Electrical igniter for gas engines A keystone to understanding by Magluvin

[forest]

deleted

[delboy]

1. How long lasts capacitor discharge into low selfinduction coil ? I mean HF oscillations

Let it be:
Lp - primary inductivity [H]
Cp- primary cap. [F]
Rp - active resistance of primary (switch + primary) !
Assuming that voltage level of primary cap droped to 5% of its initial value meaning transient proccess finished, settling time you can calculate by formula : Ts=6*Lp/Rp
Here are examples for primary:
1. GOOD ONE
Rp=0.02 ohm
Lp=10 uH
Cp= 1 uF
----------------
Resonant frequency will be approx 50kHz ant settling time Ts=0.003 s =3ms
2. BAD ONE
R=1 ohm
Lp=0.1 H
Cp=1uF
---------------
Resonant frequency will be approx 503Hz ant settling time Ts=0.6 s
(you wouldn't discharge cap running switch even low as 10Hz)
Now you decide is it better to discharge primary cap into primary coil with core or without (aircore), I mean does it Lp have to be big or small one?
Beside Lp you can manipulate also with Rp to get better results. That is where problems with switch come into play, because primary cap is discharging over switch to primary Lp. You will have small resistance of Lp by using large copper tubes but if your switch have high resistance on time when it make contact you will get bad results! That is why Tesla abandoned spark gaps in first place !
I forgot to say that formula Ts=6*Lp/Rp is correct only if there are oscillations meaning that must be Rp<2*sqrt(Lp/Cp)

[delboy]

One more thing about primary. When you choose Lp and Cp you are automaticaly selected max primary current Ip!  Here is formula for first peak of primary current :

Ip=Uc/sqrt(Lp/Cp)

Let Uc (voltage on primary cap) be Uc=1kV , then for first example max current will be:
Ipmax=Uc/sqrt(Lp/Cp)=316 A
and for second example max current will be :
Ipmax=Uc/sqrt(Lp/Cp)=3,16 A
You see, in first example current will be 100 times bigger!

[bboj]

Del. So the res. f. is for secondary and we are switching the Cp faster than it can deplete?

[forest]

I realized that I might have a nice solution back in 2006 year 
What I have done was very simple, accidentally done. I put a very fast recovery bidirectional diode (~1ps dumping speed) called transil and a varistor across MOSFET drain and source because I wanted to protect mosfet from damage.
I wanted to create plasma globe using car ignition coil and text suggested to put quite large radiator on FET to let dissipate heat. My FET had run freeze cold without heatsink at all.
I think what I have done was the separation of switching low voltage high current from higher voltage low current from flyback , even if they flow in the same path.And when you have it done some magic can be done in the same circuit.