Re: Dump the RSG!

To: "'Tesla List'" <tesla-at-poodle.pupman-dot-com>
Subject: Re: Dump the RSG!
From: Tesla List <tesla-at-stic-dot-net>
Date: Wed, 5 Nov 1997 16:41:06 -0600
Approved: tesla-at-stic-dot-net

From: 	Malcolm Watts[SMTP:MALCOLM-at-directorate.wnp.ac.nz]
Sent: 	Wednesday, November 05, 1997 3:09 PM
To: 	Tesla List
Subject: 	Re: Dump the RSG!

Hi Greg,
           Thanks for noting a rather serious flaw in the 
plan....but, but.... I'll go back to the photos I took of my MOSFET 
"gap" last year. I recall it had a similar apparent flaw and yet it 
worked. The interesting thing was that had it not worked I would have 
been all over it trying to uncover the problem. 

> From:   Greg Leyh[SMTP:lod-at-pacbell-dot-net]
> Sent:   Tuesday, November 04, 1997 3:54 PM
> To:     Tesla List
> Subject:    Re: Dump the RSG!
> 
> Malcolm Watts wrote:

<re-arranged> 
> The back diode does not allow the switch to quench in one direction,
> because as the sec energy starts to return to the pri, the diode 
> will rectify this energy and charge the pri cap once again!  
> 
> This sounds undesirable at first, but I believe it's a good thing
> since the first beat is doing most of the spark work anyway.  
> Why not return the remaining energy present in the second beat 
> safely to the primary cap?  Starting with a partially charged cap,
> the DC resonant charger will then automatically draw less power 
> from the AC mains for the next shot.

In fact it does that by default since the diodes are holding off 
against the charging voltage. I even started to think about how I 
might connect the supply in series with the diodes and bypass it 
for the RF currents. That might also be on.
 
> >              I've been sitting on some Westcode devices (SCRs and
> > diodes) for some time now while agonizing over exactly how to use
> > them without destroying them. The kind of ratings I am talking about
> > (from memory) are 2.5kV holdoff/device (fwd and rev) with current
> > ratings around 2kA or so. di/dt should not be a problem with a
> > sufficiently low frequency coil. The agonizing bit comes from the
> > implications of stacking them. If one turns on fractionally before
> > the other (say two stacked), what happens to the sluggard? Bit of a
> > worry I think?
> 
> At SLAC we have SCR control stacks 15 units high for use in some 
> 90kV, 3 MW DC regulated power supplies.  Each SCR has its own firing 
> circuit which is powered and fired magnetically.  The circuit also
> fires its own SCR if the anode volage exceeds a set limit.  This
> approach seems to work quite well in protecting the 'sluggard'.

Thanks for that idea.
 
> >     The idea is to use DC charging and use the SCR/s to conduct in
> > one direction with the diodes to conduct in the other. This should
> > allow a per cycle quench capability with a sufficiently narrow gate
> > firing pulse (SCRs recover while diodes are conducting).
> 
> My biggest concern on your SCR's would be the di/dt rating. 
> Is the rating higher than 2*pi*Fpri*Ipeak?  Ironically, solid-state
> switches seem to be better suited to larger, lower frequency coils,
> due to the di/dt limit.

I don't have the figures handy but I did figure with a high enough 
surge impedance and the rather low primary voltage I could stay 
within the ratings at around 70kHz or so. A *big* primary cap would
take care of the rather large Cs of such a coil but again, the 
balance between Zsurge and Ep creeps in. No point in settling for a 
hi-L low Q secondary.
    I'll dig the spec out and have another look shortly. That current 
rating was the continuous one BTW.

Thanks,
Malcolm