Hi Srikanth,
This is a grubby little circuit i made up one afternoon about 8 years or more ago... so it has stood the test of time now in many variants.
I finally had some time to take some measurements for you..... I'm sorry they are no where near your simulation... if your interested in this stuff, you are well advised to build one and fiddle with it, the spice is nowhere near the real world results....... where to start.....
First I guess is the little 1n4007.... no, the circuit uses only a max peak measurement of 46ma using a tip35c, and 30ma using a MJE13007 as the oscillator transistor. The other thing of note is that the 1n4007 has a max peak surge current of 30 amps... yes that says 30... check your data sheet, they are rated as 1amp continuous peak, 30 amps surge for short pulse duration, and any where in between.
EDIT: reread this.... will check again, but I think I meant 460ma and 300ma... will check later today and get back. Will leave as is until I check it out... it makes sense too, as say .35amps x 12v=4.2w.... for 1.2seconds its about 5 joules used up... inverter is pathetic, so 2.8 joules into the cap is about right. (600x600x.000016)/2=2.88joules or for 20 uf 3.6 joules.. last night it was using 19uf@600v
Edit 2: Couldn't help myself.. went and measured it three ways.
First the DVM in series with the leads...MJE13007 was .34a max reading sampling rate was poor....
second with dvm clamp meter and peak hold.... got it up to .56a on one occasion.
third with 1R resistor in line and oscilloscope... this was different.
The wave was a surprisingly clean saw tooth... maybe 30us long. It reliably peaked at 1.2amps... so... one for the simulation... closer with the scope.
The tip35c ( mostly don't use these now... have hundreds of MJE13007... so use them presently) seemed slightly more maybe 1.3-1.4a.
Average was around the 300ma range. So you could justify the better diode... but I will continue to use the 1n4007 as they have not blown ever in these units..... collectively over 100 years I guess ( across a number of units that is). Thats reliable enough for me.... and the worst that will happen is you will loose the reverse connect protection, the unit will still function normally ( .5 volts better actually).
Transformers:
.4mh for the primary of the etd49 ( 3C80 ferrite core)
.3 henry for the secondary...... don't know if that helps you out.
I fail to see why you would introduce another complicating factor like voltage limit for the caps.. that is taken care of by the trigger.
There is flexibility for setting the time constant and the voltage... it is a simple compromise if you don't want to change components, or if you want to tune it to a specific voltage and specific time constant, then that is a once only thing for each unit.... remembering that the voltage will swing through the day from the solar inputs.... 14.5 in the day, 12v at night.. so needs to be timed for max voltage anticipated for >1.2sec. Simple and less to fail.... change cap and pot setting... simple.
In a rural setting, you don't add sensitive components for fun, bugs and spiders and condensation will get it sooner or later.
However, if you wanted to do something useful with the voltage of the caps, then take it a step further, and use a sense of the output voltage on the fence, and then charge the cap to deliver a certain voltage eg 6000v under all conditions... this would save power on a clean fence, and keep the voltage up better on a heavy fence.... that would be a complication with an advantage that is worth the extra parts.
Main transformer:
Simply... keep the impedance to a minimum... thick short windings will keep the time constant down.... like 2 in hand 2.2mm wire.
I tested three inductors in series with the primary for you to consider.
1 an air coil of 50 turns,
2. ferrite of some sort with 20 turns torroid
3. filter from a computer supply.... maybe 20 turns around a ferrite inner
All resulted in a time of less then 10 us. Your simulation was 400us.
The air core was in the 5us area, the torroid about 8us, and the computer core gave about 6us
Sadly it is not perfect, there is considerable ringing on the spike itself on the oscilloscope.... but dissipates in less then 5-10 us dependent on the inductor used.
This is about what I expected, as if 2-3 joules lasted 400us... then the laws of the universe have been battered to death.
These things( 240v trippler units) I have seen run at 40 amps@7000 and more volts... thats 280000 watts.... and thats output..... probably only 30-40% efficient or less. That was using about 8 joules from memory.. very very roughly, thats 8/800000 or about 10 us... which is in the ball park. Tonight I saw 6500v@13amps... thats 84000 watts for about 3joules.. pencil in the massive losses in the transformers etc, and probably we would see near 240000 watts or about .000012 seconds... so it looks like the losses may be even worse than I suspected.
Remember also, the time constant will change with loading. Those figures above are for a fence with 500ohm loading.. standard measurement point for these things. If you load it with a very low impedance load.. eg plasma short.. then the time will increase.... markedly.
I don't know how your simulation sees the resistance in the inductors and circuit paths, but time of discharge is very very important to keep the volts up and the time down. Transformer resistance is key to short time periods, as is the lead thickness to the caps etc etc.... keep the impedance down to absolute minimum, and the losses will be a bit less too.
I like how uncomplicated this thing is, and it is well inside the guidelines re time, and timing ... well timing is user settable, and juggling the caps and the pot will do it, or change transistors works too SC4242 MJE13007... and anything else in the junk box will have different effects on the timing and the power draw.
In winter, power draw becomes much more problematic, as the sun hours are so much less intense and duration.
Have fun doing whatever it is that your doing.... but this thing does not simulate very well, as the losses are tremendous, the oscillator is raw and so full of harmonics it is a mess, and the currents are absolutely huge.. burning up a few joules in micro seconds is very lossy.
If your going to run it with no inductor, then you don't need the 4uf on the transformer, and your time will shorten even more and voltage will be crazy without a load ( 14000 volts and more I have seen with no inductor)..... also the shock in the silicon in the triacs seems to be severe in the extreme... and life expectance is reduced accordingly...... dont be doing it unless you have very strong triacs/scrs.
Thanks for the effort.. but not for me..... I may do the variable power part using an arduino perhaps to keep the fence voltage as constant as possible regardless of impedance changing.... dew, rain, grass etc etc etc... but then I would also ditch the oscillator, and go pwm then.... light display for fence power and battery voltage......... once you have a micro on board,..........it just gets to easy to do everything.
Odd how the simulation got no where near anything, in fact orders of magnitude wrong...... I guess thats armchair electronics exposed... you need to see results in the real world. simulation of wildly erratic circuits like this is not their cup of tea by the looks of it.
The little inverter easily gets to 600v in 1.2secs with well lless than the power estimated by your simulation.... not 3 sec @ 3 amps
The time constants for discharge and wave form even with inductors is magnitudes less than calculated... 5us is easy to achieve, and with bigger inductors still hit 10us... 10 times faster than we need from the looks of it.
So it just works... who knew???
I would just build it as is... it works, it cannot ever run the caps past your set point, so your solving problems that don't exist.... and I have never heard of any one wanting to turn the fence down....never.... they all want it turned up.
Stop your unit with a NO relay controlled by your sms... that way failure will terminate the unit.
...............oztules
