From: John Popelish
Organization: This space not available for advertising.
X-Mailer: Mozilla 4.7 [en] (Win98; U)
Subject: Re: 12 vdc to 6vdc @ 30amps
References: <firstname.lastname@example.org> <3D8295F3.6B10E9CB@rica.net> <3D82A7ED.DDEE6E7C@rica.net> <3D83E001.338C233E@rica.net>
Date: Sun, 15 Sep 2002 19:04:40 GMT
NNTP-Posting-Date: Sun, 15 Sep 2002 15:04:40 EDT
Winfield Hill wrote:
> John Popelish wrote...
> > John Popelish wrote:
> >> No, it is just a self oscillating autotransformer DC to DC converter.
> >> It has no short circuit protection. A fuse should work, though.
> >> Here is what I had in mind. I switched from junction transistors to
> >> fets and reduced the parts count a bit.
> >> +12v
> >> +-----+-----+
> >> | | |
> >> A M A
> >> | . | . |
> >> +-mmm-+-mmm-+
> >> | |
> >> +---\ /---+
> >> |D \ / |D
> >> --| / |--
> >> |--/ \--|
> >> --| |--
> >> |S |S
> >> +-----+-----+
> >> 0v
> >> (I picked parts from Digikey catalog)
> >> Where A is one of two schottky diodes in an I.R. 47CTQ020 (40 amp
> >> 20 volt dual), M is the motor load, mmm-+-mmm is a toroidal core
> >> with a bifilar 14 ga. winding (as many turns as will fit) on a
> >> shield bead like a Panasonic KR16TT281815 28 mm O.D., 18 mm I.D,
> >> 15 mm height, And the two Mosfets are IRF1010E, 60 volt, 0.012 ohm.
> >> The one I made with junction transistors was well over 90% efficient,
> >> and I think this might beat it.
> > P.S. that twinned black and red speaker wire is pretty good for the
> > bifilar winding for this tapped saturating inductor. The insulation
> > takes up some room, but the wire is flexible and tough. You can also
> > take two large shield beads, side by side, and loop the wire through
> > the two holes to get more core cross section and protect the winding.
> Very interesting. Are you sure this will work as drawn with
> FETs? There's no equivalent to a capacitor paralleling a base
> resistor... Talk us through determination of the oscillation
> frequency, as well as a transition from one FET full ON to the
> other full ON.
> Q. Is the transformer's leakage inductance our friend, or enemy?
I think that perfectly coupled windings are most efficient.
The period is based on saturating the core.
> BTW, I generally prefer driven oscillation: easier to think
> about and control, and well suited for lazy engineering. :-)
(snip your version of this circuit)
The circuit operates on this concept (approximately):
When the power is first connected, the motor passes current through
the auto transformer to both fet gates and drains, causing both fets
to begin to conduct simultaneously. However, the transformer and
crossed gates produce a loop gain of much more than one, so one fet
drives the other on more which drives the other on less, and one goes
ot saturation, while the other switches off. This causes only one
half of the transformer to see the motor current, and would put most
of the supply voltage across that half of the winding. But as soon as
the volts per turn produce about 6 volts across the driven half, the
other half also produces about 6 volts and forward biases a diode,
limiting the total voltage across both halves to 12 volts, so the
center tap produces a 6 volt node. The motor, thus sees the
difference between the 12 volt source and that 6 volt node.
This continues as long as the core magnetization continues to rise.
Once the core saturates, the drop across the winding goes toward zero
volts, and the motor sees a spike of most of the 12 volts (briefly).
At that moment, both fets are brought back into a high resistance but
non saturated state since the gate voltage on each is what is left of
the 12 volts (less than 6). At that point, the stored energy in the
transformer core switches which fet is on, and the core again sees a
change in magnetization, and the windings produce about 12 volts
across them, but in the opposite direction.
I have not built this 6 component (counting the motor and bypass
capacitor) power multivibrator using mosfets. I have built one that
used junction transistors and pair of base current limiting
resistors. After running a trolling motor successfully with it, I
brought the idea to a design review at the GE printer business, when
they were trying ot figure out how to recirculate stepper motor
flyback energy to the supply, instead of just dumping it. This
multivibrator eventually ended up in thousands of printers, and
knocked something like 120 % off the power requirement of the power
supply. I would like to see someone who has the Spice capability to
model a saturating inductor see what the fet version can do.