Re: tank circuit of VTTC

["Bert Hickman" <bert.hickman-at-aquila-dot-com> (by way of Terry Fritz <twftesla-at-uswest-dot-net>)]

Tesla list wrote:
> 
> Original poster: "Herwig Roscher" <herwig.roscher-at-gmx.de>
> 
> Bert Hickman wrote to the list:
> 
> > In that case we need a bit more information to properly answer
> your question:
> 
> Bert,
> 
> Thank you for your patience.
> 
> > - Planned plate voltage?
> 10 kVdc. If this caused problems I could use 6 kV as well.
> Some tube data for use as grid pulsed oscillator: peak dc plate
> voltage 15 kV, peak cathode current 12 A, average plate
> dissipation 1000 W
> 
> > - Planned tank circuit C and L?
> 2.2 nF parallel 130 µH. L/C ratio is rather high as tetrodes like
> higher load impedances than triodes.
> 
> > - Planned duty cycle (are you planning to drive from half-wave,
> > full-wave, or filtered DC)?
> Pure dc (filtered by one or some 15 µF/10 kVdc cap) as I'm
> planning on trying audio modulation. I've acquired a 3 phase HV
> tranny from a mobile russian radar that will provide more power
> than I'll ever need (weight 51 kg).
> 
> > - How many strings of caps will be connected in parallel?
> 1 string of 20 WIMA FKP1 2000 Vdc, 47 nF caps with dv/dt = 5000
> V/µs
> 
> > Assuming you're running in class C operation the peak-peak RF
> voltage on your tank cap can be 2X the applied plate voltage, and
> the peak circulating tank circuit current can be as much as
> Vplate*Sqrt(C/L).
> - So, 40 kVdc and 235 A (I peak) for the cap should be ok? And
> AWG 8 is an overkill and AWG 10 is sufficient?
> 
> >Unlike a disruptive coil, a VTTC typically has a much higher duty
> cycle... as much as 100% in the case of a system driven off a
> filtered DC supply. The combination of high RF RMS currents and
> high duty cycle make your application a potential capacitor killer,
> especially as you begin to increase plate voltage.
> - I've stated this already burning some ceramic rf caps.
> 
> > Flimsy leaded caps need not apply for this job... :^)
> - Usually I'm wiring my MMCs with thick leads, but the internal
> wiring of the caps....????
> 
> Regards
> 
> Herwig

Herwig,

No problem! At 10 kV plate voltage, your peak circulating tank current
will be over 41A, and almost 25A at 6 kV. Since you will be running off
DC the RMS value of the circulating primary tank current will be about
0.707 of this, or about 29A and 17.5A RMS respectively. The real problem
is that, because of skin effect, the skin depth of your copper
conductors is only be about 0.15 mils (0.004 mm), and only 0.19 mils
(0.0051mm) for aluminum At the operating frequency of your coil (about
300 kHz). 

Because of skin effect, a circular #10 AWG conductor that has almost
8200 square mils of conductor area at DC is reduced to about 48 square
mils of effective area at 300 kHz, making it behave as though it had
170X the resistance of the same wire at DC! To the RF current, it's
dissipating the same amount of power as the equivalent DC current
flowing through a 32 gauge wire. It's going to get very hot if you run
it at 29S RMS. I'd suggest beefing up the tank circuit by going to
larger diameter copper tubing or flat copper strap, and possibly silver
plating it as well. You are likely going to melt the PVC insulation on
your primary under extended running conditions. 

Flimsy capacitor leads won't cut it either. At these currents, simple
joule heating of the leads and capacitor plates can conduct and generate
substantial heating. For a single string, it'd be similar to heating
both leads of your MMC capacitors with a soldering iron while the unit
is running.  

However, I have no doubt that a properly constructed MMC could safely
handle this task. However, your MMC would need to be constructed of a
significant number of parallel chains to divide the current per
capacitor lead to a more manageable value. A single string won't work...
at least not for for very long. :^). 

Let's look at the MMC dissipation tests that Terry performed awhile back
using a 3 Amps RMS 370kHz CW RF source. The experimental data is
reproduced below, with the caps sorted in best-to-worst order:

* GE 42L32222 3.23 Deg. C (big cap but must be polypropylene) 
* Phillips 6.17 Deg. C (56nF version)
* WIMA FKP1 6.8 Deg. C (medium size polypropylene) 
  Arcotronics 14.9 Deg. C (Smaller tubular poly caps)
  Panasonic 19.04 Deg. C (medium size polypropylene) (old test)
  Panasonic 19.74 Deg. C (new test)
  PHILLIPS 367 KP/MMKP 39.76 Deg. C (4.7nF version small poly cap)* 
  FCI 0.015uF 69.52 Deg. C (big cap so probably polyester) 
  Sprague 430P 116.6 Deg. C (big cap so probably polyester)

Now, assuming you wanted to keep the capacitor temperature rise to less
that 10 degrees C., you will need to design your MMC so that each cap
sees no more that 3-4A RMS, using capacitors from any of the top three
vendors above. This implies that you'd need to use at least 4-6 parallel
capacitor chains to handle 17 Amp RMS and 7-10 parallel chains for the
29 Amp case. This also assumes the caps are separated enough to permit
individual free-air cooling. You may be able to squeeze more out of the
MMC by blowing air across them, beefing up the leads, or immersing the
unit in oil. You MAY also be able to get by with a bit lower design
voltage on the MMC - perhaps going to 25-30 kV - to save on cost.
However, there's not very much empirical experience with the long-term
performance of MMC's in a CW RF environment...

Hope this helped, and good luck in your design!

-- Bert --
-- 
Bert Hickman
Stoneridge Engineering
Email:    bert.hickman-at-aquila-dot-com
Web Site: http://www.teslamania-dot-com