SISG Power Supply Theory (fwd)

["Tesla list" <tesla@xxxxxxxxxx>]

---------- Forwarded message ----------
Date: Tue, 26 Jun 2007 23:47:48 -0500
From: Mark R Dunn <teslamark@xxxxxxxxxxxxx>
To: tesla@xxxxxxxxxx
Subject: SISG Power Supply Theory

All:

 

A number of TCML members have requested that I write an explanation of how
the power supply charges the tank cap and triggers the SISG circuit in an
SISG Tesla Coil Application.  For the experts this will likely be trivial
and boring.

 

SISG stands for “Sidac IGBT Spark Gap”.  The concept was developed in
April/May of 2006 by Terry Fritz.  Terry used a test coil to prove the
concept.  I converted a DC based coil to SISG in June of 2006.  Terry
documented and built the Piranha Coil as a result of his testing and
analysis.  A number of other TCML list members now have SISG coils operating
or are in the process of making them operational.  Finn Hammer has built a
triggered version which really is a new concept in and of itself.  You will
see why later.

    

The SISG coil uses a pulsed DC power supply to charge the tank cap.  When
the tank cap reaches a “threshold” voltage (set by the # of SISG circuits)
the SISG fires and acts like a switch that allows the coil to resonate just
like a spark gap coil.  In fact view the SISG as replacing an RSG – with the
additional power supply modifications.  Terry’s basic SISG circuit fires at
900 volts, so the choice of threshold voltage is a multiple of 900 volts if
you use his circuit.  I built some boards (identified as SISG4), which a
number of you have, that have (4) SISG circuits on each board.  So each
board gets 3600 volts threshold.  (2) boards 7200 volts.  (4) boards 14,400
volts.

 

The way the SISG circuit works, it can only be triggered with a DC voltage
of correct polarity.  If you study Terry’s circuit, you will see that the
gate will never fire if the polarity is reversed.  Also, once the SISG
fires, it cannot turn off until the DC voltage is removed.  Thus the need
for a pulsed DC power supply.  The rising DC turns the SISG on and when the
DC is removed the SISG turns off.  Now for discussion purposes we can talk
about the BPS(breaks per second) for the SISG at steady state in a similar
way we would discuss BPS for an RSG.

 

Just like a spark gap, when the SISG turns on the coil “rings up” and “rings
down”.  The issue that causes some headaches, is that you do not get to
independently control the BPS like in an RSG.  The BPS is tied to the
charging rate.  So if the tank cap is charged very quickly then the SISG BPS
will be very high.  For example, when I first converted my DC coil and had
(4) MOT’s charging a 37 nF cap, my SISG was operating at 1200 BPS.  This was
a terribly inefficient condition.  As I reduced the impedance of the
transformers and increased the value of the tank cap the BPS came down.  I
eventually, wound up with (2) MOT’s ( a voltage doubler) and a tank cap
value of  ~118nF to get the break rate down to 120 BPS.  Terry has used a
165 nF cap with his single MOT design.

 

Now you see why Finn’s development is interesting.  By independently
controlling the IGBT firing rate we could work with any MOT and cap
combination (as long as the cap wasn’t too large to charge before we hit the
peak) and simply fire the gate at the appropriate time. 

 

OK back to the pulsed DC charging.  Our pulsed DC is produced from rectified
AC so the AC- peaks are flipped over giving us 120 Hz DC pulses.  Another
way to look at it is repeating ˝ cycles of AC always with the same polarity.
Thus, if we want to run with a threshold voltage near the peaks of our DC
pulses then 120 BPS will give us the highest power throughput.  If we run at
60 Hz, then power will be cut in half because we will charge on every other
pulse.  If we try to run at higher than 120 Hz, say 240 Hz, then we will
have to lower the threshold voltage so than the SISG can fire twice for
every pulse.  Bang energy varies with the square of voltage so this option
is poor as well.  Our highest threshold voltage would occur if we can get
the SISG to fire right at the top of the pulse.

 

In computing the available threshold voltage, don’t forget that the peak
voltage is 1.414 * RMS voltage.  So if you use a single 2000 volt(rms) MOT
with a doubler your peak voltage will be 2*1.414*2000 = ~5600 vdc.  You will
find that the MOT will saturate around 100 to 110 volts so this would be
reduced accordingly.  With (2) MOT’s max threshold would be around 11,200
vdc.

 

In terms of charging time, the DC pulse takes 8.3 milliseconds to complete
its cycle (120 Hz).  So to rise from 0 to peak voltage occurs in 4.15
milliseconds.  So this is max charging time for the tank cap and still
maintain the 120 Hz.  Note:  The actual charging period is longer because
the coil is only firing for about 50 microseconds after the SISG fires at
the peak, so the tank cap is charging on the back side of the DC pulse as
well.

 

Enough talk.  Let’s work an example calculation that shows how to estimate
the cap size that will work.  We need to evaluate the impedance of the MOT.
I have studied about (20) MOT’s and they are quite variant, but if we focus
on the smaller light weight ones, they will pull about 10 amps at 100 VAC in
Tesla service (OK – so you’ve pulled 13 amps – then work the example with
your numbers).  So this is about 1 KVA.  Unfortunately, they have a power
factor in the .5 to .6 range(or worse) so the power throughput is only 400
to 600 watts.  Thus, if we are using a single MOT and a voltage doubler to
get ~5000 vdc, then the current is only ~ 100 mA.  This gives an impedance
of Z=V/I of 50000 ohms.  So the charge time constant for a 165 nF cap would
be t = (50000 ohms)*(165 E-09 F) = 8.25 milliseconds.  Recall the available
charge time was only 8.3 mS.  Thus, we only have (1) time constant to charge
the cap so we will only be at 64% of charge.  If we could double the current
(by using a bigger MOT), then we would reach 2 time constants or 87% of
capacitor charge.  Alternatively, we could decrease the cap size and
accomplish the same thing.  This is not an exact science, but can get you in
the ballpark.

 

Some have asked about using pure DC for this application.  There are two
problems with this.  The first was previously mentioned – the SISG will fire
once and then never turn off if fed pure DC.  The second is that to get the
pure DC you need filter caps and the filter caps will add to the tank cap
during resonance.  There are ways around the 2nd issue, but not the first.

 

Well that’s it.  Hope you can follow my ramblings.  

I did recently receive a shipment of SISG IGBT’s that I ordered last fall so
if anyone needs any contact me off list.  Took about 38 weeks.  I think they
were swam here from Singapore.

Everyone should try SISG.  You’ll never go back.  It lasts forever and never
wears out.  Been running over a year now and never blown an IGBT.  Some
think SISG is expensive, but I spent a lot more screwing around with RSG
variations for a couple of years than what SISG costs.

 

SISG schematics and other SISG info are here:
http://www.teslaboys.com/SISG/index.html

 

Mark Dunn