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

Arnold Toroidal Ferrite?? Cores



Hello All,

Mike McCarty and I got into a long discussion last night concerning
these Arnold choke cores from Hosfelt electronics.  I thought I would
share some of these thoughts and information with all.  Mike, I did
many calculations on this this morning and have discovered some interesting
facts.  There has been some discussion in this area lately by others in
the group.

First, the old question of maximum flux density came up.  Being tired of
hearing what the core's material "should be", I tracked down the exact
information from the people that bought the rights to that part of the
business from Arnold.  That business is now owned by Pyroferric Int'l,
217-849-3300.  From the Arnold catalog, part number FE-2500-1201, is
NOT FERRITE, IT IS POWDERED IRON.  I repeat it is POWDERED IRON!  So, we
can throw all of our wonderful ferrite data on this core out the window.
But, this is a good thing.  Powdered iron can take a higher flux density.
These cores have a max. flux density of 8,000-10,000 gauss.  The optimal
operating gauss is considered to be 5,000.

First I started to do some calculations on flux density.  I was only
considering the high frequency component.  Then, as I mentioned to Mike,
what does the 60 Hz component do to the flux density?  Can the 2 voltage
componenets be considered separately?  Or not?  After talking to the
technical gurus at Pyroferric Int'l for almost 45 minutes, I was still
unsure.  It seems they had no experience in this area either.

Before continuing, let me give you the formula for flux density, as given
by Arnold:

                                 (RMS voltage)x10^8
flux dens. = -----------------------------------------------------------
in gauss     4.44x(cross section in cm)x(# of turns of wire)x(frequency)

I have made some chokes with 165 turns.  I also am (was?) going to use 4 of
them in the circuit.  The voltage will be divided among the 4 chokes.  If I
figure on using a 15 KV supply and having about the same amount of votage
in RF to filter out I can use the following figures for flux density of the
2 separate components:

For the supply voltage:

      3750 x 10^8
----------------------- = 2,116,412 gauss
4.44 x 4.031 x 165 x 60

For the RF components, assuming a 15 KV RMS value and a 100 KHz frequency:

      3750 x 10^8
---------------------------- = 1270 gauss
4.44 x 4.031 x 165 x 100,000

As you can see from the calculation of the supply figure, the core would
be more than a little saturated by the 60 Hz component.  The flux density
from the 100 KHz component is well under saturation level.

So, now the question becomes; Can the 2 components be considered separately?
Again, Pyroferric said they were unsure.  Or does the fact that the core will
already be saturated by the 60 Hz supply make it useless as a  RF choke?
Based on the tests reported on in the last issure of the TCBA newletter, I
would be lead to believe that they can be considered separately.  So,
therefore even though the core is very saturated by the 60 Hz component, the
RF component still sees inductance due to the fact that it is not saturated at
its frequency.  I don't know if this assumption is true or not.  I would like
to set up a test on this, but I am afraid I do not have the equipment.

ASSUMING that the 2 components can be considered separately and all of this
info now secured, I did some calculations on the maximum flux density
expected in my cores.  I have approximately 165 turns in 2 layers on my cores.
I have 15 KV powering my system.  I intend on using 4 cores (2 on each side)
in my circuit.  So that means that each toroid will see 1/4 of the total
voltage drop of RF per side.   (While one side is seeing the positive side of
the waveform, the other is seeing the negative.  The potential across any one
choke should be 1/4 the total)  I realized that I needed to make some
assumptions here.  I do not know what the amplitude of the RF is that is
superimposed onto the 60 Hz source.  I assumed some very nasty RF was there.
Assuming a good resonant rise in the tank circuit, I used 20 KV RMS as a worst
case figure.  (This again is only a guess, by feel so to speak.  Keep in mind
I am playing devil's advocate here and am trying to make it a very bad
condition.)  So, I used a 20 KV RMS figure to calculate the flux density
in these cores.  Again, each core will only see 1/4 the total, so 5000 VAC RMS
is the figure I used.

Based on my set-up, here is the calculation, based on a 100 KHz frequency:

                      5000 x 10^8
flux dens. = ---------------------------- = 1693 gauss
in gauss     4.44 x 4.031 x 165 x 100,000

So, this should be fine, based on a 8K-10K maximum flux density.  Even if
only one choke per side is used, this number only doubles to 3386 gauss.  If
half the turns are used, the density goes to 6772 and you are still within
operating range.

This also changes if you are running a higher or lower frequency.  The lower
the frequency, the worse the situation becomes.

So, I calculated the minimum number of turns per core to keep the flux density
below the a 8,000 figure at different frequencies and voltages.  I made
a some charts using Lotus to look at various values.  The worst case I figured
was for 50 KHz with just 2 chokes (one on each side) and a 15 KV supply.  This
figures out to be a minimum number of turns of 105, to avoid saturation.

I have attached this lotus file for you amusement.  It is CHOKE.WK4.

I would still like to know more about how the 2 componenets affect the core and
the seen inductance values by an operating coil.  It is something that I would
love to have a concrete answer on.  It is also the governing factor on whether
these chokes will actually work, or not.

Please don't pick to death all of my "fudge factors" that I have used in the above
calculations.  They are there simply for the sake of having a figure to work with
that I felt were conservative figures.

If you find fault or have any comment on the above figures/calculations, please post
a message.  I would like to put this issue to rest in my mind.

Thanks,
Scott Myers