously been insured with 20A, or if the load rating is 20A. Otherwise, everyone's understanding is not on a channel, communication will be a problem, and the current level is still directly related to the cost, and the cost difference between 10A and 20A chips is not doubled. In addition, the resulting cost reduction of wiring harnesses, simplified electrical design and other values have been discussed in detail before.

(4) Power properties

The concept of power supply attributes is well known in the automotive industry, after all, it has been used for so long, and even the well-known KL15/KL30 name is also proposed by Bosch in 1984, which shows its long history. In the past, the reason for defining the power supply attribute was to facilitate energy management, but after switching to the intelligent electrical architecture, you will suddenly find that the power supply can have no attributes, all lines can be defined as arbitrary power supply attributes, and fixed power attributes are naturally no longer needed.

The value brought by any power supply attribute includes, but is not limited to, supporting more free and complex energy management strategies, vehicle electrical architecture design optimization, wiring harness system optimization, network management design optimization, etc.

(5) Each road is individually controllable

Under the traditional architecture, the vast majority of circuits are uncontrollable, such as the regular electricity circuit, as soon as the whole vehicle is powered on, these circuits have electricity, you want to turn off is not off, passenger cars do not have a general gate, there must be static power consumption management, trucks rely on the general brake to manage.

In the intelligent electrical architecture, each circuit is individually controllable, but it is difficult for people who do traditional electrical design to have this cognition, in the traditional experience, how can so many circuits be independently controllable? As a result, they are unable to realize the value of this independent controllability.

Other values brought by the intelligent electrical architecture such as: the main switch of the power supply can be canceled, power control and energy management, harness optimization, configurable, programmable, upgradeable, iterative, etc., which we have detailed analysis in the last two articles, and the small partners who want to see it can go to see it.

(6) Independent protection of each road

Under the traditional architecture, the power distribution must be hierarchical, similar to the waterfall architecture, all the secondary circuits are through the first-level large fuse, and then distributed to the multiple secondary small fuse, and then to the electrical equipment. If something goes wrong at the first level, many second levels will be affected.

Also, in the traditional design, the fuse is responsible for protection, the relay is responsible for control, the protection and control are naturally separated, if the protection is shared, the control is separated, and the load goes wrong all the way, and all the loads will be completely powered off.

In the intelligent electrical architecture, because of the increase in reliability and the lack of power supply attributes, the power supply of the secondary power distribution is much less, so it can be considered that all the secondary terminal electrical equipment and the first level are parallel, rather than connected in series, which is equivalent to the second level of protection is actually independent.

Therefore, the natural advantage of using intelligent electrical architecture is that line protection and control are integrated, and all circuit protection is naturally independent and is not affected by other circuit failures.

(7) Protection and diagnostic functions

This point has been discussed in more detail in our previous articles, and I will list a few basic points here:

Recoverable after protection, recovery conditions can be defined by software, such as by ON, switching signals or powering up and down.

Customizable overload protection feature that traditional fuses are not supported.

Current sense function

Open circuit detection function (ON/OFF status)

Voltage detection function (overvoltage/undervoltage)

Some Tesla enthusiasts tested the self-recovery protection function of eFuse on the 2022 Model S and found that it can indeed be automatically recovered after the fault is lifted, such as the 12V auxiliary power socket (12V accessory power socket). Some are not easy to say, such as the drive control button (drive control button), the tester from the 400mA load, pulled to 1.5A, found that the output is immediately protected, but also the central control screen fault alarm, the information is accurate to which function has failed, and how to view. The tester said that the failure was not automatically recovered, but waited until the vehicle software was upgraded to recover, so the specific recovery conditions depended on the software strategy.

eFuse protection for Tesla 2022 Model S (Source: Teslatap)

Because the traditional architecture has no diagnostic function at all, so everyone can say that there is basically no cognition and is blank, just like when you use a feature phone, you can't imagine the impact of smart phones, including mobile payment, scanning code and other applications, which are new applications derived after the gradual popularization of smart phones. The protection and diagnostic functions based on the intelligent electrical architecture can also derive a variety of new applications, generate new value, and have a huge impact on the intelligence of the vehicle.

Leaving aside the system dimensions and cost factors mentioned earlier, changing people's perception of a new thing may be, in a sense, more important than all other efforts. As a disruptive and innovative technology, intelligent electrical architecture can bring value far beyond our imagination, so only by changing our cognition of it and breaking the limitations of traditional thinking can we explore its potential value and assess its impact on the automotive industry, and then jointly promote its landing as soon as possible.

