Occasionally (perhaps once a day or once a week) the camera will get in a strange mode and the images and values will be unexpected. First, try the "Reset the Camera" widget on the right column of the Acquire window." Next, try restarting NeuroPlex. If that doesn't solve the problem, try rebooting the PC and cycling the camera power. If it is convenient you might start each day with a PC reboot. After cycling the camera power, the camera setting may not be the same as the setting indicated in the Acquire widget. You will need to cycle the frame rate to a new rate and then back to the one you wish to use.
Both the voltage output of the CCD camera and the signal-to-noise ratio are dependent on three factors: the frame rate (Kfps), the camera gain setting, and the on-chip binning. Attention to the effects of these factors is necessary to maximize the resulting signal-to-noise ratio and to minimize saturation.
rates are not limited to the basic rates given in the list of
configurations. The software allows the user to add time to the basic
frame interval. For eample - to achieve a frame rate of 25Hz one should load a
basic rate of 40Hz and add 15ms to it.
The camera has four NOMINAL gain settings: X1 , X3 ,X10 and X30 . The high gain is about 30 times larger than the low gain.
The dark noise (actually read noise in the cooled SciMeasure CCD) is affected by the frame rate and the gain. The read noise values for all gains are given in the following table. The read noise is somewhat higher at the lower gains. This should not be a problem because shot noise will dominate at higher light levels.
The gain setting also affects the usable well size. At a gain of 1 db the full well size is available. At high gain (30db) only a fraction (~ 20%) of the well size is available.
Optimal Use of the CCD Camera
The S/N ratio at moderate and high light levels is the square root of the light intensity. To increase the S/N one should pump in (and collect) as much light as possible at the lowest gain (1X). Only at this gain (1X) one uses the full well-size of the CCD and this is the preferred mode of measurement for most bath-stained tissue, such as brain slices and cardiac preparations. The saturation at higher gains (>=3X) is probably amplifier saturation. If saturation occurs at 1X, try to use a higher frame rate (up to 2KHz for the NeuroCCD-SMQ, unbinned). This will allow emptying the CCD wells faster, thus avoid saturation w/o losing light. One can temporally bin the data later to increase the S/N ratio. If saturation still occurs when using the highest frame rate, one will have to reduce the illumination. Higher gains should be used at dim light levels to better utilize the 14-bit digitization.
On-chip binning during acquisition reduces the spatial resolution but also reduces the relative dark noise.
These effects result in the following
For lower light levels:
1. Use the high gain setting.
2. Binning during acquisition will increase the signal-to-noise ratio.
3. Slower frame rates will improve the signal-to-noise ratio by reducing the relative dark noise.
For higher light levels:
Saturation can be minimized by using low gain, faster frame rates, and not binning during acquisition. At higher light levels the noise is dominated by shot noise. In this situation binning after acquisition is just as effective as on-chip binning in improving the signal to noise ratio (while minimizing saturation)
For images with high contrast (both high and low light levels):
Careful consideration of the above trade-offs will be needed to optimize the information obtained from the measurement. Additional possibilities are to move the brightest part of the preparation out of the field of view or to differentially illuminate the preparation.
The SciMeasure camera has a four-quadrant readout. This allows a lower bandwidth read amplifier for lower read noise. However, using the pseudocolor display in the Movie section, at high display gains, you will sometimes see quadrant to quadrant differences in the signals. In one camera we looked for quadrant-to-quadrant gain differences. They were difficult to find (< 5%) at many gains and frame rates. However, in this camera they were more obvious (5% - 10%) at all gains at 2000 Hz, no binning, and at a gain of 1 dB at other frame rates. These gain differences may become a noticeable problem in measurements with large signal-to-noise ratios. In a second camera the quadrant differences were less than 5% at all tested rates and gains. You also see the quadrants clearly if you acquire one frame without subtraction. This measures the offsets.
The SciMeasure camera may pick up noise from other electrical devices. In one instance, moving an arc lamp power supply away from the camera and its power supply eliminated the noise.
