[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
OLTCs, DRSSTCs and untuned primary SSTCs
Original poster: "jimmy hynes by way of Terry Fritz <teslalist-at-qwest-dot-net>" <chunkyboy86-at-yahoo-dot-com>
Hi again,
I have been meaning to get this email out for a while now, so here it is.
I did some more thinking about "M", and I realized my definition for M is
also flawed.
It is possible to get a good M for really short, high current pulses. The
voltage drop would stay around 5-6 volts, but more IGBTs are needed, so it
isn't as good. If we are trying to find a number for (energy to secondary/
number of IGBTs), assuming heating is the limit, then big coils would have
an advantage, because the lower freq. and longer drive time reduce the I^2
term. Because of difference in coil size, this wouldn't be a fair
comparison of OLTCs, DRSSTCs, and untuned primary SSTCs. To "fix" that
problem, I guess we could design coils on pspice that all have identical
energy transfer. I "think" this could be our equation:
energy/(rmscurrent*time^0.5)
I'll attempt to explain my derivation
E=energy transferred, R=resistance, I=current, T=time, N=number of IGBTs
First of all, since this isn't meant as guide as to how many IGBTs you
need, proportional to is as good as equal to, so I use = for both.
First we start with M=E/N, since the limitation is the heat in the IGBT, it
is the same as M=E/(I^2*T*R). Now we have the problem to figure out what
"R" is. If M=E/N, and R = 1/N then R*E = M. Therefore R = M/E. If I
substitute that in, I get M=E/(I^2*T*(M/E)), or M^2=E^2/(I^2*T). Finally, I
get to M=(E^2/(I^2*T))^0.5, which can be simplified to
E/(I*T^0.5)
Of course, you could still cheat by using a really long pulse, but I can't
think of how to penalize for that. There isn't enough info about frequency,
rise time, and envelope shape on streamers. "Normal" SSTCs have shown that
shorter pulses are better, but it isn't clear how short it needs to be.
Aside from losses, I dont think there will be a big difference if you take
2 or 4 cycles to ring up. It should be an easy test with an OLTC. All that
you would have to do is change k and slightly change the input voltage to
account for losses. If you could do this test with your mini OLTC, I would
be very interested in the results.
There is also a 0.5 factor for H-bridge coils. H-bridges have four times as
many switches, but they are only sucking up heat half the time, so the
multiplier is 1/2.
I have decided to use matlab to do a more accurate analysis of it, with
more detailed IGBT models (forward drop, and resistance). It turns out that
switching loss is pretty insignificant. If I use the datasheet's values for
switching loss, and scale it up linearly with current, then even when hard
switching, the loss is only 1.6%. I have been spending all of my time on
matlab for a while so no progress on the DRSSTC. I have some plots for
everyone, they're on my website
<http://www.hot-streamer-dot-com/chunkyboy86>www.hot-streamer-dot-com/chunkyboy86.
Jimmy