Kapanadze Cousin - DALLY FREE ENERGY

[T-1000]

I build the verpies loss-less clamp circuit and after some initial problems, it works rather well.

I used the same MOSFET module which was driving the normal yoke setup, and added an interface to the 12 turn double layer primaries.
No 28 turn secondary installed yet, just the 3 turn to link with the wima cap/inductor series LC.
The Grenade has the 40W bulb connected to check for resonance.

Hi itsu,

I would rather make another Yoke driver board and remove any parasitic capacitance and additional primary windings from the yoke.
You can try this circuit we used:
http://i.imgur.com/DZ4wieI.png

[Dog-One]

You can try this circuit we used:
http://i.imgur.com/DZ4wieI.png

That is a pretty sound design there Arunas.  Let me offer a few tips to make it even better for this type of experimental work.

C14 -- By itself, this will not provide enough impulse gate current.  I recommend a 10uF tantalum in parallel with it.

R26 -- When there is considerable kickback (which you will get initially while trying to establish tuning), use the attached resistor to dissipate waste power.  Trust me, these work really well.  Pricey, but worth every penny.  The amount of heat they can handle is phenomenal.  The heat they bleed off gives you a direct relationship as to how well you are tuned.  When you finally get things right, they should run just barely warm to the touch.  Before you are tuned, they will save you down time by doing their job without a hitch.  Put your emphasis on the heatsink for this part more than the MOSFETs.

C16 -- Use a high quality, high voltage poly cap, pulse rated preferably.  Placing a 100pF high voltage disc capacitor in parallel with this is also a good idea.

C (non-existent) -- You will want some sort of large filter capacitor very near the MOSFETs for current dump.  Start with 1000uF electrolytic.  If the MOSFETs are physically apart from each other, use two of them.  I would also recommend a 4.7uF tantalum and 0.1uF disc capacitor in parallel with the electrolytic.

D5 & D6 -- I use STTH4L06 diodes since they are faster than the internal protection diode of the MOSFET and can handle 4 amps versus the 2.5 amps of the internal MOSFET diode.

Q1 & Q2 -- If you are at all afraid of amperage overload (which is entirely possible with the dump capacitors mentioned previously), upgrade to IRFB3077 MOSFETs.  They handle slightly less voltage which should still be okay with the snubbers, but three times more amperage.  I have never noticed the heatsinks ever going above ambient temperature.  Absolutely no need to use a fan.

Layout & Placement -- Think this part through carefully.  Avoid ground loops, use large traces where appropriate, grounded backplane foil, keep small signal traces away from heavy current traces, etc.  Be smart, take your time and do it right.  A rugged, well thought out PCB will accelerate your research cycle by months.  You want something that will just work and not be an issue while you focus on the critical aspects of the coil and tuning.

Optical Coupler -- With this board being modular in design and not having its own on-board oscillator, I would highly recommend using a high-speed optical coupler on the input, immediately followed by a pair of Schmitt Triggers for signal cleanup.  You won't be going into the MHz range, so really "high speed" would be overkill.  Just something that doesn't add a lot of latency should be sufficient.

Voltage Regulator -- With the opto and Schmitt Trigger added, you'll want 5 volt power available and 12 (or 15) volt power for the 21844.  I would recommend the LM317HV adjustable along with a 5 volt fixed regulator; reason being, you can run the MOSFET voltage up to 48 volts if need be without smoking the regulator.  Just remember, this is an adjustable VR, so you'll want to start with your power supply at low voltage and set the trimmer pot to 12 volts without going over or you'll blow the gate driver.  I would use DIP sockets and leave the chips out until I have the trimmer set.  Once I'm confident the VR is holding a solid 12 volts, then kill the power and insert the chips.  The other option would be to use the LM317HV to regulate for the IR21844 and a DC-2-DC converter to power the 5 volt chips.  Or the other way around.  Use the DC-2-DC converter to power the gate driver and a fixed 5 volt VR to power the opto and Schmitt Trigger.  I prefer using a DC-2-DC converter because it detaches the logic from the positive rail, so any spikes that might go that way never find themselves interfering with the logic devices.  The main trouble with DC-2-DC converters is they only have a small range of input voltage they can accept; this limits what voltage range you can run across your MOSFETs without having separate supplies.  Ideally you want to be able to go from 5 volts all the way to 48 volts because this makes it stupid simple to control current draw.  Besides, when you get your device working, I really want to see you start it with just a nine voltage battery, not a big lead-acid battery.   


I'm here for you Arunas and happy to help anyway I can.

M@

[itsu]

Hi itsu,

I would rather make another Yoke driver board and remove any parasitic capacitance and additional primary windings from the yoke.
You can try this circuit we used:
http://i.imgur.com/DZ4wieI.png

Hi T,

nice try, but i started with that "green box" snubber system from Oleg some weeks ago, it showed terrible ringing, i think Jeg had the same experience.

Much better is this loss-less clamp design allthough it seems somewhat complicated, it really is not.

I will take a look at Dogs comments below, perhaps some better/extra components could do the trick.

Itsu

[T-1000]

Hi T,

nice try, but i started with that "green box" snubber system from Oleg some weeks ago, it showed terrible ringing, i think Jeg had the same experience.

The Oleg's version of PLL driver was also with mistakes in circuit and I had to spend time fixing on assembled PCB already. This is why there is different version of circuit.

[verpies]

Some of us including myself do own a spectrum Analyzer.
I do know there are some funny looking harmonics which keeps going up and down constantly.

And some of us own a big brother of the scalar spectrum analyzer, also known as the vector network analyzer.
The difference is that the latter also pays attention to phase, while the former does not.
The I&V phase relationship in the grenade coil allow to distinguish the significant resonance peaks among the myriad of peaks in the v(f) only Bode plot.

May i ask how do we identify those "strong magnetic component" from the grenade coil.

By electrically shielding a magnetic search coil....unless you want to pick up the electric component as well.