[Prev][Next][Index][Thread]
Re: Resonant charging and quench time

To: teslaatpupmandotcom

Subject: Re: Resonant charging and quench time

From: Terry Fritz <twfatverinetdotcom>

Date: Tue, 05 Jan 1999 18:15:02 0700

Approved: twfatverinetdotcom

InReplyTo: <C22566F0.00486E64.00atnotesmail.tellabs.fi>
Hi Marco,
If the primary gap attempts to break while there is still current passing
through the gap, it will NOT stop conducting. If you try to break the
current, the voltage will skyrocket and keep the gap conducting. Of
course, the computer can do it with ease but the situation is not
realistic. I did an experiment a few week ago to look into the Corum's
claimed voltage rise after the gap opens (coherence) where I used a very
special gap. I was able to break to gap in mid current but it was very
difficult and the spikes it caused were messy. In a normal situation, the
gap will only stop conducting at the notches where there is time for the
gap to cool while little current is trying to flow. There is always a
little voltage or current running around right after the quench which
probably is causing some spiking. I always break the gap at a current node
when the current passes exactly through zero. Reducing the switching time
in the model to around 100nS is probably more realistic than the pico
seconds the program could simulate. This will reduce the spikes to much
more realistic levels. I doubt if these spikes will recharge the cap in
real life. The energy left is probably burned in heat, light, etc.
BTW  Always use a saftey gap in a resonant charged system!!
Note that there are complex dynamics involved in when the gap fires with
syncronous, and in many cases, nonsync rotaries. As my last paper showed,
you may get higher primary voltages by firing significantly after the peak
in the AC cycle. This can be used to advantage or it can bite you too!
If the models don't agree with real life observations, it is always
tempting to question reality :))
So far, reality has proven correct every time! But I am still trying....
Terry
At 03:20 PM 1/5/99 +0200, you wrote:
>I am designing a DC tank supply employing resonant charging. My target is
>to have a bang power (primary capacitor charging voltage) stable and
>repetible from bang to bang.
>
>Playing with a MicroSim model of it I bumped into a sad thing: when the
>spark gap OPENS there is, of course, some power still left in the primary
>capacitor/coil circuit. The current left in the primary coil will make up a
>high voltage spike that, in turn, will charge back the primary capacitor to
>a certain potential.
>
>The polarity and magnitude of this voltage DEPENDS ON THE QUENCH TIME, i.e.
>on the exact moment in time the spark gap opens!
>
>If this voltage is positive, the following resonant charging cycle will not
>charge anymore the primary capacitor to a voltage equal to about twice the
>tank supply voltage, but to a lower value. Sad thing, because the next bang
>will have thus LOWER energy.
>
>If this voltage is negative, the following resonant charging cycle will not
>charge anymore the primary capacitor to a voltage equal to about twice the
>tank supply voltage, but to a HIGHER value. Again, bad news because you'll
>possibly break the capacitor or, at least, trigger the safety gap, if you
>are lucky to have one.
>
>This voltage left on the primary capacitor when the spark gap opens is
>almost random, as it is random the exact value of the quench time from bang
>to bang. Also those not using resonant charging will have their
>calculations and setups messed up by this factor.
>
>Has anybody noticed this effect? Any comments about this? Am I wrong ?
>
>For those of you lucky to have MicroSim (evaluation or real version), my
>source file can be downloaded from:
>
> http://www.saunalahti.fi/dncmrc/complete4.sch
>
>Try reducing L2 to 0.01 uH and check how the circuit works fine (ideal
>condition, only primary capacitor).
>Then go for L2 = 30 uH and play with the quench time: try values between 50
> and 80 us. You'll notice how the bang voltage varies widely!
>
>Your help is really welcome...
>
>
>