Technical perspective

Earlier, we analyzed the problems faced by intelligent electrical architecture from the perspective of system, cost and cognition, in this chapter, we will talk about what technical problems will be encountered if we want to go to intelligent electrical architecture.

01

Technical scope

Before I get into that, let's talk about my experience. If you encounter a pure electric taxi, I will generally chat with the master for two more words, ask about the endurance, the cost of 100 kilometers of electricity, driving experience, and the difference between fuel vehicles and so on. I found that the master general feedback a problem is that the pure tram repair is very expensive, even if it is a small problem, the master does not dare to move himself, must open to the 4s shop, the ordinary roadside shop is not ok, do not dare to do, why? Let's think about it.

Most of the problems of fuel vehicles are mechanical problems, which are visible to the naked eye, electric vehicles in the electrification, mechanical problems are very few, the problem has become an electrical and electronic problem or software problem, invisible to the naked eye, the technical requirements for troubleshooting will then become higher, plus electric vehicles are new things, everyone does not understand, naturally do not dare to move.

Comparison of intelligent electrical architecture and traditional architecture switch box design

From the traditional architecture to the intelligent electrical architecture, there will be a similar problem - the traditional electrical architecture is all in the mechanical and electrical category, in the OEM is the electrical sector, and electronics are not in line, but after upgrading to the intelligent electrical architecture, it is fully electronic.

Traditional distribution boxes belong to labor-intensive industries, fighting is low cost, the design of the technical content is not high, but after the electronic, the original traditional distribution box manufacturers are ignorant, do not understand at all, which touched their cognitive blind spots. The OEM's electrical department also does not understand, although the electrical schematic of the smart distribution box looks simpler, but in their view it is a black box, because there is software logic and configuration in the middle, and it is useless to look at the schematic alone.

In addition to the cognitive level and technical ability level, there are also some technical issues that were not paid much attention to before.

For example, the load characteristics, in the traditional electrical architecture to do electrical design is not bad on the line, because the insurance is originally graded, you can only 10A, 15A to choose, the wire diameter margin is generally enough, you do not need to understand the load characteristics in great detail, according to experience to the problem is not big, no on the fuse on the upgrade of a gear, the problem is solved. However, when doing electronic design under the intelligent electrical architecture, it is completely impossible to understand the load characteristics, which we will analyze in detail later.

Here we simply popularize a car wire small knowledge, the car wire overcurrent ability is actually far beyond everyone's imagination, extreme, we take the smoke time to talk about - that is, how much current you give a wire, how long can it smoke (because of the type, some wires are not smoke), the car wire can be 5 times the rated current 5s within 5s do not smoke, and the fuse, even if the slow melting insurance, 5 times the current is burned within 1s.

Overcurrent capacity of automotive wires (Source: Infineon)

As shown in the figure above, the short-term overload capacity of the wire is extremely strong, far exceeding the chip insurance, and the biggest problem of improper use of the wire lies in the insulation damage caused by long-term overload and heating.

In-vehicle fuse blow time characteristics (Source: Littelfuse)

As can be seen in the table above, no fuse can withstand 5 times the current up to 5s.

Therefore, the problems faced by traditional distribution box manufacturers when they get involved in intelligent electrical architecture include, but are not limited to:

Electronic system cognition, including electrical and electronic architecture, control logic, network communication, etc.;

Electronic product design experience, including system, hardware, software design;

Electronic product testing methods, reliability design, etc.;

Automotive electronic test standards, including electrical, environmental, durability, ESD, EMC, etc.;

In-depth understanding of load characteristics and operating modes;

Balance between design options and costs.

Then you say that the manufacturers of traditional distribution boxes can't be sure, and the manufacturers of electronic modules? They understand electronic design, yes, but you are only right about a little bit of this sentence, they only understand the small current electronic design in the profession, and the vehicle electrical architecture and high current design have touched their knowledge blind spots.

The problems faced by traditional electronic module designers when they get involved in intelligent electrical architecture include, but are not limited to:

System cognition after electronicization, including vehicle electronic and electrical architecture, electrical principles, electrical design;

Vehicle energy management, power-up and power-up strategy;

Vehicle load type, load characteristics;

Protection characteristics and harness matching (traditional fuse matching is experienced and recommended design);

Housing design, including heat dissipation, protection level, installation method, etc.;

High-current wiring design, including connectors, bolts, etc.;

High-current board-level design, including device heat dissipation, PCB current carrying, etc.;

Selection of high-current devices, understanding of MOSFET characteristics and parameters, etc.;

High-current product testing and verification methods.

