Detailed Contents
Preface
1 Microsoft Language Utilities
2 Board Installation
3 Operating System Setup
4 Installing Software
5 Configuration Utility
6 Setting up the Video Tape Recorder
7 Other Software
8 Other Hardware
In this section we discuss briefly some aspects of setting up other hardware items commonly used in conjunction with DigImage.
The Cohu 4910 and 4990 series video camera offers a high quality and cost effective option for imaging experiments. There are, however, a number of factors which need to be considered in order to attain the best results.
The Cohu camera contains a number of internal jumpers and external switches.
The optimal setting for the internal jumpers is given in the table below. Note that there are at least four different versions of the camera, not all of which have all the jumpers noted below. Moreover, versions of the camera later to D may have additional jumpers. If in doubt, please contact DL Research Partners.
| Jumper | Preferred Setting | Factory Default | Notes |
| JB1 XTAL/LL | LL (Line Lock) | XTAL (Crystal Lock) | To use the camera in Line Lock you need to have an AC power supply, whereas the factory default Crystal Lock may be used with either AC or DC supplies. While the default Crystal Lock setting offers a slightly more stable image, it can suffer from a beating with the mains frequency, particularly in conjunction with fast shutter speeds and fluorescent lights. Moreover, if a mechanical shutter is desired (see §8.1.2 ) then this may be achieved much more easily if the camera is locked into mains frequency. |
| JB2,JB3 NOR/RESET | N/A | Do not change. | |
| JB4 FRAME/ FIELD | FRAME (Frame Integration) | FIELD | This setting provides the maximum vertical resolution for the camera and, when used in conjunction with a mechanical shutter (see §8.1.2 ) provides the ability to capture both video fields at the same time rather than with the normal 1/50s (1/60s for NTSC) lag between the odd and even fields. |
| JB5 AUTO BLACK | OFF | ON | When activated, this automatically adjusts the "black" level of an image if the mean intensity becomes low. For most applications this, like automatic gain (see below) is detrimental to obtaining experimental measurements. |
| JB6 VD/PRE | N/A | Do not change. | |
| JB60 | N/A | Do not change. | |
| JB80 75 Ohm Sync Sel. | N/A | Do not change. |
For the external controls, the optimal settings are
| Control | Preferred Setting | Factory Default | Notes |
| ELECT IRIS | Off | On | This only plays a role if the optional Electronic Iris board is fitted. The effect is similar to an automatic iris on the lens, except that it adjusts the shutter speed. The net result is that the intensity output by the camera is affected by the mean light level, a situation undesirable for quantitative measurements. |
| Shutter Speed Switch | Off | Off | Controls the electronic shutter. This can be used to "stop" fast motion, but note that as it is applied to each field separately, it does not remove the 1/50s (1/60s) time lag between the two video fields. Better performance may be achieved by leaving in the Off position (which gives a 1/25s or 1/30s integration time when using Frame Integration) in combination with a mechanical shutter. See next section for details. |
| PEAK/AVG | N/A | Midrange | This only has an effect if the Automatic Gain control is in the On position. |
| GAIN | Minimum | Midrange | The manual gain potentiometer is a valuable tool in producing an image of appropriate brightness. However, as the gain amplifies both the noise and the signal, it is preferable to keep the gain to a minimum and increase the light level or lens aperture. |
| AGC (Automatic Gain Control) | Off | On | If "On", the Automatic Gain control will adjust the gain of the camera to provide an almost constant mean intensity to the output signal regardless of changes in the light level. In almost all circumstances this is not desirable. Rather, quantitative measurements require a fixed relationship between the intensity and camera output, so this switch should be turned off. |
| GAMMA | 1.0 | Normal CRT display devices are nonlinear. To counteract these nonlinearities, most video cameras produce the opposite nonlinearity, called a gamma correction, normally of the form of the intensity raised to some power g 0.45. This is, however, undesirable for quantitative measurements. Setting the gamma correction potentiometer to 1.0 effectively turns off this "correction" and makes the output of the camera close to linear in the intensity. | |
| SHARPNESS | Minimum | Minimum | This provides the ability to sharpen or soften an image by introducing a high pass or low pass filter to the video signal. Normally such manipulations are a disadvantage. |
One of the fundamental problems with video technology is that a single frame consists of two interlaced fields transmitted sequentially. The first field contains all the even numbered video lines, and the second the odd numbered lines. The field rate (50Hz PAL, 60Hz NTSC) is therefore twice the frame rate (25Hz or 30Hz). Most modern CCD cameras have effective shutters which open and close at the field rate with an exposure time less than or equal to the field period. This means that the information contained in the even field represents the state at a time 1/50s or 1/60s earlier than the information in the odd field. In a large number of applications this time delay between the two fields is not acceptable, and it is therefore necessary to discard one of the two fields and thus reduce the vertical resolution. In many cases this reduced resolution is acceptable.
