wattsup, sounds like some interesting results.  It's somewhat hard to tell what's going on voltage wise even with the AC setting if the frequency is too high for the DVM and most of them don't go that high.  Did you look at this with your scope?  It looks though like you have an effect to dig deeper into.  Good work!  I'll have to leave it to some of the smarter ones here though for a possible explanation.

...
I put the positive of my power supply to the drain of an IRF9540 PNP mosfet and sent the source to one side of the primary. The other end of the primary went to the source of an IRF840 NPN mosfet and the drain went to the power supply ground. Both gates were in parallel and connected to only the positive of my pulse generator. The secondary which was also either in bucking or @gotolucs mode were connected to my capacitor tank via a 1N5817 high speed diode. I put my LED bank on the cap and my volt meter.

I put the power supply at 5 volts and set the pulse generator to 16.5 volts. When I got to a resonance of around 3mhz, the LED lit up supper brightly, the amps on the power supply read like .01 to 0. The FG amps was again micro amps. But the DC voltage meter on the cap tank read 0 volts. WTF is going on here. I was totally puzzled as to why it read 0 volts since this is the first time that has happened. Then I put the capacitor volt meter on AC and it read 9 volts.

So the toroid primary pulsed from both ends is producing strictly AC. Enough so that I changed the LED for a small 130 volt 11 watts rated bulb and the filament lit up semi-bright, something I could never do before. I wonder if this is enough to light up a vacuum tube cathode. I think so.........

I was always wondering what SM said about the AC in his devices. Where the hell could an AC come from in a DC pulsed device. Now I see it could come from the center toroid in bucking mode when pulsed from both ends. I cannot explain technically why it is doing that but it does. I tried at lower frequencies and to my amazement it still worked from 2khz to 10khz as if it had a large bandwidth there that kept the effect going.

I will have to document this more precisely during the week.

Hi wattsup,

If I understand correctly your circuit description, I would suggest to replace the p channel MOSFET drain with its source electrode ( its source electrode ought to connect to supply positive) and also replace the n channel MOSFET drain with its source electrode (its source electrode ought to connect to the supply negative), ok?  And repeat the test.  Maybe you will find the same behavior as before but then the FETs would receive correct DC polarities.  Please double check your setup with respect to your above circuit description, I can reflect on only what you describe.
(A p channel MOSFET must get negative supply voltage at its drain electrode with respect to its source, an n channel must get positive voltage  at its drain with respect to its source.) 

If you achieved resonance in the secondary then you must have some form of AC voltages in that LC circuit developed, don't you? (At resonance, an LC circuit always tries oscillate in sinusoidal form, even when excited with pulses;   forest also refers to it.)

Did you connect the pulse gen negative to the supply negative too?

rgds,  Gyula

Yes, I also thought of this schematic, after his text description.

I tried to correct as I think these P or N channel MOSFETs prefer getting DC polarities across their drain and source electrodes. Maybe they work in this particular setup with their drain-source electrodes mixed up, I never used them in reverse.  But if wattsup indeed mixed the electrodes up then the FETs body diodes were immediately get forward biased, when he switches on the 5V supply and those diodes effectively would short the drain source electrodes, regardless of the gate control voltage, so I am puzzled how they could have been controlled from the FG.  This is why I wrote, I can reflect on what I read as a circuit description.

Respectfully,  Gyula

....
I had described my connections correctly and thank you for your explanation which I followed also to realize the way I connected the PNP was in reverse so actually, when I removed the pulse to the PNP gate, it did not change anything, yes because of the internal diode. So now I connected them as you have recommended. Also, before I was sending the FG positive to both gates, now I am sending the FG positive to the NPN gate and the FG negative to the PNP gate. The life of an EE neophyte is not easy.
So now I go from the power supply positive, to the PNP source, PNP drain to the primary, other end of primary to the NPN drain and NPN source to power supply negative.
...

Ok wattsup, now what I do not get is why you connect the FG negative to the gate electrode of the N channel MOSFET?
Do you find significant difference between:
connecting the FG negative to the power supply negative and the two gates of the MOSFETs are connected together   OR
as you wrote the FG negative goes to the gate of the N channel MOSFET?

I think the preferred method would be the FG negative i.e. the zero ground point of the FG would be tied to the negative pole of the power supply.

One would think there is a difference in the switch on and mainly the switch off times of the p and n channel MOSFETs... that could play any  role in the switching actions?  It is sure there is a difference in their reaction time.

rgds,  Gyula

Hi @gyulasun;

Well what I indicated is the positive of the FG is going to the gate of the NPN or N-Channel mosfet and the negative of the FG is going to the PNP gate or the P-Channel mosfet. I think you may have understood that in reverse.

hi Wattsup,

Yes, I mistyped it, sorry for that.

