The pulser did not like the open ended coax, it blew the IGBT (all 3 leads shorted) and toke the 250V driver circuit (ir2153 and 2x irf540) with it.
A new IGBT is in, as an ir2153.    Tomorrow i will replace the both irf540's.
This reminded me to my initial plan to insert a fast fuse in the 250V supply line 
Itsu

Hi Itsu. Crikey, that was a bad break.
Maybe you can try driving the coax with whatever termination on its end that you are planning to test
with directly with your signal generator first, and monitor the signal generator output voltage
at the input to the coax to see how much it is loading down the output of the signal generator.
If it is loading down the signal generator output a lot, then that may be a configuration to avoid using
with the pulser. A 50 ohm resistor on any length of 50 ohm coax should like like 50 ohms resistive load at the
input to the coax, but the open circuit and short circuit at the end of the coax are transformed into a
different impedance depending on the length of coax and frequency you are testing at. Doing the signal
generator test should give you some indication of what kind of load the pulser will see if it were connected
instead of the signal generator directly. Just an idea... I wonder if the input to your coax winding with the open
circuit end looks like a very low impedance at its input at that length and frequency, or if the problem was that it
is a very reactive impedance which maybe your pulser did not like.... You should be able to get some idea about that by
trying the signal generator test, if you are interested in trying it. 

I told you so 

A bidirectional TVS diode that clamps the collector to emitter closely to the IGBT's maximum rated voltage, might have prevented this disaster.
Also, this is a lesson that D1 should have higher maximum ratings.

Finally, if I were you, I'd put the fuse in series with L1.

We are back in business, parts are replaced, and some tests done.

I have changed the 1n4148 (at the gate of the IGBT) for a Bat46 and added a fuse in series with L1
The D1 diode (600V same as the IGBT) did survive.

I omitted the "coax open end" test to avoid further damage for now.

I did do the "50 Ohm termination at the end of the coax" test and redid the "shorted coax" test, see screenshots below.

When adding a 100 Ohm trimmer potmeter to the begin of the coax with the far end being shorted, it initially showed (collector voltage half way at 130V) the same
signal as the "shorted coax" signal (pot halfway) but when turning a few turns the peak suddenly doubled so i stopped.
The potmeter measured to be in hundreds of KOhms now, so probably was fried, basically presenting an open (or high ohm) load.
Guess i need a somewhat sturdier potmeter instead of the trimmer pot used now.




Void,  thanks for your insights, but yes, the 8 meter coax at 15KHz will not present much of an impedance transformation.


Regards Itsu

Updated diagram, ...

Thanks Itsu.

I used a range of fixed series resistors from 5, 12, 22, 32, 51, 68, 100 Ohm with the collector voltage on 130V.

See the attached .doc file for the screenshots, the first one is without any series resistor, so 0 Ohm.
Yellow signal is the HV probe across the DSR diode.

Itsu

The screenshot says it all, not very much!

Darn, that's a bit depressing.  I was really hoping your much improved test fixture (over Dally's) would show us the way.

So we're not back all the way to square one, but we are certainly missing something.  Any ideas as to what?

Missing that Dally device was fake from day one perhaps? Why should one dismantle his working OU device three days after presentation without taking any measurements and never show something like it again?

Yes thats strange,  but before we start again the "fake" discussion.

But do we have a theory about how the system should must work?
And can we figure out why Itsu don´t get big nano spikes in his grenade-inductor coil?
Do we get different results with karcher systems and is this explainable in Itsu's situation

Is it possible to put those two systems next to each other and figure out the aspects as mentioned above?

I' m asking this because it would be frustated if someone starts again building from scrap, without getting the current nowledge about the issue's we run into.
I think Itsu is getting right on track also in the way he presents his setup and results.

Greetings
 

I used a range of fixed series resistors from 5, 12, 22, 32, 51, 68, 100 Ohm with the collector voltage on 130V.

This is not what I expected to see.  Most likely the probe position is wrong.

I synced the nano-pulser to the Push pull driver (TL494) instead of the FG.
I will see if shifting the pulse compared to the sine wave (± 90°) has any effect.

How did you synch it?
Is the phase shift between the sinewave and the nanopulse, smoothly adjustable between 0º and 90º in your circuit ?