New Chip: 5-10% Better Hybrid MPG

One of the articles mentions setting up a new fab at Toyota, if it was cheaper to reuse silicon fans, I think they would have

 

This will also mean the inverter can be smaller, and need less cooling. Both "good things" (tm). Less heat may also translate to better reliability.

But as already stated, SiC technology is not here yet, though it's very close. I believe International Rectifier is making some parts now. Mostly diodes, in puck cases for very high power applications when I last checked.

 

i think 6 years gives you enough time to fail and people won't remember. but i'm looking at this from a layman's perspective, not scientific. for me, it's just another announcement in a long list of announcements. some pan out, most don't.

 

I can't imagine that simply changing these parts would pop a prius from 50mpg to 55mpg. How lossy are the current parts?

 

According to the book on the development of the Prius, Toyota specifically wanted to have all the production transistors built by them in their factories optimized for their cost and quality approach. It specificially mentioned they were determined to invent (or reinvent) the transistor fab infrastructure to be wholly Toyota owned and controlled. I would expect a whole lot of process improvements have happened since these original fab lines were made, so there may be a whole lot more planned than just a material change to SiC.


When I visited the Intel Developmental Fab in Portland OR many years ago, I learned that fab facilities have to be designed from the ground up every time a process change is made. Since the entire production fab is automated, just upgrading piecemeal is not really possible, since the process dimensional improvement affect every single stage of the fabrication line. Photoetching, Ion Implantation, Passivation, Wafer Testing Probes, etc are all improved with different machines designed to work in the smaller geometries involved. That was the whole reason for Intel having a "prototype fab center". Once they got the new process working in Portland, then they overhauled their fabs in the US, Asia, and Europe.

 

That sounds quite strange, but I pulled up this description of gen II
Special Issue: Inside the Toyota Prius: Part 5 - Inverter/converter is Prius' power broker | EE Times

We have a group of chips but toyota custom IGBTs and diodes. I had no idea that they fabbed them themselves. I can see how you can cut costs if you change material so that it can be air cooled instead of liquid cooled. I just can't figure out 10%. I thought those electronics were over 90% efficient, and most of the power flows from ice to axle mechanically, so less than half the power goes through inverters.

 

One of the recent projects I was involved with was replacing Si with SiC Power Amplifier Chips. The toughest engineering task is heat removal. The power density is higher since the SiC has so much more current handling capability. So while the I2R losses are lower since the R is somewhat lower, the current is dramtically increased and the heat to be removed is more significant. I doubt that a shift to air cooling is taking place. The overall heat load is down but the heat per device is most likely a lot more.

Likewise, changes like this never take place in isolation. There has to be a lot of fundamental redesign of the entire power train to take full advantage of this. So while higher mpg results, it has to be through a combination of direct and indirect engineering changes of the entire power flow...and then other unrelated improvements may be involved.

 

Cool. I thought there was no way toyota can get that much of an efficiency improvement from these, losses just seem too low. Next generation mg's are supposed to change, that can have a much bigger impact, but 10% seems really big considering how efficient the hsd appears currently.

 

Minor correction (I think): Current density is increased, not current. I think that's what you meant to say.

For my part, I think it will be interesting to see if they can get away with air cooling. It is just a matter of surface area - as water cooling is air cooling, but with access to a much greater surface area with which to do it. SiC itself can tolerate much higher temps, so if they can address other temperature limited aspects (solder, for example, by using MCMs or Chip on Board) they just might. It would be fascinating to have direct insight into the cost/efficiency/marketing trades that are taking place and will continue to take place. It sure does point to why it makes a lot of sense for them to make their own electronics though. If they can get the answer right, it will pay off in a big way.

 

For power transistors, the "current density" is not what a power design engineer works with. It is the current going through the pins that is parameterized on the data sheets. Current density would be something a semiconductor designer would work with in designing the part. I was speaking from an end user of the part.

"Surface area" is exactly NOT the problem when working out cooling of SiC parts. This may seem counter intuitive, but the real nasty heating problem is at the actual die. The hurdle is attaching the small and finite surface area of the die in a manner to get the die cool. After that, transfering the heat to a secondary cooling surface can use conventional techniques. The surface area of the die is literally "set in stone" (considering a SiC part to be like a rock). So you have to get the heat out of the surface area of the die using exotic things such as diamond substrates or direct spray cooling on the die itself to take advantage of SiC power densities.

Now this is where Toyota making the die and all the cooling infrastructure as one component can really excel. On the Prius, they did a direct die mount on the aluminum die base/cooling plate and then packaged the transistors by submerging them in a clear transparent silicone compound. They got rid of all the "middlemen". They will do something similar for SiC, but will still have to get rid of a lot of heat in a small area with ultra high reliability once they get the heat out of the die. To this end, liquid cooling is superior.

 

In another article, Toyota said that electrical loss represents 25% of all loss in a hybrid. 5% overall improvement suggests at least 20% improvement in the SiC transistor.

The size and weight are reduced also. This may finally allow the 12v battery back under the hood.

 

Losses are system wide and they are cumulative. If you're lucky 25% of energy from a conventional ICE gets to the wheels. The rest is lost as heat through the radiator, engine block and the energy used to dissipate it (turning a water pump). In electrical/electronic circuits it's internal component resistance (the reason the batteries get hot) and resistance in physical connections. The forward voltage drop in a silicon device is resistance from the hysteresis effect of forcing it to change states. It adds up everywhere.