Re: Ballasting ..... the never-ending thread ;-)
> Original Poster: Mark Finnis <mefinnis-at-medicine.adelaide.edu.au>
> At 19:20 2/12/98 -0700, Malcolm Watts wrote re: Off-axis primary inductance:
> >I'm intending to take this idea one step further by using a variable
> >reluctance coil. I have constructed two "cores" , both heavily gapped
> >internally by recovering 3C8x ferrite cores from old computer SMPS,
> >smashing them to bits and stuffing the chips into PVC and acrylic
> >piping. The largest of the two is about a foot long, 4.5" diameter
> >and weighs a bundle and has an Al value somewhere around 300nH/turn^2
> >if memory serves. By sliding this core in and out of a modest winding
> >I have a variable inductance. This might also be useful for adjustable
> >primary ballasting with a stiff power transformer. The power losses
> >should be considerably less than the gap. I did experimentally check a
> >primary with one of these cores for losses and it looked better on
> >the scope than an equivalent inductance and much larger air wound
> >primary, gap included.
> Now this has me somewhat inspired. I had ballasted my HBP (home-brew pig)
> with my neighbours welder plus some series R elements with initial success.
> Just bought a new house and am moving end of January, so coilin' takes a
> back seat for a while (but it does have a work-shop <big grin>). Problem
> is my new neighbour is very unlikely to have a spare welder ...... so back
> to the drawing board :-(
> My second-favourite junk shop has *lots* of ferrite rods going cheap. Lets
> say we fill a 4" PVC pipe with these & set the lot in epoxy resin, allowing
> a 1cm channel though the middle to take a threaded rod for position
> 1. How long will I need to make the core ?
See comments at the end of this post.
> 2. How many turns & what guage wire ?
First you have to build a core and measure its induction factor (Al).
Then turns = SQRT(L/Al) Core losses at 50Hz will be very low. The
core would need ??? Ampere-turns applied to come even close to
saturation because it is heavily gapped end-end. The losses will also
be comparably low at "high" break rates - those being a fraction of
the frequency such ferrites were designed to run at. In choosing
winding wire and knowing the maximum target inductance, the
turns*PI*mean diameter gives the wire length and knowing the RMS
current, you can choose wiresize to give a nominated copper loss
(R*I^2). The ferrites I used have very low hysteresis/power losses
when operated at Bmax of around 250mT, typically 800W/m^3 at a
breakrate/frequency of 1000 and 20 degrees C. For the core I
described, volume = 0.3*0.06^2*PI m^3 = 0.004m^3 which makes core
loss at this flux density and frequency around 3 - 4W !!!!
Ampere-turns applied to this core area is what governs flux
density. Knowing core area, turns, inductance and peak current you
can work this out: Bmax = (L*Ipk)/(N*Ae) where Ae is the core cross
sectional area. Note that *peak* current is required in this
> Given my aims are to limit current over the range 5-35 amps (more likely
> 10-30A), with primary mains at 240V/50Hz.
> Thinking further. Could we make it whole set-up shorter by using several
> winding layers, or would heat dissipation be a problem ?
Depends on copper losses which can be arbitrarily reduced by
increasing wire diameter. I would definitely use a multi-layered coil
as then you can get away with a shorter core. I would allow a
core/winding overlap of an inch or so at each end to maintain
I haven't used this core in anger yet so can't comment further. I
obtained the bulk ferrite loss figures from the Philips core data
book. They have worked reliably in practice for other ferrite chokes
and transformers I have designed using the equations above. Comments