The SciMeasure camera has a frame transfer time of either 40 microseconds or 7 microseconds depending on the frame rate. For 1000 fps, 500 fps, 125 fps, and 40 fps unbinned and 2000 fps 2x2 binned, the time is 40 microseconds. For 2000 fps unbinned, 3000 fps 2x2 binned, and 5000 fps 3x3 binned, it is 7 microseconds. It takes 40(7) microseconds for the electrons of a pixel adjacent to the horizontal center line to reach the frame storage areas at the top and bottom of the active area. During this 40(7) microseconds, the electrons are at 40 different pixels and during their stay at each of these 40 pixels, electrons will be added in proportion to the light reaching each of these 40 pixels. These electrons will contaminate the signal. Thus, a small fraction of the signal on each pixel is made of photons coming from the wrong place in the image. This fraction depends on the frame rate and the image contrast. For example, at 1 kfps unbinned the fraction of time spent at the wrong pixels is 4.0% and at 5 Kfps 3x3 binning it is 3.5%.
Problems caused by this smearing will be more likely to be noticed when the frame rate is high, the signal-to-noise ratio is large, and in dark regions of images with high contrast.
At 30 dB, a single offset will not be satisfactory at all frame rates. You must adjust the offset for different frame rates. (This problem does not occur at 10 dB and the read noise is only marginally higher.)
Flaws at the Edges of the CCD Image
Because of imperfections at the edges of the CCD chip, the top and bottom rows and the side columns may have higher or lower sensitivity. If this causes aesthetic problems, you can use the Omit Array to block them out.
Interaction with a network can cause several problems in data acquisition. One type of problem is misregistration of the camera frame and BNC inputs. A second problem is occasional scrambled frames (e.g. the outer sides of a frame appear in the middle or dark frames appear in arbitrary positions occasionally). These problems are computer dependent. Some of the Dell computers we have shipped seem to have no problems, others do; the problems are not correlated with CPU speed.
Two kinds of solutions have worked. One solution is to unplug the ethernet cable during data acquisition. This solved a scrambled frames problem. Another solution is to disable the TCP/IP Service and the WINS Client (TCP/IP) on a NT computer. This solved a misregistration problem.
(On a WinNT PC this can be done as follows:
>> Settings >> Control
Panel >> Devices >> Wins Client (TCP/IP)
a Win2000 PC this can be done as follows:
>> Settings >> Control
Panel >> Administrative Tool>> Computer
Management>>Device Manager. Right click>>Property >>
the end of an experiment the ethernet
cable can be plugged in or the services restarted. This can be done by,
>> Settings >> Control
Panel >> Devices >> TCP/IP Service
and now ping and ftp will work.)
Cross-talk between BNC Channels
There will be crosstalk between BNC channels with signals and BNC channels with no inputs. You can get rid of this crosstalk by using a BNC Short (e.g. Pasternack Enterprises, PE6012) on the unused inputs.
analog output function requires a DAP840 (but won't work with a DAP820)
a-to-d card. To determine the type of DAP card you have:
If you have a DAP840 you may need to create an empty text file with a name "dap840.txt" in "c:\rsi\idl52\" if there isn't one already.
If you don't have a DAP840and want to use the Analog Output function, contact Chun Falk at RedShirtImaging for instructions.
To use the Analog Output the connection inside the control box, DAC0OUT to the Analog Output BNC needs to be made. (Ground to proper ground.)
the online manual for the requirements for the trigger pulse.
Each frame readout of the CCD camera is synchronized with the first BNC input channel (BNC1). The subsequent BNCs (BNC2-BNC8) would have a shift of 1/8 of frame interval from the previous channel. However, the BNC outputs and BNC input sampling is not completely synchronized in the current setup. Meaning, there can be 1-2 point jitter from trial to trial. In most applications, this jitter is of little concern as long as the stimulation signal or the electrical response is recorded with a BNC input channel as a timing reference. However, for users who need precise synchronization between the output and input, it can be achieved by modifying the system in the following way:
(1). Add a "jumper" to J22, which is located on the Microstar a/d board at the bottom center. Unfortunately, however, adding this "jumper" will disable the "external trigger" function.
(2). Open the file sm_takedat.pro in the c:\rsi\idl52\np\ directory, use "Search" to find "For precise synchronization", and then uncomment the two lines that follow, in two places.
Once this is done, the BNC inputs and outputs should be precisely synchronized, but the "External Trigger" function will be disabled as a result.