Varistors are fine, what sparks are describing is also fine.That's how we return back into parametric resonance oscillators 

Hi Wattsup,

Well, I had thought of them as expensive components.  (And if you order less than 10 qty, the price may go up even higher for a single unit.)

I think you can "make" a normally open FET from the cheaper normally ON  SJDP120R085 type by using a ground independent voltage source like a battery (either a dry or a rechargeable).  Because you can bias the gate-source with the battery voltage by connecting it across the gate-source via a series resistor, with the appropiate polarity.

I think this as good for mainly test purposes, it will behave as if it were a normally open type and you have to control its gate-source with an input voltage that takes the FET to the ON state, whenever your input pulse defeats the battery voltage bias.  Very little current is taken from the battery.  The simplest control would be via an audio or a pulse transformer but a capacitive coupling can also be considered. In fact a 1:1 to 1:2 or max 1:3 transformer would be useful, working in the frequency range you would wish to operate the JFET switch, this may involve a wideband transformer which again may be expensive, so the capacitive coupling is more attractive.
If you need to see such a schematics, I can draw it if you wish.

rgds, Gyula

Today I spoke with SemiSouth and ordered two of each. Those prices are unfortunately their retail price for quantities under 100. Above 100 u nits and price drops 40% and above 1000 units price then drop over 50%. But still it is expensive.

They told me the 63 model is their best seller but would not indicate the type of devices using it. For sure these must be expensive circuits if only one component can cost so much. Must be because they work.

@gyulasun

Thanks for your information. SemiSouth said they will ship ASAP so I think for now I will wait till these arrive for initial testing.

Incredible that these JFETs are used in audio. Must be high end audio systems. SM was an audio guy also and he did express the fact that his audio produced pure signals, so this just may be the way to go.

I will start very carefully because the prices for these are rather extreme. I will surely need your help when the time comes to make sure I set these up correctly for testing.

I just got notice that my order for a good 200 standard JFETs has arrived at the post office. I will use these in parallel and do some testing while I wait for the SemiSouth JFETs.

My main way of pulsing is using my FG which is a HP 8111A. It goes up to 22mhz, 16.5 volts but very very low amperage which is a good thing.  A copy of the manual is available on my ftp OU site located here;
http://purco.qc.ca/ftp/Equipment%20Manuals/

The FG goes to the gate but I then use my power supply on the source/drain line to pulse the required power into coils, etc.

So I can't really put the positive of the FG on the source and the negative in the gate, since the positive is connected to my power supply. Man why is EE so complicated. Can't it work like regular plumbing. lol Just jokin.

Thanks again for looking out for these JFETs.

wattsup

...
I will start very carefully because the prices for these are rather extreme. I will surely need your help when the time comes to make sure I set these up correctly for testing.

The most important thing to see perhaps is power dissipation in the switch (i.e. in the JFET, MOSFET etc):  if you have a coil you place in series with the drain (or source) electrode of a FET, you HAVE TO consider the possible maximum current drawn from your power supply when the FET is switched ON. 
How can you figure out the maximum current? 
First by simple Ohm's Law.  You have a coil with ,say, 1 Ohm DC wire resistance and you use ,say, 12V DC supply. When your FET is ON, and suppose it has a 0.1 Ohm ON resistance, the current will be 12V/(1+0.1)=10.9 Amper. (I neglected your power supply inner resistance which also in series with the coil and the FET and I considered it as zero.)

Now what power dissipates in this FET? Still using a static state (you do not pulse the input yet and the FET is ON continuously) the voltage drop between its drain and source electrodes is 10.9A*0.1Ohm=1.09V and the heat dissipation in the FET will be 1.09²/0.1=0.118W This 118mW will heat the FET, no real need for a heat sink to the FET body in case of a TO220 or similar casing.
Of course when you use AC pulse control for the FET switch, the current flowing via the coil (and the FET) will depend on the duty cycle and the frequency so power dissipation surely decreases with respect to the above static case, I used the example to show that the FET you happen to use must be able to handle the 11 Amper in the given circuit parameters.

Now you understand that in case your FET has a 10 Ohm ON resistance instead of the 0.1 Ohm, then the maximum current will be 12/(10+1)=1.09 Amper, using the same 1 Ohm coil and the 12V supply voltage. But the FET dissipation INCREASES because the voltage drop across the drain-source will be 1.09A*10Ohm=10.9V so the dissipation in the FET will be 10.9²/10=11.88W  now this is a serious heat that would quickly make the FET tost soon, a decent heat sink is strongly needed. BUT the best to avoid using a FET switch with 10 Ohm ON resistance in this example circuit I used with the 1 Ohm coil.

And now, after the current handling consideration, there comes the drain-source maximum voltage consideration for the FET. Because the coil is in series with the switch, the switch-off instant imposes the voltage spike across the switch so you have to use at least a 300-400V FET or even higher, not ruin it at the first switch-ON.
The induced voltage when the current is switched off in a coil depends on how suddenly you switched it OFF, how much current was flowing in the coil and what self inductance the coil has. V_i=L*dI/dt  where V_i is the induced voltage, L is the coil inductance in Henry, I is the current change and t is the time under which the switch-off happens. 
If you have L=100uH I=1A t=1us then V_i=0.0001*(1/0.000001)=100V spike.  If you have a faster switch with ,say, 100ns OFF time, the spike willl be 1000V!  AND YOU WILL HAVE a fast switch because your HP generator and either a MOSFET or the JFET is able to switch ON or OFF around 100ns or even quicker.

Now it is obvious you may wish to start any such switching test with the smallest power supply voltage possible, say, starting with maybe 2-3V DC instead of the 12V or whatever, to keep the current first at a low value, thus everything within safe parameters.

