Ok,
I have scribbled out a quick circuit that should be right....
The oscillator can be anything you chose if you don't want to wind a feedback style transformer, but I find it is bullet proof and rugged., mine use larger transistors, (up to tip31 etc) and can draw an amp or so.... it is completely flexible in this circuit.... it just has to oscillate, and make the current you need for the storage you want in the time constant you choose.... I generally go for 1 second pulses... but more is fine, less may be illegal for your country (it is here).
The 640R and the 2k2 can be changed to suit your current requirements, as can the npn transistor. For your project it could be a TO92 size, drawing only a few ma to an amp or so.
The 50uf cap is about the smallest I use for larger fences, but for your size, even 2-4 uf will be fine, and this will allow your current from the oscillator to be small, (TO92 and higher values for the resistor bridge etc.)
Note the start and finish of the windings in this transformer..... otherwise it won't oscillate. (the dots are the starts).
We then rectify the secondary, and it drives the capacitor. It should be that it takes about a second to drive the caps voltage up to around 500-600 volts.
We then use the 1m 1m 1m 330k divider to get a sample of the voltage, the variable resistor is just to get the timing better, and the finish trigger voltage more precise, but is unnecessary really. The .7uf capacitor is just to load the divider, and so also controls the timing. If this is too big, the real (big) cap voltage will possibly overshoot, as the voltage is not a true reflection of whats happening to the main cap, it is slightly retarded, but stops false triggers.
When the voltage across the .7uf gets to the breakdown of the diac, it triggers the triac/s.
The triacs short the storage capacitor across the transformer primary, which for me is about 9 turn 2 in hand of 1.8mm wire. This presents a very heavy low impedance load on the cap and triacs. It is this impedance that will dictate the pulse width and shape. I use about 20 turns around a ferrite, or whatever takes your fancy to impede this current flow some, and so widen the pulse. This helps in transmission over long distances, as the spike is not so high, and a bit broader, it also suppresses RF Interferance... which over here is no bother.
It also helps keep the avalanche currents down a bit in the triacs. I used to use very expensive scr or triacs of the 50 to 100 amp range at 1200-1500volts, but found a bank of 20 cent traics of the 800v 16 amp variety to be every bit as good..... if you have a bit of inductance in the line ( the 20 turns etc).
This impedance lowers the output voltage by a good deal, but the two 2uf capacitors across the transformer fixes this up (smaller for you), and actually improves it a touch.
The diodes on the tranny are just to stop the back emf from the coil, which is necessarily savage. They need to have a 1000v rating and good surge current for big systems, and not that exciting for smaller units.
The main output transformer for me is a microwave rewound, you may find for your small fencer a normal power transformer with 110v:10v windings will do.. if it can be stopped from internally breaking down...other wise wind your own around any core you can get hold of.
In truth, you can wind it around just the "I" of a big transformer... ie an open inductor, and it will work just as well. The transformers are driven so deeply into saturation, that most of the core is seen as air gap anyway
.... so using air instead of the rest of the "E"'s will make very little difference in practice.
Questions or comments?
...............oztules