Kapanadze Cousin - DALLY FREE ENERGY

[Hoppy]

quote from verpies:
Will the 1500v TVS diodes have any effect on the fets, there? 

Nick,

The TVS diodes are 1500W (not volts) rated at a clamp voltage rating below the mosfet's rating.

[NickZ]

    Verpies:  But, it's still working on the same device. And can be just as damaging as we get into applying higher and higher voltages from the HV circuits, interacting on both of the induction circuits.

   Hoppy: Ok thanks, 1500W TVS. So, will they help at 10000V but at low wattage from the Kacher or HV circuit, which is induced into the induction circuits, and still able to keep the fets happy? Maybe that not a problem? But, it seams to me that some of the burn outs, that some of the guys are experiencing may be from the HV interaction, and not "BEMF". 
  Itsu mentioned that he wanted to avoid using a Kacher circuit.
  The full bridge rectifier diodes can also become a problem, and overheat, as well. So, which are the best diodes to use for the bridge? Remember that Ruslan's 4000w device was frying those diodes, and he had to turn off the second bank of bulbs.

  I'll be ordering some parts to build up the commercial TL494 board, as well as the fet driver circuit. But, I'm still confused about which is the best simplest mosfet driver circuit to use, and which snubber components to use for the fets, also. 
So, any new or updated advice on that is always welcome, especially if the circuits have already been tested and working as expected. 

[Dog-One]

Wow, busy day in this thread--five pages since I last posted, quite a storm.

On the matter of the loss-less clamps, this is no doubt a superior design.  Thanks verpies for getting it out there.

The last post I made in regards to Arunas' implementation was to make it more robust so a small mistake doesn't end-up blowing components leading to down time.

There is an advantage to not using loss-less clamps, though verpies listed it as a disadvantage:  Heat dissipation.
I mention this because monitoring the large snubber resistor is an easy way to determine if you are truly getting power flow through the yoke or not.  With the loss-less clamp design, you just see draw at the power supply to be minimal, but it isn't clear as to why.  In effect all the power is being recycled by the clamps, which is a good thing, BUT one might be tempted to just boost input power, which is a bad thing.  At this stage of the game, being able to detect reflected power (from a hot snubber resistor) is important to know so you can take the appropriate steps to change the yoke or the components down-stream of it.  With the snubber design you will know straight away whether the input power is going to where it should or not.  If it is, the snubber resistor will not get hot.  You will see volts/amps on the power supply and know this power is getting through the yoke, instead of being reflected back.

To Nick's point, I don't think any one of us knows for sure where the magic OU is manifesting from.  It could initiate anywhere in these circuits/components and be amplified by the rest.  What I'm confident of is the more we can see, the better chances we have of getting something to work.  Optimizing circuits too soon in this development could easily knock us completely off the path.  Once we get a runner, then we can try all sorts of improvements.  I just don't think now is that time.

Having said that, it is also important to put enough care and thought into the push-pull driver so it doesn't blow up at the slightest loose connection or over-voltage condition.  If a throw-together kind of driver is the only kind that will produce a self-runner, we're probably dead in the water already.

M@

[verpies]

Verpies:  But, it's still working on the same device.

Yes, but in a different circuit that is isolated by a magnetic barrier - a transformer. 
"Kacher's" HV can get back to the yoke's primary circuit only through some capacitive coupling, which constitutes a parasitic effect in a transformer.

And can be just as damaging as we get into applying higher and higher voltages from the HV circuits, interacting on both of the induction circuits.

The dissipative or non-dissipative clamps act only on the yoke's primary winding circuit.  They will clamp any other undesirable voltages there.  They will not affect the "Kacher's" HV influence on the grenade coil.

Itsu mentioned that he wanted to avoid using a Kacher circuit.

I would be wary of the "Kacher" too as it destroys test equipment. I think Itsu has paid dearly for that.

I'll be ordering some parts to build up the commercial TL494 board, as well as the fet driver circuit. But, I'm still confused about which is the best simplest mosfet driver circuit to use, and which snubber components to use for the fets, also. 
So, any new or updated advice on that is always welcome, especially if the circuits have already been tested and working as expected.

MOSFET driver chips are easy to evaluate by their peak source and sink currents, In/Out voltages and rise/fall times and polarity.  For example the non-inverting driver that Itsu is using can source 9A in 20ns into a 10nF gate capacitance and can be supplied with 15V maximum voltage.

The full bridge rectifier diodes can also become a problem, and overheat, as well. So, which are the best diodes to use for the bridge?

Diodes with the lowest forward voltage drop (V_F) you can get.

For very high powers and low losses it is possible to use MOSFETs instead of diodes in a technique known as Synchronous Rectification.
This works because MOSFETs have lower voltage drops than the best diodes, up to a certain current - see the graph below:

[verpies]

...being able to detect reflected power (from a hot snubber resistor) is important to know so you can take the appropriate steps to change the yoke or the components down-stream of it.

Interesting point, but guys like Itsu will just place their Hall current probe on one leg of the recovery diode and see the reflected power faster than the resistor can heat up.
For those without current probes, an additional CSR would be needed to sense that reflected current.