Once the current has risen to its steady state value, large power supplies take over and maintain the current. Remotely
controlled, they are set for the required constant current (CC) and a voltage about 0.5V above what will be required.
Once set, they are left as is. At the end of the charge phase and the coil voltage has risen to its predicted steady
state value, diodes switch in the large power supplies. The switchover is seamless, clean and simple.
Originally, HP HP6292B (40V, 50A) supplies were used with two in parallel for the MOT coils and one for the transfer
coils. Disappointingly, these were found to drift over hours and have been replaced by Kepco ATE15-50M (15V, 50A)
in the same configuration. Of the two MOT supplies, one is remotely controlled with the other slaved to it.
The magnetic fields have to be removed from the traps fast, in well under 1mS. Because the magnetic fields
are maintained by the coils and the magnetic field strength is proportional to the coil current, removing the
current from the coil will collapse the magnetic field, excluding eddy currents from nearby metals. The problem
is to remove the energy from the coil, fast. Because coil currents continue after an external current source
has been removed, the only way to dissipate the coil's energy is by making it dissipate power, by forcing as
high a voltage as practical. Using the conventional equation relating voltage, inductance and current in
which the voltage V is the dependent:
V = L * dI/dt where V is the coil's counter-EMF, L is the coil inductance, I is the coil current
then making V independent and dI/dt the dependent:
dI/dt = V / L
The rate of reduction of the current, dI/dt, is constant as long as I > 0, causing the current to reduce to zero linearly.
By maximizing V using TVSs (Transient Voltage Suppressors) to about 1KVp, dI/dt is also maximized, reducing the
current to zero as fast as possible.
TVSs are similar to zener diodes. Above a certain voltage, they conduct. The difference is that
TVSs can absorb lots of power for short amounts of time. The limitation is that the breakdown voltages
are lower that what is required for this project. However, conveniently, TVSs can be stacked in series to increase
the apparent breakdown voltage. Two parallel stacks of twelve TVSs in series limit the flyback voltage
V to about 1KV. The TVS used is Micro Semiconductor's
15KP60A. Its breakdown voltage is nominally 70.4V, peak currents are 154A, rated peak power is 15KW (assuming
an exponential decay with half power at 1mS or less), On resistance typically 0.18 Ohms. In this project
with a peak design current of 100A, the peak power per TVS is a mere 8.8KW for <<1mS. At 100A, the
actual voltage drop across each TVS will be 70V plus the IR drop of 18V (from 50A and 0.18 Ohms).
The switching device used is an IGBT (Insulated Gate Bipolar junction Transistor). This type of semiconductor
is a cross between a conventional BJT (Bipolar Junction Transistor, AKA "transistor") and an enhancement mode
MOSFET (Metal Oxide Shield Field Effect Transistor). Although retaining the worst of both worlds (poor
saturation Vce-sat, high Miller capacitance Ccg and its other effects), they are capable of handling high power.
The IGBTs used, International Rectifier's
IRG4PSH71KD-ND, can work up to 1.2KV, 78A, 350W and 150°C. The 1.2KV determines V in the above
equation and ultimately the time to collapse the magnetic field. The other parameters are not being pushed.
An oscilloscope screen capture is available of the
MOT coil current falling from 60A to zero in 149uS.