Ghazanfar Ali Generator - Utlilizing trapped energy

[poynt99]

@poynt99
As far as 800W output sim file is concerned, dont waste your time on that. It is unstable schematic. Damping occurred after 4 minutes in my case. However schematic of 42 W output is stable. I made it the hardware run for 1 and a half day and the bulb was continuously lighting.

What happened after the 1 and a half days? Did it quite running?

.99

[dole]

Hi,
I was very concern about 337 uH and so far I did not even try so low inductance. ( I am at mH range)
For those who want to “save” FG I may suggest hex Schmitt trigger HCF40106B, what I tried as FET driver at 50% duty cycle.
First I wanted to see if there is any relation at resonance frequency of the tank circuit
(capacitive and inductive reactances equal to each other so the total impedance will increase),
but I am not sure if this is possible with a square wave (Gyula ?)
I may only report that I was able to hold around 4-5V in 4700 uF capacitor without R.
But changing properties of tank circuit this was even not possible at all, so it may be that there is some relation there.
So far nothing miracle and I did not burn anything jet
But no rush there is only one week left and Ali will be safe home and uncover everything….. I guess.
d.

[gyulasun]

Hi Dole,

If you go up with the coil in the mH range from the 330-350uH values (this latter depends on the preferable measured 18uF capacitor),  then you inrease the resonant impedance of the LC tank circuit. 
Here is a link, unfortunately it needs Windows Internet Explorer to run it, it is able to calculate the resonant impedance of a parallel LC circuit once you give it the coil series resistance (the DC resistance is already ok to be in the ballpark).
http://www.thetaeng.com/calc_LC_tank.htm

And when you increase the resonant  impedance in this circuit then the current taken from the 4700uF capacitor will be less of course when the MOSFET switches ON.  If we knew how long the 1mm wire Ali had in his coil, we could calculate the DC resistance, hence we would know his tank circuit's resonant impedance at 2 kHz.  From the link I gave in earlier post, a 3m long 1mm diameter copper wire has got 0.064 Ohm DC resistance,  let's 'ruin' this to 0.07 Ohm, this results in 278.5 Ohm resonant impedance when the L=351uH and C=18uF.  So this 278.5 Ohm is which  governs the current via the switch when it is controlled by 1kHz, the current taken from the 4700uF in this case is 40V/278.5Ohm=143.6mA. 
Now if you have say a 3mH coil with 1 Ohm DC resistance, then your tuning cap value is about 2106nF and the resonant impedance 1424.5 Ohm, much higher than with the 351uH coil  (OF COURSE this impedance strictly depends on the actual wire resistance either Ali or you actually have got).

You ask on the square wave drive if it creates resonance, yes it does, no problem. 

Gyula

[dole]

@Gyula
Thanks, I perfectly understand all that you are saying but as I see it from my testing the biggest problem is the tuning.
At the point when I am out of the resonant pulse, there will be a large increase of 4700uF dissipation due to decrees of
LC resistance and not only that but this will immediately destroy parallel LC oscillation and result in FET overload.
Theoretically at the resonance LC resistance = infinite, but I was never able to get there.
Finally mixing this with nonlinear R (load) in the series LCR seems not so easy to achieve!?

So it must be some way to self-tune to this nonlinear oscillation (Gyula ?) because we are in the real world and not in an ideal simulation.

But with hard work.. maybe
d. 

[poynt99]

So I took the plunge and downloaded Multisim in order to see what is going on with Ali's OU simulation.

There are a number of simulation settings that Ali changed when compared to the standard default settings, but only one had the effect of enabling the continuous oscillation, and it is the RETOL setting. This can be found in the "Simulate->Interactive Simulation Settings->Analysis Options tab->Use custom settings-> customize...->Global tab".

The default setting for this is a value of 0.001, whereas Ali has it set to a magnitude higher at 0.01. This is the limit at which NI recommends you set it, but evidently, this setting causes excessive errors in Multisim, because changing this parameter in OrCad PSpice does NOT cause this error of continuous oscillation. This explains why Ali did not upload an equivalent OrCad file. btw, increasing RETOL should only be done when convergence is a problem, but this circuit converges just fine with the 0.001 setting.

IMHO, Multisim is a cumbersome program to use in comparison to OrCad PSpice. It seems to have some limitations as well, but it does have some modern (but minor) features that PSpice seems to be lacking. And now having seen this error in the coding, I'd say Multisim is well suited for educational purposes, but it's not a serious engineering tool.

See the attached for a little info on the "RETOL" parameter.

Sorry Ali, tweaking a faulty program to coax it into doing something it shouldn't be capable of, doesn't cut it in terms of making a claim of OU imo. But kudos to you for finding and exploiting that flaw.

Now, if you can show a video of this actual circuit continuously powering a load, then we will have something of real interest to discuss. 

Regards,
.99