Toyota testing semiconductor technology - SiC

Unfortunately Moore's Law doesn't really apply to power electronics. Logic switching speeds are primarily increased by advancements in lithography, ie the ability to make smaller and smaller transistors and wires on chip. This is also great for production cost and density. But one of the consequences of shrinking the transistors and wiring is lower voltage breakdown, and reduced current capacity. This is part of why we've seen logic levels fall from 5V, to 3.3V, to 2.5V, to 1.8 and now to 1.2V and even 0.8V over the years. That creates its own challenges but is generally good news for power efficiency vs. speed. But its really lousy for high voltage / high power applications, so many of those applications are "stuck" on older coarse geometry processes. Silicon is generally not great for those applications anyway, but the huge volume production advantage still makes it a common choice. Many high power / high speed chips have still hung on in more expensive alternate technologies like GaAs, InP, etc for applications like cell phone power amps. But their much higher production costs have spurred development of a number of Silicon compatible technologies to replace these products in different applications, such as SiGe for high speed analog and SiC for high power. Though they have their own challenges, these processes bring many of the advantages of these alternative technologies to the high volume silicon production world. This can result in greatly reduced production costs, or as in this case greatly improved performance, density and efficiency compared to just getting by with plain silicon.

Pretty cool to see Toyota going this route.

Rob

 

Agreed, but there's no reason it shouldn't (except for the reasons I mentioned previously).

 

The above discussion somewhat overlooks what causes the delay between engineering development and production volume fielding. It is not until you get every single last detail designed that building the manufacturing line can commence. For semiconductors, the foundry process requires quite a few optimizations to be established and quite a few preproduction wafer runs to be completed and verified. Even then, the production line must be up and running a full output prior to any car assembly line incorporating all the new electronics. The few years it takes to get all this accomplished and working indicates Toyota is moving fast, not slow, with this technology.

 

True, unless they design it upfront to be acommodate it.

 

Not to mention to retest for EPA and CARB emission certificate.

 

SiC power MOSFETs and schottky diodes are being manufactured in production volumes right now if you can live with 1200 v breakdown and the Prius can. The full benefit of SiC is crazy high voltages, but crystal defects are holding that back. Toyota could do SiC today if they really wanted to. Supply and demand will dictate a premium in price, but this can be made back by other savings.

 

There's actually a very good reason why it doesn't. In the article you link, it states that Moore's Law is a) not a law but an observation that has held true as much through self fulfillment as anything, and b) applies to "the number of transistors in a dense integrated circuit."

The IGBTs used in the inverters are anything but part of a "dense integrated circuit." Each IGBT die is roughly 1 square inch in area, and effectively contain 1 transistor. Compare that to a "dense" circuit like the current gen Apple A8 processor which contains 2 Billion transistors in approximately 0.138 square inches.

In a power device such as these IGBTs, the challenge is not density per say, but are primarily
1) getting heat out fast enough to keep it from exploding
2) keeping voltage breakdowns high enough to keep it from exploding
3) keep loses low to improve efficiency and keep it from exploding
4) keep yield, matching and reliability high enough to keep it from exploding

Pretty much everything that drives Moore's law in processor's is in direct opposition to these requirements. The associated speed / density / cost improvements are primarily achieved by making both the transistors and their wiring smaller and smaller. Smaller transistors mean lower voltage breakdowns. Current high density Si processes have breakdowns <1V. Smaller wires mean they can handle less current. It likely only takes uAs to pop the ~50nm wide metal traces on these processes. When you are planning to put 500V and 100A through a piece of silicon the size of your thumbprint, none of these things are your friend.

Fortunately I'm a receiver guy, and don't have to deal with the pains my PA comrades go through to keep the magic smoke/fire safely contained within their chips. The again Moore's Law isn't much use to me in high speed / low noise analog design either.

Rob

 

Toyota just announced that the test showed 5% increase in fuel economy. Their goal is to improve 10% with further optimization. Target launch is around 2020 so, probably for Gen5 Prius.

Toyota started the test in February 2015 in Toyota City, Aichi Prefecture, Japan. At this point, it has already confirmed a fuel efficiency improvement of 5%. If the motion control is optimized, it is possible to realize a fuel efficiency improvement of 10%, which is the company's goal, it said.

In the test, a car prototyped based on the HEV model of the "Camry" was used. SiC power semiconductors (MOS transistors and diodes) were used to replace Si power semiconductors (IGBTs and diodes) in the boosting converter and inverter (for controlling the motor) of the power control unit (PCU), which controls the driving power of the motor. The SiC transistor is a trenched MOSFET manufactured with a SiC wafer whose diameter is four inches (100mm).

In addition to the fuel efficiency improvement, the employment of the "full SiC" system is expected to reduce the volume of the PCU by about 80%. While the previous PCU using Si power semiconductors is equipped with 18 to 24 power modules (cards), SiC potentially enables to reduce the number to about 14, Toyota said.​

Toyota Announces 5% Fuel Efficiency Improvement With SiC - Nikkei Technology Online

 

​

I don't see how a camry hybrid can get 10% better fuel economy - company goal - just by changing electronics efficiency. Do even 10% of the losses come from electronics? I'm curious. I have to assume they will put a more powerful battery in there, and use different software to optimize engine use to get the gain.