For the board-level high-current design scheme, the traditional electronic module manufacturers should lack corresponding experience. If nothing else, as big as tesla boards, so many busbar current-carrying designs, most people have never seen it. Because the electronic design generally does not exceed 10A current, most of them are mA level, static power consumption is generally A level, for hundreds of amperes of current everyone has no concept, can not imagine; even, the traditional electronic module manufacturers may not even have the corresponding DC power supply equipment (commonly used DC power supply in 30A, the largest is not more than 100A), more than 4 square wires have not been seen much (home into the home wire diameter is generally 4 square).

The author believes that the transformation of traditional distribution box manufacturers is more difficult, because there are too many knowledge blind spots involved, from hardware design, software development, to electronic product testing experience, the knowledge architecture is different, and it is difficult to make up.

Traditional electronic module manufacturers cross-border electrical design, the difficulty is not small. For example, the understanding of the electrical principles and electrical design of the vehicle, the in-depth understanding of the load characteristics, etc., as well as the design of the structure, electrical, wiring harness and other aspects of the traditional distribution box, the knowledge structure is also different. There are also many pits in this regard, such as how big the bolts of the high-current wire are, and how much torque range is required during assembly, which is a traditional electronic module manufacturer without any concept.

Therefore, the author believes that the most likely to achieve the landing of intelligent electrical architecture is Tier 1, which has the design capabilities of traditional distribution boxes and electronic modules, and they can gather the two departments that have completely failed to line up in the past, and work with OEMs to complete the design of intelligent electrical architecture and gradually land.

02

Chip solution issue

Small current design is good, you can use a mature integrated chip scheme (HSD high-side chip), this previous article has been analyzed, passenger car applications below 25A are very mature, commercial vehicles below 10A have.

High-current solutions, passenger cars currently have mass production of 30A HSD chips to choose from, the future current level will continue to increase. However, in terms of commercial vehicles, according to the author,the major chip suppliers do not have new roadmaps for the time being. The high-current scheme can only use the drive chip + MOSFET discrete scheme, which has the following problems:

The overall scheme is complex and the comprehensive cost is high;

Complex and costly current sensing (shunt+amp op amp scheme);

Less protection function, slow protection speed;

The protection circuit is complex and the protection strategy is complex;

The diagnostic function is small, and the design of the diagnostic function is complicated;

High-current applications require MOSFET parallel design;

This scheme requires the addition of discrete circuits according to the application requirements, and shunt and amp must be added to the current detection, including protection functions and diagnostic functions, the more functions, the more complex the circuit. Because the protection function is mostly implemented by the MCU, the speed is not to say that the speed is slow, and the software policy is also complicated.

Let's first appreciate Tesla's plan, everyone has a bit of intuitive feelings. The black square in the following figure is the power MOSFET, the silver white is PCB Busbar, used to do large current carrying, brass color is shunt, used for current detection, large current shunt global can do not do much, don't look at it is a copper sheet, but the accuracy and temperature coefficient of the material are extremely high, it is a basic material discipline, Germany is doing very well, of course, it is not cheap.

Tesla's drive chip + MOSFET discrete solution

Isabelle enhuette's car grade shunt in Germany

In addition, although mosfets have their corresponding rated current and pulse current parameters, soa (Safe Operating Area safe operating area) must be considered when designing, which is very important for MOS design. Specialized driver chips generally take into account the MOS drive voltage, current, junction capacitance charging and discharging time and other switch-related parameter design, but if you build your own circuit, you have to consider more. In addition, even if the driver chip has a protection function, it is generally limited to short circuit protection (based on VDS) and MOS overtemperature protection, if the chip does not support shunt current detection, other protection functions derived from current detection such as overcurrent protection, open circuit protection, current limiting, etc., it is more complicated to build a circuit by yourself, including hardware circuits and software strategies.

Driver chip + MOSFET discrete solution (Source: Zuo Chenggang)

There is also the problem of MOSFET parallel design that must be used for larger current levels, which requires high requirements for the consistency of the MOS device itself and the hardware design of the product. For example, PCB current sharing, transient energy, peak shutdown voltage, parasitic oscillation and other issues, especially for the inductive energy release problem when applying inductive loads, it is easy to have problems if it is not handled well, and the circuit with the most fragile or least impedance will be blown up first.

