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

More thoughts on protection chokes


----------
From:  Gary Lau  23-Feb-1998 2109 [SMTP:lau-at-hdecad.ENET.dec-dot-com]
Sent:  Monday, February 23, 1998 8:23 PM
To:  tesla-at-pupman-dot-com
Cc:  lau-at-hdecad.ENET.dec-dot-com
Subject:  More thoughts on protection chokes

I've given some further thought to xfmr protection circuits and done some
modeling and simulation on Pspice.  Below is the circuit I use and how I
modeled it.

VSIN is 7.5KV RMS = 10.6KV 60 Hz, half of 15KV/60mA NST secondary.
L3/R3 are lumped secondary components, 352 Hernys and 2.75K.
C3 is my bypass cap, 600 pF.
L2/R2/C2 is my ferrite core choke, 14mH, 0.5Ohm, 14.8pF
R1 is my series damping resistor, 500 Ohms.
SW1 & SW2 are switches representing spark gap - on-at-8.333ms, off-at-8.7ms
R4 is spark gap resistance, arbitrarilly set at 4 Ohms.
L1/C1 are primary tank components, 30.1 uH, .01 uF (290 KHz).

  +------+--R1--+--L2--R2--+------L3--R3-+
  |      |      |          |             |
  |     SW1      ----C2----+            VSIN
  L1     |                 |             |
  |      |                C3             |
  |      |                 |             |
  |      R4                +-------------+--GND
  |      |                 |             |
  |      |                C3             |
  C1     |                 |             |
  |     SW2      ----C2----+            VSIN
  |      |      |          |             |
  +------+--R1--+--L2--R2--+------L3--R3-+

I've found some unexpected things.  Looking at R2/L3 (xfmr terminal), I
don't see any evidence of the tank resonance energy, regardless of how I
include or exclude the C3 bypass caps.  This may be due to my simplistic
modeling of the gap conducting for > the duration of the tank ringdown.
When the gap is conducting, there is essentially no voltage across it,
and when the gap opens up, there's no longer have a tank circuit at the
frequency of the ringing secondary for it to couple into.  So I'm having
trouble remembering what we're trying to protect against!

If I model the gap such that it turns off while the tank circuit is still
ringing, there is a huge voltage spike that occurs as the current still
in the primary inductor tries to flow in the now open gap circuit, but
I'm guessing that the gap is unlikely to quench this early.  And the
protection network modeled above IS effective in blocking it.

What I do see in the PSpice simulations however is my choke and bypass
cap resonating and ringing down as soon as the gap conducts.  The choke
ringdown current here peaks at tens of AMPS and is due to the bypass caps
discharging through the chokes, gaps, and damping R's.  If L2 or R1
get smaller or C3 gets larger, the choke current increases, and I know
my ferrite cores begin saturating at currents above 300 mA.  I don't know
how to model a non-linear saturating choke, I'm wondering what happens to
that damped ringdown when the chokes saturate.  Could there be that as
effective inductance goes down upon saturation, that the chokes are
ringing UP and CAUSING NST damage?  (wild, unfounded, paranoid
speculation, I hope, someone please correct me!)

The post from Robert Stephens and Jim Fosse concerning tank circuit
arcing when using air core chokes is timely, although I don't think it
relates to 60 Hz resonance.  I had the same problem when I used air core
chokes, despite using 3K damping resistors.  The problem went away as
soon as I started using ferrite chokes.  I wonder if this difference could
be due to air core cores NOT saturating, whereas ferrite chokes may?

The more I ponder this protection circuit business, the less certain I am
of anything!

Flames, further speculation welcome,
Gary Lau
Waltham, MA USA