Pharsighted E9 series ultra high-speed cameras

Selection should begin with the required full-frame rate, minimum exposure and recording duration. The E9·150S provides 489,000 fps at 640 × 480 and a 34 ns minimum exposure, while the E9·100S provides 326,000 fps and 59 ns; the E9·80S and E9·50S provide 272,000 and 164,000 fps respectively, both with 98 ns minimum exposure. Reduced-height rates extend to 2.72 million fps on the E9·150S and E9·100S, 2.457 million fps on the E9·80S and 1.98 million fps on the E9·50S. Internal memory also scales from a maximum of 216 GB on the E9·50S to 648 GB on the E9·150S. Applications requiring shorter exposure or more full-frame temporal resolution will normally point towards the E9·150S or E9·100S, whereas lower-rate events may not require their additional capacity.

The large backside-illuminated pixels can reduce the illumination burden, but they do not remove the need for suitable lighting and optics. The sensitivity ratings are ISO 200,000 for monochrome and ISO 80,000 for colour, while monochrome EMVA characterisation at 630 nm measured 68.2% quantum efficiency, 49.94 dB dynamic range and 21.1 e⁻ temporal dark noise. Because the output is digitised at 9 bits, highlight clipping and exposure placement still require attention, particularly with flames, arcs or bright reflections. Lens aperture, spectral transmission, illumination wavelength and the selected exposure time will all affect the usable signal level. Quantitative techniques such as PIV, DIC and spectroscopy should therefore be validated using the actual optical path and acquisition settings.

At 640 × 480, the smallest memory option for each model provides approximately 1.14 to 1.38 seconds of recording. The corresponding base values are 1.15 seconds for the E9·150S and E9·100S, 1.38 seconds for the E9·80S and 1.14 seconds for the E9·50S. Selecting the second or third memory option doubles or quadruples the recording time and frame count, while reducing image height also increases duration; at 640 × 16, base-memory figures range from 2.81 to 5.47 seconds. The circular buffer allows the event to be positioned before or after the trigger, with a minimum trigger offset of two frames. For repeated tests, the memory can be divided into as many as 1,024 partitions and combined with automatic saving to the SSD.

The cameras support PTPv2, IRIG-B, external frame sync and hardware triggering. PTPv2 operates through Ethernet with specified accuracy below 10 ns and allows an E9 camera to function as either a grandmaster or an ordinary clock, while IRIG-B accuracy is below 1 µs when used with a suitable time source. External frame-sync edges can be received through one of the six configurable user I/O ports, and those ports may also carry event, ready, record, shutter or trigger signals. The dedicated trigger input accepts compatible logic signals and voltages up to 28 V, whereas the general user inputs must not exceed 5 VDC. Electrical levels and port assignments must therefore be checked separately when designing the interface to a test controller or data-acquisition system.

The manual Nikon F/G mount is the standard configuration, with active Canon EF and 1-inch C-mount alternatives available. Full-frame Nikon or Canon EF lenses are required to cover the complete 33.28 × 24.96 mm sensor area; APS-C lenses may not provide full coverage. The active EF option enables electronic aperture and focus control through SLOW, while the Nikon F/G mount provides mechanical iris control where the lens supports it. A standard sub-1-inch C-mount lens has an image circle of roughly 16 mm, so vignetting may occur at larger image heights, and the internal 58 mm filter position cannot be used with this mount. Lens weight must be supported independently, and back-focus should be checked whenever a new mount or internal filter is installed.

The camera weighs 10.88 kg and measures 240 × 212 × 175 mm, so its support structure should be designed before the optical layout is finalised. Tripod installation requires at least two 3/8-16″ studs and a tripod rated for no less than 14 kg, with additional allowance for lenses and accessories; monopods and bipods are unsuitable. Active cooling draws air through the left side and exhausts it through the right, so neither path may be obstructed, and side mounting should place the left side down. Primary and backup inputs accept 24–28 VDC, with the backup input able to maintain the volatile recording buffer if primary power is lost. Model power requirements extend to 230 W with an SSD on the E9·80S and E9·100S, while final model-specific E9·150S figures require confirmation.

The .SLOW format should be retained when raw sensor values, camera settings and frame metadata are required for later processing. SLOW can convert recordings to AVI, MP4 or TIFF and can generate CSV files containing frame numbers, timestamps, timing-lock indicators and event states. Image and colour adjustments remain metadata when working in .SLOW, but they are applied permanently to pixel data when exporting to another image or video format. Each camera includes an approximately 1.6 TB U.2 NVMe SSD, and recorded takes may be saved automatically after capture or transferred through the USB 3.2 Gen 2 dock. A practical workflow is to preserve .SLOW files as the engineering record and create separate converted copies for analysis, reporting or presentation.

The camera supports 1, 2.5, 5 and 10 Gigabit Ethernet through an RJ45 connection, with Cat 6A cabling suitable for normal installations. It obtains an address through DHCP when a server is available or assigns a link-local address in the 169.254.x.x range when connected directly. Camera discovery uses UDP port 8890, while control commands use TCP port 80, which matters when routing or firewall rules are applied. The Ethernet cable should be connected before power-up, and the camera requires approximately two to two and a half minutes to boot; connecting the cable afterwards can require a restart before detection succeeds. The host network interface, storage subsystem and cabling should all support the selected link speed, as the interface rating alone does not define end-to-end download throughput.