Power MOSFET high-current parallel applications (Source: IR/Infineon)

The following is Infineon's summary of the parallel application of Power MOSFETs, the first and third articles both mention the current balance problem, even if the MOSFET is a positive temperature coefficient device, it naturally brings some self-balancing advantages, but there are also its limitations (steady state and switching state are no problem, but the short circuit problem is very large, not mentioned below). The second is the SOA issue, which shows its importance.

Power MOSFET high-current parallel applications (Source: IR/Infineon)

Here by the way to popularize some of the basic knowledge of MOSFETs, MOS RDSON is the on impedance, the unit is mΩ, the smaller the value, the greater the current, the more expensive. The consistency and stability of vehicle-grade MOSFETs is inherently higher than that of consumer and industrial grades, but even the same batch of MOS has huge parameter differences because they may come from different wafers. Of course, these differences do not exceed the parameters specified by datasheet, but these differences will still cause many problems when applying in parallel.

MOSFET batch parameter differences (Source: NXP)

We look directly at the conclusion, this application document analysis for half a day the final conclusion is: not recommended multiple MOS parallel applications (simply nonsense), think that a single low RDSON MOS is better, even if parallel n, in the end will certainly not reach n times the effect (unless you are not bad money, the 1st, 2nd is worthless advice, in fact, this everyone knows, that is, the cost is not allowed). Finally, it is recommended that if you really want to parallel, do not exceed 3, and then group parallel, Tesla is playing like this.

Interested partners can go to see the application documentation of NXP and Infiniton, which will not be elaborated here.

Power MOSFET Parallel Recommendation (Source: NXP)

Finally, let's take a look at the cost and feel why everyone must use it in parallel.

NVMFS5C410N-D is tesla in use of the largest MOS, 0.92mΩ, 1.8 US dollars a piece; larger than it, 0.63mΩ to 5.2 US dollars, the price doubled 3 times, the current is actually not much larger; small double 2.3mΩ, the price dropped to 37%, the larger the more expensive, and the price is completely disproportionate. So in the previous picture you can see teslas are all 4 in a group of parallel start. The DC-DC input is directly 2*4 in parallel, and the usage is the same as NXP's recommendation, which is grouped. But Tesla was bold and directly went to a group of 4 (just ask Volkswagen Ford if you dare).

Power MOSFET price (Source: Onsemi)

03

Load characteristics - reliability issues

The reliability mentioned here involves not only reliability, but also robustness robustness.

Usually, what we call reliability generally refers to durability and failure efficiency, in fact, it can be used for a long time, but there is very little bad (MTBF and FIT value dimensions). In the intelligent electrical architecture, as a distribution box that realizes the distribution and control functions, the function is often more important than the box itself is not broken. If the function is invalidated, although the box is protected, it is not broken, which is meaningless for the actual application, and from the user's point of view, it is unreliable and always bad.

So the reliability here needs to consider more robustness, robustness and robustness, which and the sensitivity and accuracy of semiconductor device protection is a pair of contradictions, and the design is to balance this contradiction, to achieve stable work, reliable protection, but for the vast variety of vehicle load characteristics, this design is difficult, such as:

When the line is normal:

1) The load inrush current should not be protected;

2) The load should not be protected when it is overloaded for a short time;

3) The protection can be restarted, and after the line fault is eliminated, the line should return to normal;

When the line is abnormal:

1) Hard short circuit requires rapid protection;

2) Soft short circuit according to the needs of protection;

3) The protection speed is appropriate to ensure that the wire cannot be damaged;

4) After the device is protected, no damage or parameter degradation can occur;

Above we have mentioned some load characteristics of the problem, the traditional electrical design fault tolerance is strong, because the margin in all aspects is large enough, the design constraints are small, such as the difference in the cost of different current levels of the fuse is very small, plus the fuse is the need to replace maintenance, etc., which leads to the extensive design of the traditional electrical design.

However, in the era of electronic design, the design constraints have completely changed, such as the improvement of chip sensitivity, while bringing accurate protection, it must require accurate matching in the early stage of design, otherwise there will be problems in the later stage. For example, it is easy to misbehave, or the chip is burned when it fails; if the margin is increased in the early stage, it will lead to excessive BOM cost, because the chip cost is strongly related to the current level.

