Ultra-high-speed framing cameras have long been the workhorse for acquiring elusive data in scientific fields where very fast phenomena are studied, such as high-voltage discharge, crack propagation, detonics, and impact studies. Today these cameras are also increasingly used in the fields of biomedicine, nanotechnology, and space research, where the ability to “freeze” the motion using very short exposure times (typically rates of 500,000 frames per second or more) gives access to valuable new information.
With all of these applications, the quest for detailed analysis means that the images must be of the highest quality and resolution. Once considered temperamental technology, the high-speed camera has been developed into a simple "plug and play" system that can be used by non-experts as well as dedicated users. In this article, we will review the technology behind ultra-high-speed cameras and provide examples of how the tools can be best used.
Technical Developments: A Simplified Optical Path
Traditionally, multiple-channel fast framing cameras have used rotating mirrors, plate glass beamsplitters, or pupil split optical systems to direct the light either to film or individual image sensors. Each of these optical system designs came with disadvantages such as vignetting, parallax, astigmatism, and large instrument size. In the last five years, demand from the scientific community has driven companies such as Specialised Imaging Ltd. to develop a new set of multiple-channel fast framing cameras with technological innovations that address previous limitations.
Using sophisticated optical design tools enabled Specialised Imaging Ltd., for example, to create an optical design where the optical path of its SIM series camera was folded, while maintaining equal transit times on all 16 viewports, resulting in a compact model. As it is not possible to predict the wavelength of light that enters into the beam splitter, special attention was given to the lens design so that the focal plane is coincident for all visible wavelengths (350-900nm).
In the SIM optical design, each of the viewports is populated by a sensor to capture the image. With some fast framing cameras, the sensors are conventional interline transfer charged-couple devices (CCDs), which can be electronically shuttered down to a few hundred nanoseconds. Electronic shutters, however, have limitations in so far as the shuttering signals have to be synchronized to the readout of the sensor, which can result in long delays from an asynchronous trigger to the point at which an image is captured. The very nature of electronic shuttering also means that each sensor can only acquire one short-exposure image. If a second short-exposure image is required, it can only take place after the first has been fully read out from the sensor, which can be up to 200 milliseconds later.
Another high-speed imaging method fits each viewport on the beamsplitter with an 18mm high resolution image intensifier coupled to an interline transfer CCD sensor. Intensified CCD (ICCD) sensors can have a 6μm pitch micro-channel plate (MCP) (see Figure 1), which provides a staggering 50 line pairs per millimeter of spatial resolution. The sensors are also capable of exposures as short as 3 nanoseconds.
ICCDs overcome the timing limitations of interline CCDs because the sensor remains in darkness while the intensifier is gated off, thereby allowing the sensor readout to be re-synchronized as soon as a trigger is received. This arrangement allows the first short-exposure image to be taken as soon as 50 nanoseconds after an asynchronous trigger. Similarly, because there is no ambient light between exposures, the image intensifier can be gated a second time as soon as the charge has been transferred into the interline readout registers to allow two images to be taken in quick succession by each ICCD. In the SIM type of cameras, this time has been reduced to around 500 nanoseconds.
The light-sensitive photocathode of the ICCD converts incident photons into electrons, which are then multiplied within the micro-channel plate (which has approximately 6 million channels), before hitting a phosphor screen, which converts the electrons back into visible photons. The electrons, in turn, are relayed onto the CCD sensor by a fiber optic stub.
The ICCD sensor mechanism provides three benefits:
- Short Exposures - Exposures as short as 3ns are produced by rapidly turning on and off the photocathode, which then functions as a very fast shutter. The phosphor screen at the output of the intensifier has a relatively slow decay, which allows the CCD sensor enough time to capture the image.
- Increased Sensitivity - The MCP multiplies the electrons and can provide variable gain as the potential across the plate is varied. The light amplification is invaluable when taking very short exposures, but this gain control also allows the sensitivity of each of the ICCD units inside the camera to be matched, and provides the fine control necessary to vary the sensitivity from frame to frame when taking images of events that have rapid increase or decrease in intensity, including explosive events, plasma, or electrical discharges.
- Resolution - The combination of the SIM system’s fine-pitch MCP and a fine-grain, high-sensitivity P43 phosphor, carefully coupled to the CCD sensor using a low-distortion 4μm fiber optic stub, has resulted in a spatial resolution in excess of 50 lp/mm.