When tracking particles it is desirable to keep the particles as small as possible. The lower limit on the size of the particles is imposed by the wish to locate them with subpixel accuracy. Subpixel accuracy requires that the particle extends at least two pixels in both directions. However, if the particles are moving more than approximately 50% of their diameter between one video field and the next, it is possible to use information in only one of the video fields. In order that the particle extends at least two pixels vertically in the reduced resolution of a single field, it must extend at least four pixels vertically relative to a complete frame, thus requiring the particles to be twice the size. Moreover, while this gives subpixel resolution, it is subpixel relative to pixels twice the size.
To complicate matters further, different cameras construct the two video fields in different manners. In some cameras the even field corresponds to the even lines of pixels in the CCD chip, and the odd field to the odd lines of pixels in the CCD chip. In such cameras not only must the vertical resolution be reduced by discarding one of the fields, but because there is no information about what is happening in the other set of pixels, the effective resolution for small entities such as particles is significantly worse than half the full frame resolution.
Slightly better are cameras which produce an average of the even lines and the preceding odd lines for the even field, and the odd lines and the preceding even lines for the odd field. With such cameras it is still necessary to discard one of the fields, but at least the remaining field has information from all the pixels in the CCD chip.
Many more recent cameras, such as the Cohu 4910 & 4990 series, offer the choice of either of the above methods of generating the video fields. In "field integration" mode it uses the average of the even and odd rows of pixels for each field, while in "frame integration" mode it uses only the even rows for the even field and the odd rows for the odd field. With "field integration" the shutter speed can be set anywhere between 1/50s (1/60s NTSC) and 1/10000s. Realistically shutter speeds up to 1/250s are of practical value for particle tracking. Most useful, however, is the ability to have a separate 1/25s (1/30s NTSC) shutter for each of the fields in "frame integration" mode.
For normal cameras the slowest shutter speed is equal to the field period (i.e. 1/50s or 1/60s), with both the even and odd rows of pixels being reset at the same time. Both rows are also transferred to the CCD shift registers at the same time, with the adjacent even/odd pairs of rows being summed. In frame integration mode the camera proceeds as follows: the accumulators for the even rows of pixels are reset at some point up to 1/25s (1/30s NTSC) before the even rows are transferred to the shift registers. 1/50s (1/60s NTSC) after the even rows were reset, the odd rows of pixels are reset, and 1/50s (1/60s NTSC) after the even rows were transferred to the shift register, the odd rows are transferred to the shift registers.
The following diagrams illustrate the timings of the two integration modes. The resetting of the accumulators is indicated by R, and transfer to the shift registers by T. The frame boundary is indicated by F and field boundary within a frame by f. The contents of the shift registers is e for even lines only, o for odd lines only, f for the sum of the even and preceding odd lines, or p for the sum of the odd and preceding even lines. The field read out is given by EEEEE for the even fields and OOOOO for the odd fields, with S representing the vertical sync.
FIELD INTEGRATION: Timing ---F-------f-------F-------f-------F-------f-------F-------f--- Even pixels R--T----R--T----R--T----R--T----R--T----R--T----R--T----R--T--- Odd pixels R--T----R--T----R--T----R--T----R--T----R--T----R--T----R--T--- Shift reg. ppp fffffff ppppppp fffffff ppppppp fffffff ppppppp fffffff ppp Read Out OOOSEEEEEEESOOOOOOOSEEEEEEESOOOOOOOSEEEEEEESOOOOOOOSEEEEEEESOOO FRAME INTEGRATION: Timing ---F-------f-------F-------f-------F-------f-------F-------f--- Even pixels R--T------------R--T------------R--T------------R--T----------- Odd pixels --------R--T------------R--T------------R--T------------R--T--- Shift reg. ooo eeeeeee ooooooo eeeeeee ooooooo eeeeeee ooooooo eeeeeee ooo Read Out OOOSEEEEEEESOOOOOOOSEEEEEEESOOOOOOOSEEEEEEESOOOOOOOSEEEEEEESOOO
The frame integration feature is not by itself a major improvement over standard field integration video cameras. However, if a mechanical shutter is used with the camera in conjunction with a 1/25s (1/30s NTSC) electronic shutter, it is possible to expose the two fields at the same instant in time, thus allowing the full vertical resolution to be used. The diagram below uses the same symbols as those above but with a full frame electronic shutter. The external mechanical shutter is open for the period indicated by MMM.
