Xenon Strobe

In pulsed power, the strobe is a workhorse. Xenon strobes illuminate the picture taken with a camera as well as pump the lasers at the National Ignition Facility. When a very bright and short pulse of light is needed, it is often achieved with either a spark gap or a flash tube. This work uses a flash tube as a consistent, safer, and brighter source of light than an open arc gap.

Xenon emission spectrum from "Perkin Elmer High Performance Flash and Arc Lamps"

Xenon emission spectrum from “Perkin Elmer High Performance Flash and Arc Lamps”

The Xenon lamp is an ideal light source for pulse applications because it has a nearly flat emission spectrum from the UV into the infrared. The spectrum of the flash comes from the current density in the lamp. At low current densities, the xenon emission spectrum exhibits discrete spectral lines; however, as current density increases, the emission becomes more continuous until eventually approaching the spectrum of a blackbody at 9800K. When xenon is pulsed, the spectrum is flattened in the visible range.

Using the lamp as a flash tube requires careful consideration of the high power, fast transients, and high voltages. When a flash tube is off, the resistance is very high. In order to create a flash, the xenon gas has to be ionized. This is accomplished with a high voltage trigger pulse delivered to a contact on the outside of the bulb. The ionization process occurs in several steps. At the high voltage pulse, a small amount of the gas ionizes in the center of the tube creating a streamer. During the next several microseconds, the streamer expands until it fills the tube. At the point the tube becomes fully ionized, the associated resistance drops to near zero. The behavior of each one of these phases is governed by the pulse polarity, the series inductance, and a transient inductance. Following the pulse, the light is extinguished when the supply voltage drops too low to sustain the arc or when interrupted by a semiconductor switch.

Emission modes of a xenon flash tube

Emission modes of a xenon flash tube

The high current of the arc exhibits a self-inductance making it difficult to turn off. The point at which the arc is interrupted also affects the ability to turn it off. For a 10us flash, the tube has been fully ionized for less than 3us and the voltage in the capacitor remains high. In order to fully interrupt such a current pulse, a pulse-power insulated gate bipolar transistor (IGBT) is used. The IGBT is connected in series with the tube and is turned on as the tube is ionized. The flash tube is extinguished by opening the IGBT and then letting both the tube and the IGBT cool down briefly before the next flash. This cool down period is important to reduce the temperature of the IGBT as well as the flash tube. The specified maximum power for the chosen Xenon tube is 50W. This is an important parameter for design as it is the energy at which the tube will explode in just a few flashes. Keeping the single flash power to approximately 30% of the explosion energy maintains a reasonable service interval and keeps the operator safe.

Trigger and extinguish circuit

Trigger and extinguish circuit

As with the LED circuit before, the strobe should not cause interference with other circuitry. The transients should not affect the delicate electronics of the carrier lifetime circuitry or propagate back through the mains. A very effective way of improving isolation with the 120V mains is through the use of a transformer. There is no direct connection to the mains, reducing feedback as well as making the circuit safer to work on by virtue of being isolated. It is impossible to separate the trigger circuit and the high voltage rails; however, the low voltage components are separated on different transformer windings. A low voltage timer circuit triggers an optocoupler providing further physical isolation from the high voltages and currents of the main circuit. Each of these steps help isolate the system electronically and physically to reduce propagated interference and ensure long term survival of components.

At this point it is important to mention the safety issues with this setup. The whole circuit board has to be designed to handle high voltages and currents. The trigger pulse is nearly 6,000V giving it a tendency to jump from wires and components. Keeping the current inside the traces and components proves to be a challenge at this voltage. Common soldering and layout techniques used for low voltages become not only unacceptable, but dangerous. All solder joints must be „balled‟ to reduce the number of sharp points from which a corona could form. As with the solder balls, the board traces should be rounded and all sharp points eliminated. It is also essential to provide physical separation between components to reduce the chance of arcing.

The most challenging voltage to contain is the 6,000V trigger pulse. Even though 6,000V is a high voltage, the energy of the trigger pulse is limited by the value of the trigger capacitor. Using a 1uF capacitor keeps the single high-voltage pulse energy low. Unintended arcing to the technician during diagnosis remained a nuisance but it is not deadly. A potentially dangerous condition can occur if the arc from the 6,000V trigger circuit provides a path for the 350V (medium-voltage) rail. This rail idles around 350V and stores 5 Joules of energy. The most dangerous components in the entire circuit are those connected to the medium voltage rail. A discharge through an arc that reaches the 350V rail forms a direct short and is highly destructive. Physically contacting the 350V rail can easily melt tools and should be considered deadly. To improve the safety of the medium-voltage rail, high-value bleeder resistors are placed across the capacitors to slowly discharge them and all components were isolated in a metal box. Nevertheless, even when the circuit has just turned off, it should be considered lethal. Servicing the circuit can be safely accomplished only after being disconnected a minimum of five minutes.

This strobe circuit has three voltage requirements: 350V for the rails, 18V for the optocoupler, and 12V for the timer. The 6,000V is derived from an inductive pulse transformer connected to the 350V rails. Unfortunately, the timer could not be connected to the 18V line and the optocoupler could not operate on 12V. A unique solution for these multiple outputs came in the form of a tube transformer. Class-A tube amplifiers for guitars still use vacuum tubes and a market for these products still exists. The grid bias in vacuum tubes is around 500Vac center-tapped (+/- 250Vac) and the tube heater or filament outputs are 5Vac and 6Vac. When rectified, three DC voltages of 350V, 18V, and 9V were available and isolated.

Layout of the board was done to reduce the parasitic inductances as much as possible while providing physical separation of the high and low voltage components. The 350V is rectified from the center-tapped output and fed into high-voltage capacitors with low internal inductance. Traces to the flash tube are wide with no ground plane beneath them to further reduce inductance. The trigger transformer was placed close to the flash tube reducing the distance travelled by the 6,000V pulse. Several ground traces had to be lifted manually after the board was fabricated to provide additional clearance for the trigger pulse. When triggered, an arc would form to the ground plane or to secondary components causing unintended, albeit temporary, fires. In some cases, components got hot enough to melt and board traces exploded. Manual re-routing of board traces and component placement eliminated these safety concerns; nevertheless, a more elegant solution would include repeating the board design to further increase physical separation and eliminate the arc risks.

Operation of the strobe begins at a pair of 555 timers just like with the LED setup. The system can be operated in two modes: single-shot or a series of shots operating at 2Hz. When a pushbutton switch is depressed for single-shot, the first edge triggers a monostable timer for debouncing. The monostable pulse triggers a second monostable timer that controls the strobe output pulse. Debouncing the switch prevents the tube from stuttering on and off rapidly. Pulse width of both timers is adjustable by potentiometer. The second pulse triggers the optocoupled gate driver to control the final xenon pulse width.

Pulse shapes and switch debouncing

Pulse shapes and switch debouncing

Driving the IGBT to turn on and off the inductive load is made easier with a special optocoupler chip designed for applying gate drive. Faster current drive to the IGBT improves the turn on and off characteristics. The IGBT is designed to withstand 180A pulses with a total power dissipation of 378W. By optically coupling the input and output sides of the gate driver chip, additional separation between the high-current pulse and the low-voltage control electronics is created. Even with all these steps, both the IGBT and the flash tube get hot within 10 seconds if using a repeating pulse train. The highly inductive nature of the Xenon arc is challenging to turn off quickly. Due to the thermal limits of the devices, the circuit is only used in single-shot mode.

The full circuit for the strobe and triggers

The full circuit for the strobe and triggers