The analog video signal from each channel's CCD sensor is digitized to 12 bits using a dedicated video signal processor, which is located close to the ICCD to minimize any noise and interference, before being transmitted to the main control electronics. A large state-of-the-art field programmable gate array (FPGA) device then multiplexes the eight separate 12-bit data channels into a single block of high-speed memory where the image sequence is held until required for download to the controlling PC. Dealing with this amount of high-speed data requires very careful PCB design and layout to avoid the introduction of unnecessary noise.
Electronic Integration and Synchronization
Once a trigger is received, it is fed into a high-speed (1 GHz) timing generator, which closely controls the gating of all of the ICCD channels to give up to 1 billion frames per second, admittedly with overlapping shuttering times. A novel, RAM-based timing generator is implemented within an FPGA to allow leading-edge to leading-edge shuttering pulses from 1ns to 10ms. Complex algorithms offer independent control of each ICCD channel with exposure timings from 3ns to 10ms in 1ns steps. The timing signals are then converted into a train of high-voltage pulses to turn the image intensifiers on and off at the appropriate times.
The backend optics and electronics are important; however, the most important feature is simplicity, as these systems should be regarded as tools and not science projects. To this end, the majority of ultra-high-speed systems are USB 2.0, Camera Link, serial, or Ethernet controlled. The interfaces are built into the computer (although most modern laptops now do not come with a serial port as standard). Camera Link does require a specialist card, which can cause conflict during installation and subsequent operating system updates; it is ruled out with most laptops. As the computer industry adopts faster and faster speeds, Ethernet control is likely the best plug and play interface, particularly where an experiment, such as one that uses very fast or explosive subjects, is located at some distance from the controlling computer.
Capturing events that develop over a few seconds requires careful control, but imaging phenomena that last a millionth of a second or less require both extremely fine control of the timing, as well as a sophisticated triggering system. The triggering system has to remain simple yet flexible enough to accommodate the vast range of trigger sources available in the modern laboratory, including optical sensors, strain gauges, make-and-break circuits, or inductive pickup from high-voltage signals.
The triggering system of modern cameras must be capable of accepting any electrical signal in the range of -50V to +50V and should be sampled at a very fast rate so as to reduce the trigger jitter. For example, the SIM camera samples the trigger signal with a 1GHz clock, offering repeatability with a jitter of less than 1 nanosecond. The SIM camera’s triggering is further enhanced by the SURESHOT facility, which utilizes two separate trigger inputs so the exact moment to start recording can be pinpointed, even when test subject velocities are variable or unpredictable.
New Scientific Applications
An application study demonstrates how technological advances in ultra-high-speed imaging technology are opening the door to scientific breakthroughs in a growing number of fields.
For space impact studies, like the analysis of hypervelocity impact into regolith simulant material, a 16-channel version of the SIM camera was used with variable inter-frame time (from 10μs to 25μs) and 20 ns exposures (see Figure 3). Two RF500 flashlamps created the backlighting, and the field of view was approximately 45 (H) x 60 (W) mm.
Under vacuum conditions, a 25 mg projectile impacted regolith simulant at 2.8 km/s. An electrothermal gun accelerated the particle, and the camera was triggered by the shot with a preset time delay. The total imaging sequence was 400 μs. Details of the ejecta cloud and the rebound of the projectile can be seen clearly in Figure 2. The multiple exposure capability of the camera, used in the first two images of the sequence, allowed an accurate measurement of velocity to be taken.
Studies of microscopic phenomena have also traditionally been a problem with earlier generations of high speed cameras, due to the often small working apertures and limitations of the optical systems when used with collimated light. The SIM camera optical design has made it possible to fit high magnification lenses or interfaces to microscopes without the need for additional optics. Therefore, studies of micro-phenomena, such as cavitation in micro bubbles, material studies on nano-materials, or micro exploding bridge wires can be hassle free.
Ultra-high-speed cameras have come a long way since the analog devices such as the Hadland IMACON700 and Cordin rotating mirror camera. With advances in electronics, and especially in FPGA technology, the cameras are getting smaller, faster, and much more reliable. It is important for the camera manufacturers to make these systems easier to set up and provide a robust triggering system so that cost of the experiment is controlled and data is captured each and every time. The next generation of cameras must give even more images at the very highest rates and even better sensitivity.
This article was written by Wai Chan, managing director at Specialised Imaging Ltd. (Tring, UK), and Keith Taylor, Technical Director of Specialised Imaging Ltd. For more information, Click Here