Stoneridge Engineering Home |
Sorry about the delay in getting back to you. For the following discussion I'll assume that the dump resistor bank is directly across the magnet and your SCR switches (and commutation SCR) are in series with the magnet. Some possible options include using a high current mechanical switch alone, a mechanical switch in combination with a delay fuse, a mechanical switch in parallel with a semiconductor switch bank, or a bank of high current semiconductor switches (IGBT, Thyristor, or GTO) to handle the current entirely. Although a specially built electromechanical contactor using Ag or Ag-Wu contacts would have low steady-state conduction losses, it may be a bit of a challenge to accelerate the mass of the movable contact assembly so that it can fully open within 20 milliseconds from when you detect a quenching event. For now, let's assume that this problem can be solved. By paralleling the contactor with an expendable fast-acting high voltage fuse, you can prevent arcing of the contacts upon opening. Instead, the fuse would actually open the supply current to the magnet. High voltage fuses are designed to quench the arc the fuse element opens. However, for this high current DC application, you would likely need to use an explosive "expulsion fuse" to quickly quench the arc - and these are very noisy! The sizing of the fuse would probably need to change as a function of the magnet current, reducing flexibility and adding to maintenance costs. Suppose instead you used a high current semiconductor that is turned on and connected across the contacts. If you mechanically (or optically) detected that the main contactor had opened sufficiently, you could then command the semiconductor switch to turn off. In this case, the full magnet current would only be applied to the semiconductors for a relatively short interval, say 20-25 milliseconds. This would allow the semiconductors to briefly handle a much higher peak current than their average current rating. And, by "snubbing" the contactor with a semiconductor switch, you would retain the benefit of opening the contacts "cold", and you'd need only two-three semiconductor devices to handle the brief switching load. This is the approach I would recommend, since it's efficient and energy efficient. I would not recommend using a modified brute force approach with SCR's since the higher current and voltage requirements simply make current commutation even more problematic. Instead, Gate Turn Off (GTO) devices, Gate Commutated Thyristor modules (GCT - a GTO with a control module), or IGBT's might be viable alternatives. GTO's are turned on by a positive gate current just like an SCR. However, GTO's can also be turned off by briefly pulling a much larger current (typically ~20% of the anode conduction current) out of the gate. Although this complicates the gate drive circuitry a bit, it also allows you to use standard high current devices that are used by the electrical power and locomotive manufacturing industries. Since gate control of the GTO is done at a comparatively low voltage, you'd no longer need to use bank of high voltage electrolytic capacitors to force anode commutation. However, you will need to provide a high current negative polarity gate turn-off pulse. Some manufacturers (ABB, Powerex/Mitsubishi) provide integrated and optically isolated gate drive modules to provide the appropriate gate drive for mating GTO devices in order to create GCT modules. These modules can then be synchronized via fiber optic inputs and electrically stacked in series to handle large voltages. This is commonly done during HVAC-HVDC interconversion within the power transmission industry. IGBT's are also available with up to 1500 amp capability at 2500 volts (Westcode), and 1400 amps at 1700 volts from Powerex. These devices are even simpler to drive, requiring only a control voltage. At similar current levels, SCR's, GTO's and IGBT's tend to have similar forward voltage drops. Unfortunately, the voltage drop for these devices under full load is in the range of 3-4 volts. This can result in considerable (60-80 kW) steady state power dissipation if they are configured to continuously pass full magnet current. A significant number of devices would also need to be connected in parallel to handle the total magnet current, and you'd need to insure that the current was being evenly shared across devices. Although this approach would eliminate the contactor, it also adds significant cost and energy inefficiency to the system, and I do not recommend it. Finally, if your superconducting magnet can withstand the voltage stress, you may also be able to increase the dump resistor bank value since the switching semiconductors could withstand considerably more than 1,000 volts. Obviously, using a higher value more rapid dissipates the magnet's energy, potentially more LHe. In any event, it sounds like an interesting design challenge! Here are some starting points: Hope this helped and best regards, -- Bert --
http://www.pwrx.com/
http://www.abb.com/semiconductors
http://www.westcode.com/
Message Thread
« Back to index | View thread »