Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://www.naic.edu/~jeffh/radar.html
Дата изменения: Wed Jan 2 22:08:27 2002 Дата индексирования: Mon Oct 1 20:39:10 2012 Кодировка: Поисковые слова: п п п п п п п п п п п п п п п п п п п п п п |
The WAPP is the Wideband Arecibo Pulsar Processor. This device is a digital correlator which feeds data at high rates (up to 50mb/sec) into a PC. The PC is interfaced with a high speed parallel card.
The existing radar interface hardware is interfaced to an ancient VME system which writes its data to tape drives or 10base2 Ethernet. The hardware itself can go much faster than theses devices. (I'm told >10Mb/sec) So we got an idea to easily interface the VME system to the WAPP. Data is read from the radar interface hardware and sent over the VME bus to a MVME162 card. From the 162 card the data passes over the IP (Industry Pack) bus to a second high speed parallel card in the WAPP PC. Then the data is written to (LVD SCSI 160) hard drives. As a pulsar machine, the WAPP includes a host of tape drives and lots of disk space (~200 Gb). One 36GB hard drive is dedicated to the aeronomy group.
The data rates of the various pieces are:
The radar interface consists of an interface gui screen in TCL which runs on the observer's console, and some software running in the WAPP computer.
When it is working normally, you should see green messages every ten seconds telling the data rate. Once started, the data taking will continue writing whatever data it receives until the filesystem is full. It will write one file until it is near 2GB in size and then switch to a new file with a filename extension. For example:
If you see red messages, the radar interface is definately not working. These errors generally occur if the hard drive selected is full or the radar observing is started before the wapp software.
There remains a bug in the data taking with the WAPP radar interface. The parallel interface card has an internal 64 byte PCI fifo. This fifo cannot be emptied unless the fifo is cleared or an even 64 bytes of data is written to it. This is a limitation of the parallel interface hardware. (EDT PCD60)
In practice this fifo is cleared when you choose "New File" on the radar interface gui screen. I am told this is not a big issue as its easy to take at least one extra record of data. So if the last record is missing up to 63 bytes, its of no consequence.
Another bug. It should be possible to control the radar interface completely from the GUI and never need to type into a separate window. This would have the advantage of the wapp knowing when data was about to be sent. This could be done with help from an interested scientist.
Jeff Hagen, Sept 2001