​

20% the size, I think they can do that, and they can probably drop power modules even more

 

I don't know what motion control they were talking about.

 

Remember the current power electronics has a significant cooling loop and dedicated electric pump. That heat is energy loss. At lower speeds, the power required is significantly low:

It is entirely feasible that at lower speeds, the switching overhead is a significant part of the energy budget (see right hand scale.)

• 1 HP ~= 745 W

With more efficient switching, the waste heat loss goes down and it may be possible to eliminate or significantly reduce the cooling loop and the coolant pump.

Also, never forget that when the Prius is running on engine power, ~28% of the power flows over the electrical path whose efficiency is ~91% x 91% ~= 82.8%. Bring the efficiency up to 95% and it becomes ~90.2%. It is only the low power over the electrical path that keeps it efficient.

Bob Wilson

 

Sure but what percent of power is used at low speed? How much of the overhead is in inverters versus motors? On a 15,000 mile year are these inefficient low speed miles even 1000 miles (<7%), and low speed miles use less power than higher speed miles (drag).

SiC is less harmed by heat so you can definitely reduce cooling, but cooling takes very little energy. Liquid cooled power electronics are more reliable, so its probably not a great idea to eliminate it to cut costs, still smaller systems are lighter and less expensive.

Most of the inefficiency on the electrical path are losses in batteries and motors not inverters and other power electronics. So I can see the path saving 10%, but that would likely require a more powerful battery (maybe lithium ;-)) and more efficient motors (coming independently of the SiC in each generation).

 

I think the goal is to use the coolant from ICE, so there is only one loop and one pump.

Gen1 to Gen3 inverter need separate coolant since the operating temp is so different than the ICE.

I think SiC would be used not only in the inverter but also in DC to DC converter. The Power Control Unit contains both.

Every time you go from 500-650V AC motor to 201v DC battery during regen (and vice-versa), all the power flows through the converter.

Power semiconductors are found in power control units (PCUs), which are used to control motor drive power in hybrids and other vehicles with electric powertrains. PCUs play a crucial role in the use of electricity, supplying battery power to the motors during operation and recharging the battery using energy recovered during deceleration.
​

Toyota to Trial New SiC Power Semiconductor Technology | TOYOTA Global Newsroom

At highway speed, a lot of power is also routed in the energy-recirculation (Heretical) mode. It happens when ICE is making more torque than necessary. MG2 (becoming generator) power is routed into MG1 that is spinning reverse providing propulsion power to the wheels.

 

I like it! Good call.

Bob Wilson

 

Toyota Australia seem to think that the 2016 Prius will have SiC, as indicated by this article from January 2015.



Toyota to refresh hybrids in 2015

In the words of Tony Cramb, Toyota Australia's Executive Director of Sales and Marketing. Speaking before an audience of journalists earlier this week, Cramb also officially announced 2015 would see the launch of the long-awaited new-generation Prius.

"I can confirm that the Prius family will be revamped, with significant updates for Prius C and Prius V in 2015, and the arrival of the next-generation Prius..."

This new hybrid is anticipated to lift fuel efficiency up a further step from the current model, introducing a lighter Hybrid Synergy Drive system with a new semi-conductor intended to reduce power losses in the car's power control unit. Additionally, the new internal-combustion powerplant will deliver improved thermal efficiency, and the car will be cheaper to build – potentially translating to lower retail pricing for consumers.



Excerpt from:
Toyota to refresh hybrids in 2015 – Car Reviews, News & Advice - carsales.com.au

#3485 David7, Jan 8, 2015

 

Interesting. If SiC makes it into production, there will be more room under the hood. There is no reason for the 12V to be in the trunk. I expect it'll go back under the hood.

Edit: Hmmm.. He was only quoted for the 2015 launch of Gen4. The line bolded is from the author's speculation.

 

I think the breakdown that is fixed and easy to state is torque, 72% on the mechanical path and 28% on the electrical. Power, being a product of torque and rpm, breaks down in a variable way depending on the relative rpm of the two motors. For example, in any low speed situation where MG2 is turning less than a third the revs of MG1, there's actually more power following the electrical path than the mechanical.

-Chap

 

Toyota is saying 2020, this likely is because

1) Full year of testing
2) My guess is that part of the gaines they are talking about come from higher voltage motors, so further dependancy for not making this year, but ... they could do in a gen IV update.
3) Biggest problem with SiC is cost, but like most chip technologies costs come down over time. 2020 may be their prediction of when it is less expensive because of manufacturing advances.

Part of the reason battery is in the back is to move the weight cg back, which may be even more important in the gen IV. Still the prius is pretty unbalanced right now, so lower cost battery up front makes sense. Its not a miata, and lower cost may be a better trade off.

 

If we agree the engine turns at one RPM into the 72%/28% torque split, I have no problem with:

• HP_mechanical = ((.72% * Torque) * RPM) / 5252
• HP_electrical = ((.28% * Torque) * RPM) / 5252

Then when the engine is off, RPM=0, all motive power is via the power electronics.

Bob Wilson