Therefore, in order to achieve a balanced design of cost and reliability, it is necessary to understand the chip characteristics and load characteristics in detail, such as HSD chips have rated current, limited current parameters and temperature characteristics, etc., and the protection aspect has overcurrent protection characteristics, short circuit protection, thermal protection characteristics, etc. The drive chip + MOS scheme is more complicated, and many protection characteristics depend on the specific hardware design and software strategy.

At the load level, whether the load is capacitive or inductive, the inrush current waveform, peak and duration must be considered, and the motor load also needs to consider whether overload protection and stall protection are required.

We have analyzed the load characteristics earlier, but in practical applications, you will find that some consumer appliances have a variety of load characteristics. For example, some controllers, when powering on, is capacitive, because the controller has input capacitance, after working, if the load controlled by the controller is the motor, it has inductiveness, and the load is similar to the resistance of the heating device. In addition, different load conditions are not the same, which also needs to understand the electrical principles and specific functional applications of the vehicle.

Therefore, the reliability of the final function must depend on the load characteristics and operating conditions assumed in the early stage, and if the load characteristics change, the application may have problems. If for uncertain load characteristics, the protection characteristics are very difficult to design, stricter may conflict with the load characteristics, resulting in malfunction, wider need to increase the chip margin, resulting in higher costs, or the protection is not protected, increasing the risk of use.

Characteristic curve of a 12V DC motor (Source: Infineon)

As shown in the figure above, this motor load has a rated current of nearly 30ms 5 times, if it is a protection strategy for this load, it is necessary to filter out this inrush current to ensure that there will be no misoperation. Considering the uncertainty of the motor stall time, the time needs to be extended again. But if the load becomes resistive, it may be that the wiring harness is broken and needs to be protected.

Load characteristics, protection characteristics and harness characteristic curves (Source: ST)

The above figure perfectly illustrates the relationship between the wiring harness characteristics, protection characteristics and load characteristics, the yellow is the load current/time characteristic, the green is the device protection characteristic, and the red is the harness I t characteristic. Therefore, for the design, the following requirements must be met at the same time:

The protection characteristics must be wrapped around the load characteristics, otherwise they will be misactive;

The protection characteristics must be within the characteristics of the harness, otherwise the wire may be burned in the event of failure;

The three curves must be guaranteed to be free of crossover over the full temperature range.

Earlier we have analyzed in detail the impact of load characteristics on the protection strategy, but there is no specific data comparison, you may not have to intuitively feel, I will compare the previous set of data:

Load characteristics and protective parameter characteristics (Source: Zuo Chenggang)

Why do I keep emphasizing the load characteristics in this article? On the above load characteristics and fault parameters, not to mention the design for versatility, that is, for a specific type of load, you give me a protection strategy to try? Therefore, a generic design or a generic load type protection design is difficult. The design must be load-specific, and the reliability and cost of the design must be balanced. From this point of view, the difficulty of passenger car design is much lower than that of commercial vehicles, and the small partners who do commercial vehicles can be a little arrogant, although the technology is backward, but the difficulty is greater.

04

Quiescent current problem

As we have analyzed earlier, traditional fuses have many benefits, including simple and easy to use, cheap, and leathery, but there is also a point that has been ignored by everyone, that is, the current consumption as a passive device. The fuse is just a piece of metal material, as a complete passive device, it does not consume any additional current, can continuously protect the cable, prevent any short-circuit failure, and can play a protective role at any time.

When the vehicle is running, this advantage of the fuse cannot be reflected, but when the vehicle is in a parked state, the consumption of static current by the whole vehicle is required, and the advantage of the fuse is reflected at this time. It can quietly protect the entire vehicle without adding any additional quiescent current consumption, which is not possible with semiconductor chips. It is better to maintain the chip conduction alone, but at least a few tens of A, if you also need protection functions (nonsense, if you don't protect the car park there is a risk of burning), then the current level will be to the mA level.

Here to popularize a little knowledge, the general high-side drive solution, whether it is with HSD integrated chip or driver chip plus MOSFET, a channel to maintain the conduction plus protection function, generally need about 5mA, which is OEM death can not be accepted (general passenger car OEM requirements vehicle 15mA ~ 20mA or so).

Quiescent current parameters for several HSDs (Source: ST, Infineon, TI)

For example, if there are 5 functions that require constant electricity, it will require an additional 25mA or so, if the battery capacity is not large, then the car will be out of power for a period of time. For example, the original can be put for a month, now it can only be put for half a month, you came back from a long errand, found that the car can not start, what do you say you feel?