I just got notice that my order for a good 200 standard JFETs has arrived at the post office. I will use these in parallel and do some testing while I wait for the SemiSouth JFETs.

Now you may have some info how to NOT toast them.  When a lot of them is in parallel, maybe it is then not the current that ruins them but their 25-30V drain-source maximum voltage... 
Though you have to consider current too because if 10 of them in parallel have ,say, the 10 Ohm ON resistance and you use them in the above example with the 1 Ohm coil, then the 1 Amper current from the 12V supply will drive 100mA through in each, (10 times 100mA=1A) and these small signal jfets cannot handle 100mA but maybe 30-40mA maximum each, depends on their type of course.

...
My main way of pulsing is using my FG which is a HP 8111A. It goes up to 22mhz, 16.5 volts but very very low amperage which is a good thing.  A copy of the manual is available on my ftp OU site located here;
http://purco.qc.ca/ftp/Equipment%20Manuals/

The FG goes to the gate but I then use my power supply on the source/drain line to pulse the required power into coils, etc.

So I can't really put the positive of the FG on the source and the negative in the gate, since the positive is connected to my power supply. Man why is EE so complicated. Can't it work like regular plumbing. lol Just jokin.
...

You have a very useful function generator indeed. It can give out negative or positive or symmetrical polarity waveforms which are very useful. For your N channel JFETs the negative polarity is the one needed because all the waveform is UNDER the zero voltage line.
In case of the FG the output voltage is defined with respect to the BNC socket 'outer metal ring', let's name its metal cylinder as the zero voltage point. And the BNC socket's middle point in the center of the cylinder metal body gives out the polarity (and amplitude) with respect to the zero point.
I mention these so that you can think of them correctly when you mention polarity in case of a power supply. What you sound to call positive at the FG, it is the FG's BNC output metal cylinder, the zero voltage or ground level point, ok?  And you can connect it to the source electrode of the JFET (or MOSFET), no problem if your power supply's negative polarity is also connected to it.  And the FG's center pin of its BNC output can go directly to the gate electrode of the JFET (or MOSFET) because the polarity of this pin can be chosen from the FG's front panel button, Output Mode Selection shown in Page 41 in the PDF file  ( http://purco.qc.ca/ftp/Equipment%20Manuals/hp-8111a/08111-90002.pdf ).  (The Complement mode simply means the 'phase' of the output pulse is changed: where there was zero voltage, it changed to max amplitude and vice versa.)

And your power supply's positive output can go to the drain electrode of the FET or MOSFET (via a coil), its polarity has nothing to do with the FG's output polarity, you know that.

rgds,  Gyula

Wattsup,  did you mean this schematic here from Groundloop?
http://www.overunity.com/index.php?topic=8185.msg240769#msg240769

and you use two NPN bipolar transistors instead of the two MOSFETs? If so, then some modification may be needed in the schematic perhaps?

Just for me to understand your description better, with the least puzzle, ok?
I am not fully in your picture I am afraid...

Gyula

...
Also, I want to take some time and test the Tesla Ozone Patent by using 4 or more of these PN44116A's JFETs in parallel. This means I have to make a short circuit via a high impedance coil going to the negative but just do it with a very small voltage to start and I will report the results. Lucky I have 100 of these JFETs. lol

The most important thing for me is I now have a JFET that can open and close very fast and that does not have an internal diode.

Thanks, ok.
However, you cannot make that short circuit with PN4416A jfets because they have typical ON resistance (at zero gate-source voltage) of about 1-2kOhm each. And if you parallel ,say, 5 of them, you would still have between 200-400 Ohm ON resistance between the drain-source electrodes, far from a short. I repeat this just for to be aware of. 
And as soon as the voltage difference between the drain-source exceeds about 35V (DC or peak to peak AC), the 5 jfets will probably be ruined.
So just be careful.

rgds, Gyula

Something is weird with LC circuits. If I have properly constructed mechanical pendulum in action it can run many minutes,but LC circuit without power stops immediately. Looks like power source and mosfets are the case why this happens. I need to think how to use fets only to overcome resistance of circuit and use power source only to make initial push to the circuit.

Yes,  Power source can "ruin" the resonant Q of an LC circuit because a decent power source (usually a battery or a mains power supply) has a very very low output impedance. MOSFETs (or bipolar transistors or jfets etc) used as a pure switch also can only add loss to  LC circuits.
Here is a Java applet for resonant LC circuits where you can adjust L, C and the series equivalent loss resistance of the parallel LC tank circuit:
http://www.walter-fendt.de/ph14e/osccirc.htm 

If you use the 500uF, 5H values and zero series resistor value (a loss-less LC circuit) you get undamped oscillations. If you introduce any small higher than zero loss resistor, then you get damped oscillations which eventually reduce to zero (input energy dissipates in the loss resistor such as like wire resistance, core loss if any and capacitor dielectric loss).

So if you use different value loss resistors, you can compare the 'ringing' time for the different loss values, and maybe compare them to the mechanical pendulum's running time.

If you want to use a MOSFET (or any other active device) to overcome the loss of the LC circuit, then you have to build it into a circuit which makes the active device to 'simulate'  ,say, a 'negative' resistor. 
This can be achieved by building an oscillator around the LC circuit from the active device or bias it like for instance Naudin shows here:
http://jlnlabs.online.fr/cnr/negosc.htm 
However, to do so involves using an energy source to maintain the oscillations, unfortunately. So this way of making and maintaining oscillations in LC circuits does not seem to be OU, unfortunately.
(The best oscillator shown in this Forum consumed 3-4uA current from a 3V battery at 5-10kHz frequency and used a MOSFET, it was built recently by Luc (gotoluc).
(I think there are possibilities in parametric circuits.)

rgds,  Gyula