Re: Coupling coeff. vs Voltage gain (was Re: Who needs a quenching gap ?)

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "Antonio Carlos M. de Queiroz by way of Terry Fritz <twftesla-at-uswest-dot-net>" <acmq-at-compuland-dot-com.br>

Tesla list wrote:
> 
> Original poster: "Marco Denicolai by way of Terry Fritz
<twftesla-at-uswest-dot-net>" <Marco.Denicolai-at-tellabs.fi>

> >The currents cross zero. The primary voltage touches zero and increases
> >again, without a polarity reversal.
> 
> Sorry, but in my (MicroSim) simulations do primary voltage DO reverse its
> polarity. Why /how it shouldn't?!

Try this:
L1=0.1 mH;  C1=10 nF
L2=100 mH;  C2=10 pF
k=0.6
Vc1(0)=10 kV

Vc1 initially reverts polarity, reaching -5.6 kV and climbs back to 
zero in 4 uS, touching zero only.
Vc2 goes to 180 kV, and then reverts to the peak of -316 kV, at 4 uS.
Il1 goes to 108 A, reverts to -52 A, and croses 0 at 4 uS.
Il2 goes to -1.7 A, reverts to 3.4 A, and crosses 0 at 4 uS.
 
See the comment in other post about possible inaccuracies in Microsim
(Spice) simulations. Numerical integrations always introduce artifacts,
and these appear enhanced just when lossless linear systems are
simulated.

> So, if I understood well, the instant when both CURRENTS are zero is the
total
> energy transfer instant, correct?

Yes, but this is not enough. With zero current in both inductors the
voltages on both capacitors are at peaks (dV/dt=0). But for complete 
energy transfer the primary voltage must go to zero too. In the case 
mentioned in that paper, it doesn't go to zero.
 
> >Something strange in this. The maximum gain can't exceed sqrt(L2/L1),
> >with the two untuned frequencies identical.
> 
> Right, but here we consider f1 <> f2

But the gain would be higher if they were equal...
 
> But now calculate again sqrt(L2/L1)=sqrt(100mH/0.185mH)=23.25 (you did change
> L1)
> 
> Note that now 27.4 = 1.18*23.25 !! Just as the paper said.

Of course, as the system is as the paper describes. But notice that if
the capacitors weren't changed, and so the amount of energy fed into
the system wasn't changed, you can obtain larger gain with the original
system with f1=f2 and k=0.6.
 
> The paper doesn't assume you are varying C1, L1 or other: it simply tells
that
> for certain combinations of f1, f2 and k (that is L1, C1, L2, C2, k) you can
> break the gain rule of sqrt(L2/L1) and get 1.18 times that much.

I see that it says this. It's a mistaken interpretation. The problem 
is that the true maximum gain rule is not this, but sqrt(C1/C2), as 
the initial energy was in C1, and ends all in C2. This rule is
independent of L1, L2, and k, and is only reached in the usual
conditions of f1=f2 and k in the families that I mentioned.
 
> You can find a zipped version of it at
> 
> http://www.saunalahti.fi/dncmrc/th_ccoef.htm (check the last link down)
> 
> or
> 
> http://www.saunalahti.fi/dncmrc/phung.zip
> 
> The quality of the scan is pretty bad: there are also some typing errors (at
> least in one place "1" is confused with "T").
> Beware of that. I have also got a postscript version of it (about 800K
zipped)
> which is much more readable. If you can
> open/print postscript let me know and I'll email to you directly it.

The paper is readable. Do you have at hand the reference [5]?

Antonio Carlos M. de Queiroz