From: Winfield Hill
Subject: Seeking large high-permeability high-frequency ferrite cores
Date: 2 Jan 2003 08:05:30 -0800
Organization: Rowland Institute
X-Newsreader: Direct Read News 4.20
I'm seeking a source for huge high-permeability high-frequency
low-loss ferrite cores. Hopefully one of yo'all can help me!
I'm trying to design an inductor to resonate a 600kHz, 8kV system.
We're talking about say 200 to 400uH of inductance, used with 350
to 175pF of total capacitance. At 8kV the resonant currents are
about 10.6 to 5.3A, respectively, for a peak circulating power of
85 to 42kW. Clearly the Q of the L-C circuit is critical; also
note the 2x power price paid for running with higher capacitance.
For example, a 200uH 350pF system with a Q of 50 would require a
steady input of 850 watts to replace losses, and it would require
nearly 1kW of cooling as well. Not very attractive!
In the past I've used ferrite cores up to 2kV and large air coils
for higher voltages. With a special giant bank-wound litz-wire
air coil mounted in a huge box, I managed to keep the capacitance
down and to obtain an overall Q of nearly 400, so that about 100W
of input can sustain 8kV. That's a reasonable loss.
However, I'm once again attempting to accomplish a similar goal
with a smaller coil using ferrite cores. Or at least to settle
the question, what's the best that we can do?
I've found a large U core that might work, Amidon's UA-77-14002.
It's a 4 x 4.5 x 1 inch assembly, with ui = 2000. I calculate
that a stack of six cores, with appropriate gapping, is worth a
try. BUT oops! Amidon tells me that large core is discontinued.
HELP! Anyone know of some _large_ ferrite cores I could consider?
Here's some background material for those who might be interested.
Thanks to a specially-designed litz wire, my primary remaining
difficulties are high core losses and winding capacitance. To
assist in the search for a suitable ferrite core, I derived a few
"figure-of-merit" formulas to evalute the impact of ferrite choices.
Power loss is my biggest problem, P ~ V^2.2 (ui / f L le Ae)^1.1
Winding capacitance to the core is a killer, C ~ le L / ui Ae^1/2
This tells me that increasing le, the magnetic field length, helps
the power loss, but hurts the capacitance (increased wire length).
ditto for changing the inductance, L. However, increasing Ae, the
effective cross-section area, reduces both the power loss and the
capacitance. So choosing cores that can be stacked is a good idea.
These formulas can provide some guidance, but after selecting cores
and designing a winding and gapping the core to set the required
inductance, we need to calculate the exact power loss. First, the
maximum field, B = 10^8 V u^1/2 / f (pi L le Ae)^1/2, in Gauss, where
IIRC, L is nH, and the dimensions are cm. Next the manufacturer's
curves to look up the loss factor. Sometimes a formula is available,
like 7.0 x 10^-10 f^1.15 B^2.23 mW/cm^3. Multiply by area, Ae, to
get the actual loss - ouch, this step often hurts, hundreds of watts!