advertisement


Gallium Nitride (GaN) Power FETS

russel

russel

./_dazed_and_confused
My ears picked up today when a PSU manufacturer mentioned to me they were working with Gallium Nitride Power FETS, I had not heard of them being used in power supplies before, thought they are used just in RF, before but it looks like they are on the way. Class D amplifiers with megahertz switching speeds, here we come!

http://www.irf.com/product-info/ganpowir/GaNPCIM.pdf
 
Looks very interesting, the Gate charges seem much lower - well the all important G-D Miller capacitance seems much lower judging by Toff delay times.... Maybe even the Gate channel has lower resistance....

That said you can see that the Vgs On is quite high - which helps in switching apps -as you dont have to reduce the Gate voltage so far before the device starts to switch off...

Guess these devices are there Direct FET packages judging by the low ringing levels...
 
Low Vth is not always an advantage in PWM switching applications where stray circuit inductances can resulting in unwanted and destructive "parasitic" enhancement of the device.

In the past I helped my father design and manufacture fS pulse generators for Laser and Xray streak cameras with frame times of under 50pS / multi shot - with the fast edges we require packaged die are useless.... we bonded onto the die directly... Low Vth are not the order of the day with these kinds of edge speeds....

http://www.kentech.co.uk/index.html?/&2
 
All of the big boys are actively researching GaN for power - trying to get the switching speed right up for that nth degree of efficiency and keeping the bill of materials low. I would expect a torrent of releases over the next months.

Jason
 
Making the thing must be expensive. GaN is much harder to make into pure crystals than silicon and as there is no equivalent of the self aligning oxide layers, the yield will be poor
 
I don't think they use bulk GaN, but heteroepitaxial layers on sapphire or silicon.

A big advantage is that it is very thermally stable, so you can run the devices very hot, and so increase power handling from a given chip size.
 
Wonder if they will ever produce a Pch device... trying to cleanly level shift is such a pain in full bridge applications.
 
See http://spie.org/x87883.xml

Basically the only acceptor that works is Magnesium, and that has a large activation energy, so the doping becomes very temperature dependent.

I wonder if remote doping in a heterolayer would work, analogous to the way HEMTs are made?

So I guess the question would be are any of the other nitrides any easier to P dope?
 
So worst then Silicon Mosfets? or about the same - IIRC you see about a 2 to 2.5 worst RDS on etc

Edit - just read the link, sadly much worst ATM :(
 
You must log in or register to reply here.


advertisement


Back
Top