[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: OLTC maggy - Modeling and "kickback"



Original poster: "Terry Fritz" <teslalist-at-qwest-dot-net>

Hi Steve,

Thanks for sending me your model.  I had to redo some of it since I could 
not get the libraries to work right but this is what I ended up with:

http://hot-streamer-dot-com/temp/OLTCsc02.gif

I changed these things:

1. I took out the diode since it was not helping me ;-)
2. I put the initial voltage on the cap itself.
3. I used a liner transformer.

These changes are no different from what you had since it all works the same.

In order to get it to match your scope reading:

http://hot-streamer-dot-com/temp/OLTCsc00.jpg

I did the following:

1. I added a 1pF cap in series with a 220kOhm resistor to simulate the loss 
of a steamer load about 1 foot long.
2. I raised the R2 value to 25mOhm to match the primary system loss shown 
on the scope trace.
3. I used 100mOhm open and 0.001Ohm closed on the switches since the 
program likes those values a bit better.  Probably makes no difference here.
4. I forced the simulation to 0.01uS steps so the output would not be all 
choppy:

http://hot-streamer-dot-com/temp/OLTCsc01.gif

So here is what I get now:

About 170kV peak on the output:

http://hot-streamer-dot-com/temp/OLTCsc03.gif

The cap voltage matches well now:

http://hot-streamer-dot-com/temp/OLTCsc05.gif

The current peaks at "only" 800 amps :o)  The V(R2:2) ripple (or output 
voltage) may be "leaking" onto your scope probe a little since the scope 
trace shows some ripple.

http://hot-streamer-dot-com/temp/OLTCsc04.gif

But the model here matches things very well now ;-)


Let's look at the "kickback" which I don't think is "kickback" at all.  I 
was confused at first since the kickback you described never showed in my 
models or on my coil measurements.  My quenching was always solid.

Lets suppose we quench on a voltage zero crossing so no voltage is left on 
the cap:

http://hot-streamer-dot-com/temp/OLTCsc06.gif

http://hot-streamer-dot-com/temp/OLTCsc10.gif

We get slightly lower max output voltage now:

http://hot-streamer-dot-com/temp/OLTCsc11.gif

But look at this big switch spike1 (thank goodness for transorbs!!):

http://hot-streamer-dot-com/temp/OLTCsc12.gif

The problem is that even though the voltage on the cap is zero (ok, I tried 
to get close ;-)) the current in the primary is still high.  So when we 
open the primary coil with current still in it we get a big spike due to 
inductive kick.

http://hot-streamer-dot-com/temp/OLTCsc13.gif

If we try to quench at (or near) the zero current crossing, we end up with 
voltage still left in the cap:

http://hot-streamer-dot-com/temp/OLTCsc22.gif

So we can't win!!  Since the voltage and current will always be 180 degrees 
out of phase, we either have to loose energy due to "left over" primary cap 
voltage or left over inductor current.  Of course, with the OLTCs great 
quench control, we can quench at the primary current "peak" if we want ;-))

I am guessing the energy lost is about the same either way...  220 amps 
976nH = 24mJ  and 230volts on 1.4uF = 37mJ  and we start with 323mJ.  But 
there is a trick!!  On my resonant line charged OLTC, I used the stored 
voltage to help "kick" the voltage for the next charging cycle.  So it 
really is not wasted.  On your system, that voltage just means you don't 
have to charge as much for the next cycle.  So, if you store the left over 
energy as voltage on the cap, nothing is really wasted!  If you quench at 
peak current, that energy is wasted in heating transorbes up.

So just leave the voltage on the caps as you are now.  Depending on 
polarities, the antiparallel diode will do that for you automatically.  If 
you pan through the quench range, you will find that not much happens since 
the antiparallel diode delays the quench through 1/2 of the cycle.  The 
other half is crunching and grinding as the current spikes get eaten up by 
the transorbes.  When you hit the zero current crossing point, things just 
seem right...  I just turn the quench timing control for the best sparks 
and don't worry about it ;-))

If we had a full bridge control, maybe we could trap the voltage 200V 
negative on the cap.  However, since the cap's stored energy is a V^2 
function that really does not buy us too much extra power.

BTW - When the quench is set right, the heating on the IGBTs and transorbs 
is very low too.  You may try putting the quench way out too to use up all 
the energy in the primary, but usually it is best to quench early to trap 
the energy in the secondary rather than try to grab every bit of energy out 
of the primary.

Cheers,

         Terry



At 10:37 PM 4/7/2003 +0100, you wrote:
>................
>
>
>I also investigated the kickback issue a little more. Here are some screen 
>grabs of a PSpice simulation that shows the problem.
>
>http://www.scopeboy-dot-com/tesla/t3model.jpg
>http://www.scopeboy-dot-com/tesla/oltcsim1.jpg
>
>And here is a scope screenshot of the actual tank cap voltage
>
>http://www.scopeboy-dot-com/tesla/oltctrace.jpg
>
>Sadly, it looks nothing like the simulated trace. I'm going to simulate 
>the maggie design in PSpice anyway to see what happens. If it looks good I 
>think I will probably convert my mini OLTC to a maggie. This will be a 
>good experiment. Same resonator, same IGBTs, same tank cap, same power 
>input, we'll see if the maggie makes bigger sparks, or even the same 
>length sparks more efficiently.
>
>If it does, I'll double the tank capacitance, redesign the primary, bolt 
>on two extra IGBTs, and increase the BPS to 1600. That will get me into 
>the 800 watt range that Boris challenged me to achieve ;D
>
>Steve C.