FULL RESOLUTION FAST SHUTTER: Mechanical Shutter M MMM MMM MMM M Timing ---F-------f-------F-------f-------F-------f-------F-------f--- Even pixels ---TR--------------TR--------------TR--------------TR---------- Odd pixels -----------TR--------------TR--------------TR--------------TR-- Shift reg. ooo eeeeeee ooooooo eeeeeee ooooooo eeeeeee ooooooo eeeeeee ooo Read Out OOOSEEEEEEESOOOOOOOSEEEEEEESOOOOOOOSEEEEEEESOOOOOOOSEEEEEEESOOO
The mechanical shutter is opened while the odd field is being displayed, shortly before the end of the even field integration period, and shortly after the start of the odd field integration field, thus exposing both fields associated with the next frame at the same time.
Extensive trials have shown a slight improvement in the horizontal velocity resolution for rapid flows, and a dramatic improvement in the vertical resolution. Some of this improvement is due to an improved signal to noise ratio when compared with the standard camera used in this trial, but most of it is due to the improved vertical resolution.
There are a number of ways of producing a mechanical shutter. However, the easiest and most effective is to lock the camera in to mains frequency then use a synchronous AC motor to drive a disk with the appropriate number of clear and opaque sectors. We recommend using an 8-pole motor attached to a disk approximately 250mm in diameter with four open sectors covering approximately 25% of the circumference. Further details are available on request.
Producing suitable light sheets is often a difficult task. Many resort, unnecessarily, to lasers if a standard slide projector does not provide a suitable light sheet. The following subsections include brief details of two alternative and cost effective methods of producing light sheets.
Figure 5 shows the physical arrangement used to produce a light sheet from a 1kW halogen photograph lamp, and figure 4 shows the optical paths schematically. Key features are:
Figure 5: Sketch of halogen lamp elevation (top) and plan (bottom).
Figure : Sketch of optical arrangement for halogen lamp.
Arc lamps have a number of advantages over halogen lamps. First, they provide a much higher intensity for a given power input. Second, the light emanates from a much smaller region, thus allowing better collimated. Finally, the colour temperature of the light is higher, which is particular valuable for LIF work.
For a number of years we have been using 300W xenon arc lamps with an integral dichroic parabolic reflector. Higher power 500W and 1kW versions are also available, with the light output of the 1kW version approximately five times that of the 300W version. These lamps are sold as "Cermax" and are made by ILC Technology (399 Java Drive, Sunnyvale, Ca 94089, USA; Ph: +1 (408) 745 7900; http://www.ilct.com/ ).
The beam of light produced by the lamp is well collimated by owing to the large ratio between the size of the arc (around 1mm) and the relatively long focal length of the reflector. The higher power lamps have a larger reflector and smaller arc, producing a higher degree of collimation.
The method of producing a light sheet from the 300W lamp is extremely simple and is sketched in figure 6 . The basic idea behind it is to take the lamp well back from the region to be illuminated, letting the well-collimated beam expand far enough to illuminate the entire length of the sheet required, and control the thickness of the sheet by masking things at the sides of the tank, thus selecting only a tiny fraction of the light, but ensuring it is extremely well collimated (in the cross-sheet direction).
Figure 6: Sketch of optical arrangement for creating a light sheet with an arc lamp.
The improved collimation of the 500W and 1kW versions makes the optics slightly more difficult as it will normally be necessary to spread the light to obtain a sufficiently large sheet. This is most conveniently achieved through a very long focal length cylindrical lens. When using the higher power lamps we strongly recommend that you use a heat mirror or window to reduce the heat content of the light reaching the tank.
Note that for all the lamps there are two sets of light rays. The first set leaves the lamp unit directly, radiating radially from the arc. Such rays effectively emanate from a point source and decrease in intensity as 1/r2. These light rays contain a significant amount of infrared and, as a result, the light beam is very hot close to the lamp. However, at typical working distances of 3m or more from the tank, the light intensity is much smaller than the second set, and the heat content does not represent a problem for many flows. The second set of light rays make up the collimated beam which we make use of. These light rays originate at the same arc, but are directed towards the dichroic reflector. Most of the heat is transmitted through the reflector to be dissipated, while the visible light is focused into the beam.
Warning: These lamps should
never be operated directed vertically upwards as we are told (but
have not experienced) that instabilities in the arc in combination
with strong convection heating the front window can cause the
lamps to explode.
DigImage documentation page
DigImage home page
Stuart Dalziel's home page