|
The basic shadowband design calls for two shadowbands, one that occludes 2°
and another that occludes 5° of the sky. During the kickoff meeting Vogelmann, after discussions with Qilong,
agreed that we will proceed only with the 2° band for the first stage of the development. The engineering team (Reynolds/Smith/Edwards)
agree that it will not be too difficult to add the second shadowband once the first is working properly.
The state of the shadowbands are determined by two switches on either horizon of the band. We will want a design such that in the event of software lockup, the switches have full control over the motor. In this way the motor will be turned off before the band is driven into the mounting plate. |
|
The logic design, shown here, should satisfy all fail-safe requirements with a minimum of complexity. |
|
|
| Demonstration 1: A demonstration of the first phase of development of the TCRSR was conducted successfully at BNL on 25 July 2007. Attending were R. M. Reynolds, W. Wiscomb, A. Vogelmann, S. Smith (mechanical engr.), and R. Edwards (electronic engr.). |
|
Oscilloscope traces of MFR Head Band 1 output at near solar noon on 25 July 2007.
The voltage output from the head has a negative polarity so in the trace is about -400 mV in full
sky and about 25 mV in shade.
For the 2-deg band the shadow period lasted approximately 85 msec.
The shadow minimum for the 5-deg sweep lasted approximately 220 msec.
An analog-to-digital sampling period of 20 msec (50 samples/sec ) would collect at least three samples in the shadow
for the 2-deg band and 10 samples for the 5-deg sweep.
For band 1, the 2-deg band, a full sweep of 190 deg takes about 14 sec or 13.6 deg/sec. Therefore, for a sample period of 20 msec the samples spacing would be 0.27 deg/sample. A total of 700 12-bit ADC samples would be collected for each sweep. For seven channels, each sweep would yield a total of 4900 samples. There will be two bytes/sample. with no compaction, so a binary record would have approximately 9800 bytes. Let's say a full record would have 10 KB. At a baud rate of 9600 bits/sec a 10 KB data block can be transmitted in 10.4 seconds Therefore, we can transmit one sweep record while the second band is sweeping. For Band 2, the 5-deg band, the sampling rate will be 20 samples/sec, the record size will be approximately 4 KB and the upload time will be about four seconds. |
The sky was clear and deep blue. We installed the TCRSR on the BNL roof test platform and operated it from 11:57:34 to 14:24:06 (EDT). There were some breaks in the data when the TT8 stopped for no apparent reason.
In all data presentation the 2 degree band is designated by 1 and the 5-degree band is designated by 2. The MFR head channels are labeled 0-7 for approximate input filters centered on [open,415,500,615,680,870,940]. The zeroth channel is the open silicone detector. Note these data are uncorrected for zenith angle, atmospheric mass, or sensor azimuth-elevation correction.
|
A summary plot for the experiment. The shadow value and global values are the analog-to-digital converter output count
where 0-2048 represents a voltage input of approximately 0 to 2.047 volts DC. Band = 2 (5 deg) and
Channel = 1 (450 nm).
Shadow value is the sweep minimum value. If more there are more than one minimum value the first minimum is taken. The shadow values are quite stable throughout the time period and is independent of the direction of the shadowband. Global values are the means of the first and last N points of each sweep. N is determined from the configuration file. In this case the first and last four points are are averaged. In the plot, the forward sweep global count is shown by a blue curve and the reverse sweep global values by a red curve. In this example the red curve is almost hidden by the blue meaning there is insignificant difference between the two directions. Shadow ratio = (dip/noise) where dip = (global mean - shadow) and noise = stdev(global values). Only sweeps for which the shadow ratio is greater than 2.3 (see config file) are accepted. Shadow window is defined by searching away from the shadow index to determine where the shadow ratio limit is exceeded. The angles where this occurs before and after the minimum determine the shadow window. |
|
A set of 21 sweeps for the 5 degree band, channel 1 (450 nm).<
Top.
Sweeps.
|
|
A set of 21 sweeps for the 2 degree band, channel 1 (450 nm).<
Top.
Sweeps.
Note, for the 2-deg band the shadow and global values are slightly higher than the same values with the 5-deg band. This is likely because the 5-deg band blocks more of the sky the the more narrow band. We will exame this effect more in following analysis. |
Another near-perfect sunny day.
Operation time: 2007-09-24 (267) 15:33:35 to 2007-09-24 (267) 18:41:43.
In all data presentation the 2 degree band is designated by 1 and the 5-degree band is designated by 2. The MFR head channels are labeled 0-7 for approximate input filters centered on [open,415,500,615,680,870,940]. The zeroth channel is the open silicone detector. Note these data are uncorrected for zenith angle, atmospheric mass, or sensor azimuth-elevation correction.
| SAME AS ABOVE |
|
| SAME AS ABOVE |
|
| SAME AS ABOVE |
On the morning of 6 October 2007, the TC-RSR was deployed at BNL to further test its operational stability while observing the burn off of fog. It began mid morning (before which the optical path was too great for an aureole to be exhibited in the scans) and progressed into early afternoon. Once the burn off reached a sufficiently small optical path, the Sun's aureole appeared. However, interestingly, its magnitude fluctuates during this later period, indicating that the fog has notable inhomogeneities as denser portions waft in and out of the field of view during the burn off.
