How to make power conversion less sucky A stealth startup company whose power-conversion tech could save billions through reductions in energy loss, has gone public, revealing why companies such as Google and Kleiner Perkins have invested heavily in its ambitious plans. Transphorm reduces the amount of power lost when energy is converted from one form to another. …
This seems a bit off. What is it they make exactly? A look at their website seems to indicate they build power supplies, but why is that worth all that investment? Are they designing and/or building the semiconductors as well?
MOSFETs made from gallium nitride instead of silicon are already available, if somewhat more expensive, and have similar characteristics to regular CMOS. But if they were used more, the prices would drop.
As an electrical engineer, I've built a number of power supplies over the years, and I know that switching losses in power transistors are a major part of total conversion losses, but reducing those losses mainly comes from better design of the driving circuits and filters.
While I'd love a better transistor, there is absolutely no reason a good engineer can't design their own power supply. They aren't that difficult, (I'd rate them about medium) and you don't need a company for that.
Or they might be a supplier for OEMs. That could be a good idea, but information on their website is thin.
The current generation of GaNFETs have particularly fragile gates. Just a volt or two over the rated limit for mere microseconds is enough to destroy them. The effect is cumulative and the working gate voltages are typically quite low, <6V anyway. Thus impedance in the gate drive circuit has to be extremely tightly controlled usually necessitating at least a 4 layer PCB - not conducive to mass producing cheap PSUs.
The other trouble with exotic substrates has been the inability to put more than a handful of interconnected transistors onto a chip. Maybe they solved this and thus integrated the gate driver.
"As hybrid and electric cars also convert electricity from AC to DC, Transphorm's tech has the potential to give EVs a longer range as well as smaller battery sizes"
Forgive me if I'm being thick but how does this help the battery size/range of pure electric cars? It might reduce the loss whilst charging from the mains but surely it wouldn't make any difference thereafter. It would though be true of hybrid cars as they generate AC and store DC.
This post has been deleted by its author
"As hybrid and electric cars also convert electricity from AC to DC, Transphorm's tech has the potential to give EVs a longer range as well as smaller battery sizes."
Please explain,bearing in mind that:
Batteries store DC. The car runs on switched DC. There is no AC to DC conversion of any significance once the car disconnects from its charger. Improving the efficiency of the charger neither increases the range nor decreases the size of the battery. It may slightly decrease the incoming mains energy needed to charge the battery, but HVDC (600V?) charging is already being discussed for this very reason.
switched DC looks a lot like AC, 'cos when you switch current on and off, it alternates between "flowing" and "not flowing". Yes, it's not sinusoidal, but that's what EMC filters are for.
And it's the switching that gives rise to the leakage currents. If they can reduce the leakage by 90% (so the loss goes from 10% to 1%), I think that's a reasonable saving.
What little they say doesn't say that anyway. They say "up to 90%". The scope for improved efficiency from GaN devices isn't worthless but it isn't that great either.
I suggest you look at "Efficient Power Conversion Corporation" who actually produce GaN devices rather eco green bollocks posturing to attract venture capital from fools.
Are you sure? Obviously the batteries store DC, but does the motor/generator really generate AC not DC? A link would be much appreciated?
Even if the motor-generator does generate AC which needs a bit more electronics than DC to control its use in recharging the battery, its seems unlikely this new miracle semiconductor will substantially impact electric vehicle performance.
Same goes for modern standard switched mode PSUs. They're quite efficient. Typically not 100% by any means (otherwise they wouldn't get hot) but the bigger ones, where efficiency matters more, already achieve over 90% for the better ones.
Don't yet see room for miracles. See need for enlightenment. His Holiness Saint Bill will lead the way to Enlightment.
"Transphorm reduces the amount of power lost when energy is converted from one form to another. "
"Transphorm says conversion loss accounts for 10 per cent of all energy in the US....."
Was the vulture asleep when this article passed the editor's desk? Methinks the author may be referring to *electrical* energy, otherwise I'd guess that the "conversion loss" could be 100% and it still wouldn't compare much to the energy striking every square metre of the US (during daylight hours) from solar radiation. And just how would this technology help gas, oil, coal and biomass "conversion losses"?
"Gallium nitride - ..... it's an extremely low-resistance material that simultaneously holds off large voltage."
And I'd be fascinated to know how this wonder material will help the "conversion loss" in the step-down transformers within a national electrical grid - can it really handle the peak-to-peak voltages in a 400kV line?
C'mon El Reg, your readership has a high proportion of nerds....
"Gallium nitride - ..... it's an extremely low-resistance material that simultaneously holds off large voltage."
Some info gleaned:
"Gallium nitride, on the other hand, is better [than silicon] at preventing leaks by holding onto the maximum voltage when it’s not delivering power"
Sounds like it will primarily be used to stop leak current.
As for handling peak-to-peak voltages in 400kV range:
"Transphorm’s first product will be in the 600-volt range and suitable for industrial operations such as data centers, solar panels and automotive drives, said Primit Parikh, president of Transphorm. The company is working on 900-volt designs, he added."
So no, the power grid wasn't the target of their product (yet).
Citations from:
http://gigaom.com/cleantech/transphorm-the-new-data-center-waste-power-slayer/
Try google. Great way to find Enlightenment.
This is potentially a very interesting topic - but the article has so many inaccuracies and holes you could drive a bus through that it's a bit of a waste. Others have highlighted many - the one that jumped out at me was the AC/DC switch improving electric car ranges.
Don't just copy the press release, go a bit beyond to find out details (or point out the holes). That's why I keep coming back here...
You have a choice: Efficient, High Power, or Cheap. (Pick any two). There would be a market for High Power 600V switches that also have high efficiency, low loss, low rds. Say 600V, 60A,6mOhm
A typical 600V 100A switch has a resistance around 100mOhm. If you run 100A through it, you start with 60KW and loose 10KW, around 15% loss. HIgh efficiency switches, around 1mOhm do exist, but typically can only block 40V -- not enough for direct connection to a grid-interactive solar array
The available choice for a high efficiency 600V switch is quite limited, and if they can make something good, they will find sales.
It's a big IF though - this is a very capital-intensive business, and if it was easy, someone else would be doing it already.
For those who don't do this as their day job.
There's a big world outside of computers where 5v supplies don't cut it.
As the voltage to be shut off rises the the number of device types and suppliers which can handle them falls.
Same with current.
Devices can be strung in series and parallel to meet voltage/current needs but the PSU is likely to be bulkier/heavier.
Lowering the on resistance and raising the off resistance are *very* good things. A *big* part of high power device design is the *packaging* of them to that heat out, typically without fans or liquid cooling.
Not all EV motors are DC. IIRC the GM Impact was AC. Historically AC motors have had simpler* construction but are tied to the AC frequency and changing that frequency has been *hard*.
The way to do "AC" from DC is to chop the DC into little chunks and then feed them into what is basically a high power low frequency (relatively) filter. The *ideal* circuit for this uses "switches" that switch in a nanosecond (look up the spec on a power device and see how realistic that is) with no power reflected back to the DC source, 0 on resistance and feed a cheap circuit whose output is a smooth sinusoid. It has 0 loss and no more RFI than a regular power circuit.
It's a total fantasy. The amazing thing is how *bad* the simulation of AC can be to con an AC motor into working.
*IDN if that is still the case as there has been a *lot* of DC motor development.
Clever materials have improved things before by introducing new board and packaging materials. It's not clear what Transphorms Special Sauce (C Lewis Page) is.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