The reason I am (was) doing this is mainly due to how I understand the specs on these two mosfets (IRF840 and IRF9540). The PNP IRF9540 specs, when I look at the Gate to Threshold Voltage says minimum -2 to maximum -4 volts. So I thought the gate of the PNP needs only a negative voltage source. On the NPN IRF840 the same spec says +2 minimum to +4 maximum. So this is why I connected the FG in that way. 

I see. Normally when you use a voltage source (battery, mains, generator) then you always consider two poles (when it is DC) or two points (when it is AC) between which you understand the amplitude info of the voltage source like a 9V battery or 120V mains voltage etc.  You always measure voltage between two points, not one point in itself, right? 
This is so with the gate voltage in case of MOSFETs, they do not state in every point of the data sheet that it is meant between the gate AND the source electrodes or with respect to the source electrode of the MOSFET BUT they mean it! Gate threshold voltage is always meant with respect to the source electrode of a FET. And in the data sheet there are Figures showing recommended measurements setups how they mean a specific data.

One of the problems I have is when you say to put the negative of the FG with the negative of the power supply. I had been pulsing mosfets with only the positive of the FG and it has worked in the past so this is why I never used it. OK I will do as you say. 

Thanks. It can work with only the positive of the FG but two things are involved with it: it may work erraticly and you will not have any idea of something happened due to the circuit properties or due to an erratic MOSFET switch on or off (for instance the gate-source self capacitance has no possibility to discharge correctly because the gate is floating with respect to the source electrode), the other thing is you may also input spurious frequencies you are also not aware of, (the gate electrode has a MegaOhm input impedance at low frequencies like 50 or 60Hz mains and you always have field strength from the mains in a room or lab etc that can modulate the input frequency coming from the FG.
It is true an FG has a low output impedance (pure 50 Ohm) but if you use a ground independent circuit with a battery as the voltage source (not a power supply fed from the mains like the FG) then chances are the gate is still unterminated (it cannot 'see' fully the FG's output) so it can remain at high impedance, hence can pick up 'waves' as 'foreign' voltages and you may not be aware of it.

If you used only the positive output of the FG so far for driving mainly N channel MOSFETs and you found your MOSFET was switching as you expected then probably you used power supply as the DC voltage source to feed the drain-source electrodes, hence chances were the negative supply point was enough 'low impedance' point via the mains network from the FG's negative output or vice versa so the gate-source path had good chance for normal operation. Not so in case of operating from a battery, then the path for the low impedance route is missing via the mains network, unless some other measuring instrument's (scope etc) negative point brings it to the source electrode of the MOSFET or JFET.

If you study any schematic that includes a MOSFET or JFET you find that the gate-source electrodes always have a DC path between them: either a resistor or a coil or a transformer's coil, these DC conducting components insure the gate-source capacitance can discharge (the RC time constant is important of course).  If the DC path is missing (and it can be missing if you use only the FG's positive output), then the control of the gate may become erratic.  See for instance Groundloop's schematic for the two coil switch shown earlier, the 2 kOhm resistors are between the gate-source electrodes. (In case of using an NE555 timer for controlling a MOSFET directly, the DC path between the gate-source is insured by the 555's inner circuits but then you have to connect the 555 negative supply point with the N channel MOSFET's source electrode too, right?

OK, I did as you say. I made some other minor changes too and the AC and DC effect is no longer there. Only DC output. But I found some great results.

Thanks for showing the video, nice job.  I cannot judge if "the AC and DC effect is no longer there"  result is good or bad but it is ok you found some great results.  One thing is sure: if you apply correct methods for serving the MOSFETs (or any other device) then the MOSFETs will operate as they are destined to and you can depend on their performance and focus on the effect of the now correct switching.

RE on your finding the different diodes behaviour:  the so called reverse recovery time is the important data here for them, t_rr, and there are big differences between the types of course.  So called fast or extra/super fast recovery diodes are preferred for capturing voltage spikes from the collapsing magnetic fields, t_rr should be in the 20-30 nanosecond range or even better.

rgds,  Gyula

@Otto,
Ha Ha.

@wattsup you take the right side I'll be on the left and maybe one thought will emerge. Ha, Ha.

@all,
I have AOP605s on the way. These will drive my irf840s in a push pull type operation from a 555. 

Remove your watch and rings in further tests. These items make you a better antenna, an electric fence, lightning rod.

@All

I put up a new short video located here;
http://www.youtube.com/watch?v=vUH4uIbvRBo

@gyulasun

Thanks a heep for coaching me through this. I see now the error in my ways. I won't promise not to make more errors even intentionally (just to see the effects - lol). One of my breadboards looks like a cratered moon surface so I am used to it.

I thank you for your kind patience.

@otto

I answered you in @GK's thread.