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

Re: Weird safety gap behavior - Understood (I think)



Original poster: "Gerry Reynolds" <gerryreynolds-at-earthlink-dot-net> 

Hi Terry, Ed, Dr R, Malcom, and all others that contributed,

Everything I've seen now makes sense to me.  The quick answer is what Dr R
suggested as being ferroresonance.   A google search on the subject found
some interesting info.  I'll sumarize what I think I learned for others to
disgest and comment on.

Ferroresonance is a nonlinear phenomenum that can occur when ferro material
is used to increase the inductance of a coil (chokes, transformer, etc).
The characteristic of the ferro material can be seen by by plotting FLUX (or
flux density B) vs Current.  The resulting B-H curve has two relavent
characteristics.

1. The curve does not go thru the origin (ie, at I=0 there is either + or -
residual flux remainding).  Other words, there is a hysteresis to the curve
depending on the previous excitation.

2.  The curve flattens out at high enough + or - current (it saturates).

Up to now, I've been thinking of NSTs as having an equivilent inductance
where:

      XL = Vs_oc / Is_oc  (the transformer impedance assumed to be all
inductive)

This value is used to compute the resonant value of Cp where:

      Cres (nf) = 10^9 / (2*pi*line_freq*XL)

The value of L (XL/2pi*freq) is often assumed to be linear and the resonant
frequency calculated for the LC product is the linear resonance.

The problem is that the inductance of the transformer is not constant but
changes with current.  It is the slope of the hysteresis curve at the
current you are at.  So as saturation occurs, the inductance goes down.  So,
if you think you are running with a Cp at 2.5*Cres where the resonance is
38Hz and the transformer begins to saturate, the transformer inductance will
lower and raise the resonant frequency.  Resonant charging can result and
the output voltage can become uncontrolled and the current can go up.  The
increase of current can drive the transformer further into saturation and
reduce the inductance even further.  In any case, there are multiple stable
modes (non saturated and saturated).  The hysteresis explains, I think, why
I had to reduce the variac substantually to bring it out of saturation.

Terry, I repeated the current measurement and the current actually did go up
a little so we must not have been coordinated in our experiment.  The fuse
would probably protect the NST if the MOVs were not there.  The reason the
uncontrolled resonant rise stopped at 30KV (actually 29KV), I believe, was
because of the MOVs (16  1800V MOVs = 28800V).  I have not verified this and
the match of measured voltage and MOV ratings could be a coincidance.  I did
run for about 5 seconds this way, shut down, and see if anything got warm.
MOVs were cool and so was everything else.  With no spark gap in the
picture, the line power was about 1KW and this was probably mostly being
dissipated in the NST.

Interestingly, when the SRSG is turned on, the energy that it pulls from the
charging circuit seems to either prevent saturation or sink enough energy to
prevent resonant rise (probably the latter).

Anyway, this is my take on the phenomenum.


Performance with SRSG running and no safety gap firing:

Cp -at- 2.5*Cres
Optimum phasing ~1.5ms after peak (just as simulation predicted)
Vfire = ~23KV
Vmax = ~25KV
BPS = 120
NST is 15KV -at- 60 ma
Toroid is 6x24 inch
Secondary 1440 turns of 23awg for a 8x36 coil
Primary is tapped at ~17 turns

Wall Power (at 140V on the variac) = 1200 watts
PF = 0.71
VA = ~1700
Line current = 14amps rms
Line voltage = 120V rms
Rated NST VA = 900     (you can draw more real power from the NST than
rating)

Power arc to grounded object = 62 inches


Terry,  the power numbers we were measureing were low because I had some
interference in the variac and could only get it to 130V.

Comments welcomed,
Gerry R



 > Original poster: Terry Fritz <teslalist-at-twfpowerelectronics-dot-com>
 >
 > >
 > >         Certainly you CAN achieve resonance at high flux density when
the
 > >permeability is decreased after going through peak.  In my case it was
 > >at low flux densities that I observed the phenomenon.  What puzzles me
 > >is how your transformer insulation has withstood that high a secondary
 > >voltage!!!!!!!  I assume you haven't run it that way for long......
 > >
 > >         I would think the buzzing indicates core saturation, which
should be
 > >very evident in the input current waveform, which should show big
 > >spikes.
 > >
 > >Ed
 >
 >
 > That is what I found so odd.  As Gerry was very slowly turning up the
 > variac, we watched the output voltage on the scope too.  The output looked
 > pretty distorted with higher order harmonics, the current was about 11
amps
 > and the power factor was about 0.17.  Then the voltage suddenly (~1/2 sec
 > ramp) jumped up to a fairly good 30kV peak sine wave, the current actually
 > "decreased" to about 9 amps, the power factor went to 0.97 and the trannys
 > hummed like mad.
 >
 > I always thought that fuses would protect against this condition.  They
 > apparently do not!!
 >
 > It seems like the system is going from a nonresonant low power factor, low
 > real power situation (about 224W real)... to a practically unity power
 > factor situation (about 1080 watts real).
 >
 > It all has me confused %:-p
 >
 > Cheers,
 >
 >          Terry
 >
 >
 >
 >
 >