Radiohead said:
I've just started to learn about matrix converters as part of my
university course, and I've been asked to decide which is the best
device to use in a matrix converter out of MOSFET, IGBT, Thyristor.
It looks to me like MOSFETs & IGBTs are the best but I'm not sure which
is the right answer. I'm thinking it might depend on the voltage &
current, and so there is no answer that is correct for all cases. Is
this right?
If any one has any experience of this I would appreciate your coments.
Many thanks.
I have yet to see a real product made with matrix converters. Prior to
the late 1990s, nobody in their right mind use FETs above about 600V
because Rdson was proportional to BVdss^2.6. That all changed (thanx
infineon, IR) but even so, its hard to get a decent 1000V FET, and even
then they are puny. It is trivial to obtain a 1kA 3.3kV IGBT, (or 2kA
1.2kV) which is puny compared to the thyristors you can buy - 6kV, 9kA
in a hockey puck isnt exactly huge. You should see the stacks of these
things used in MV MVA (yes, megavolt, megawatt) inverters and
rectifiers. The great thing about thyristors is (if you control the
energy) they fail short-circuit (if you dont control the energy they can
and do vaporize). So big inverters have many thyristors in series
(literally stacked one atop another), with many more than are required
to meet the voltage spec. That way, multiple devices can fail and the
inverter still works, and can be repaired at a convenient time. in NZ we
have a DC link between the south (lots of hydro power here) and north
(lots of consumers) islands; if the inverter at haywards blows up, 2/3
of the population pretty much goes without electricity until it works
again (gross oversimplification)
Clearly power level is important - at 6W 300V it would not be clever to
use IGBTs, but then a matrix converter would be a pretty stupid idea
anyway. GTO thyristors allow you to turn the darned things off, unlike
SCRs.
Matrix converters have the disadvantage of no DC bus (its their
advantage too), making it much more difficult to prevent inductive
overshoot - which kills IGBTs very quickly. This generally requires a
lot of snubbing, increasing size, cost, power dissipation etc. And of
course the switches need to be bidirectional, so there are a LOT of them.
Cheers
Terry