The relationship between battery capacity, quiescent current and parking days of passenger car vehicles is compared as follows:

Relationship between vehicle battery capacity, static current and parking days (Source: Zuo Chenggang)

Commercial vehicles such as medium and heavy trucks, batteries are generally large, often more than 100Ah, plus there is a general gate, there is a truck is used to make money, the use of different models, generally not likely ten days and half a month to stop, the loan has to be repaid. So commercial vehicles are a bit better than passenger cars.

In addition, for pure electric vehicles, the current trend is battery miniaturization. For example, Tesla uses a small battery of 33Ah, because there is no need to undertake the task of starting the motor, and as long as it does not power, there is a high-voltage battery there to support, and it is not needed. However, once the high voltage is powered down, the low voltage static power consumption is not well controlled, resulting in the battery losing power, even if the high voltage has electricity, the vehicle can not start. Of course, pure electric vehicles because as long as the high-voltage contactor can be absorbed, the vehicle can be on the high voltage and start, high-voltage DC-DC can charge the low-voltage battery, which is much smaller than the fuel car starter on the battery remaining power demand, but the corresponding requirements can be reduced some of the remaining power is debatable.

So for the problem of increasing static power consumption after electronicization, Slav is ingenious, the high voltage is not powered down at all, this is a system engineering design, mainly to support all Online services, but also to solve the problem of static power consumption caused by the use of semiconductor design, of course, tens of mA power consumption increase in front of Tesla's 2.6A quiescent current is a younger brother.

But this problem may be a real problem for other OEMs, unless you can copy the Tesla system design of pure electric vehicles, and your users can still buy it. Because not all users can accept a 1% daily power loss of the high-voltage battery after parking, you can't convince the customer without bringing some additional value. At the same time, the audience groups of different brands are also quite different, the cognition is not in a channel, and there are great differences in the acceptance of some designs, such as iPhone users on the poor signal, small battery and no fast charging can be accepted, change Android users do you dare to think?

For fuel vehicles, because only low-voltage lead-acid batteries, if the requirements of more constant current load, static current can not be lowered, this problem is unsolvable, there must be a uA level solution.

In response to this problem, ST has launched a new vehicle-grade driver chip, which is designed for high-side drive external MOSFET applications, with a withstand voltage of 60V, which can be used in passenger cars and commercial vehicles. ST's chip has a standby current as low as 70 A and can be continuously powered while in standby, while also providing protection.

Drive chip standby quiescent current parameters (source: ST)

Finally, to summarize, this article analyzes the difficulties of intelligent electrical architecture landing from four dimensions such as system, cost, cognition, and technology, and the author believes that among these four points, cognition may be the most important point and the most difficult point to change. Smart electrical architectures, as a disruptive technological innovation, change people's perceptions of them, and in a sense, may be more important than all other efforts.

Due to space limitations, some technical points are not covered, such as why should MOSFETs be considered to use parallel designs, while HSDs cannot be directly paralleled? Why does the high-side power supply design have to consume power? What are the differences between automotive electronics' engineering thinking and traditional electronic design? Welcome to leave a message, follow-up we can continue to analyze.

Bibliography:

1. Munro And 3IS Compare Tesla, Ford & VW Electrical Architectures， 3IS

2. Zonal_EE_Architecture-Towards a Fully Automotive Ethernet–Based Vehicle，Visteon

3. NXP Philippe Dupuy-Improving the automotive power distribution architecture

4. Why did Tesla kill fuses and relays? Nine chapters of intelligent driving

5. What kind of electrical architecture is required for autonomous commercial vehicles? Nine chapters of intelligent driving

6. Kill the fuse and relay, automatic driving can be safer, nine chapters of intelligent driving

7. Fuse Function with PROFET application note， Infineon

8. Paralleling Of Power MOSFETs For Higher Power Output，International Rectifier

9. Using power MOSFETs in parallel，NXP

Write at the end

About submissions

If you are interested in contributing to the "Nine Chapters of Intelligent Driving" ("Knowledge Accumulation and Collation" type article), please scan the QR code on the right and add the staff WeChat.

Note: When adding WeChat, be sure to note your real name, company, and position

As well as information such as submission intentions, thank you!

"Knowledge Accumulation" manuscript quality requirements:

A: The information density is higher than the vast majority of reports of the vast majority of securities companies, and it is not lower than the average level of "Nine Chapters of Intelligent Driving";

B: Information should be highly scarce, more than 80% of the information needed is not seen in other media, if based on public information, there needs to be a particularly strong exclusive view. Thank you for your understanding and support.