New Chip: 5-10% Better Hybrid MPG

Funny that they call them computer chips, and not just once, but several times. To the writer, I guess the idea is that the big cylinders are capacitors, small cylinders, often with stripes, are resistors, and rectangular prism shapes are computer chips. Kind of like the four elements of our ancestors times.

I imagine that SiC diodes are being integrated as well, since reverse recovery times and loses are greatly reduced with their use.

 

why would toyota announce this like some shyster aftermarket product, instead of just putting it in the gen IV and announcing new mpg's?

 

2020? i hope i'm still around.

 

perhaps. sounds more like a gm announcement to me.

 

so what is going to take 6 years?

 

they mention that they are already being used on trains and a/c units. must be a big change to move them to automotive, or perhaps it's a cost problem.

 

For all electric, high power devices, you save so much power that the high cost of new technology is not critical. For Toyota to save a mere 5%, getting the cost down to dirt cheap will be paramount. (notice in Jeff N post, they are shown using a much smaller die, larger die sizes will reduce costs, as will finer detailed lithography. These things will come with experience) There are a Lot of steps where we are very good at using silicon, but we will have to relearn how to best use SiC

Semiconductor device fabrication - Wikipedia, the free encyclopedia

Once perfected, they offer lower power loss, so they will need less space and less cooling infrastructure allowing tighter packaging. With less to pack under the hood, MPG can rise.

 

My guess is that they will eventually be built with several in a single wafer for lower than Prius power uses. Much like op amps and logic gates. Use what you need and leave the other inputs open. With the lower "break over" voltage means less heat in forward conductance. Heat and inverse voltage (the voltage applied when it isn't conducting) is what kills diodes (and anything else silicon). Other than the usual noise, which tends to mess with digital and analog circuits, beefing up a normal silicon component is the best way to increase reliability in any motor control switch.

The other side of the argument is that adding carbon tends to increase electrical noise (diodes are already notorious for "noise"). Physical placement can alleviate much of this but coming within budget is the enemy when designing something that is to be mass produced and the "bean counters" ALWAYS win.

 

It sounds like from the video that Toyota's main challenge with implementing SiC tech is going to be designing something that is very reliable with the main tech challenge being handling heat dissipation and very high duty cycles

 

Interesting. The prototype had 5% better efficiency, but they decided to call it "up to 10%"?

 

Their stated goal is 10%, the prototype got half of that, yes.
"Through use of SiC power semiconductors, Toyota aims to improve hybrid vehicle (HV) fuel efficiency by 10 percent"
does not say they have done so, it only states their goal.

"Toyota, which has achieved 5 percent gains in fuel efficiency on prototype vehicles, plans to start installing the chips in its cars around 2020, Hamada said, although it does not expect to immediately achieve the targeted 10 percent fuel savings."
again contrasts what they have now versus their stated goal. It does not even predict 10% by 2020.

 

Scaling up the manufacturing. Making millions of chips a day for a production auto requires making hundreds of production machines for the chips. Making hundreds of production machines requires making hundreds of production machine parts. And you don't start building those until you KNOW that the final product will work perfectly for every car. So to recount:

1) Spend a year testing the chips to meet every single requirement and be cheap enough to sell in volume
2) Spend a year designing all the production machinery for building the chips AND the machines for building the inverters that hold the chips
3) Spend a couple of years getting the production machinery built
4) Spend a year getting all the production machines installed in the factories
5) Spend a year getting the production line up to full production rates with perfect chips, perfectly installed.
6) Build EVERY hybrid auto of a specific model with the new chips.

You may laugh, but six years for something this major is a no nonsense schedule.

 

There are two major factors improving efficiency with inverters. Higher switching frequency and lower resistive losses. I expect the prototype test just had a direct substitution of the SiC chip for the silicon MOSFETS. Changing the switching frequency would have required replacement of a substantial part of the hybrid system, including the control software, etc. So the 5% is just the resistive loss improvements. The switching frequency changes coming later to provide the other 5%.

 

Silicon carbide - Wikipedia, the free encyclopedia Hints that while SiC produces less heat, getting it out of the die may need new technology.

 

Do we know that the existing silicon equipment won't work for SiC, so all new equipment and fab lines are needed?

As new silicon production moves to the latest largest-wafer fab lines, it tends to free up older lines built for smaller wafers. Some of this ends up going into solar PV production, but I'd hope that significant chunks could be adapted or modified to SiC.