Campbell Scientific NL116 NL116 Ethernet and CompactFlash Module - Page 34

NL116 Ethernet and CompactFlash® Module 7.3.3 Table Size and Mode The size of each data table in CPU memory is set as part of the DataTable() instruction and the size of each data table on the CF card is set with the CardOut() instruction. Because they are set independently, they can be different. It is important to note that if the CPU memory is set to fill-and-stop mode, once a table is full, all data storage to the table will stop. No more records will be stored to the CPU memory or the card. 7.4 CF Card Data-Retrieval Details Data stored on CF cards can be retrieved through a communication link to the datalogger or by removing the card and carrying it to a computer. 7.4.1 Via a Communication Link Data can be transferred to a computer via a communication link using one of Campbell Scientific's datalogger support software packages (for example, PC200W, PC400, LoggerNet). There is no need to distinguish whether the data is to be collected from the CPU memory or a CF card. The software package will look for data in both the CPU memory and the CF card. The datalogger manages data on a CF card as final-storage table data, accessing the card as needed to fill data-collection requests initiated with the Collect button in datalogger support software. If desired, binary data can be collected using the File Control utility in datalogger support software. Before collecting data this way, stop the datalogger program to ensure data is not written to the CF card while data is retrieved. Otherwise, data corruption and confusion will result. 7.4.1.1 Fast Storage/Data-Collection Constraints When LoggerNet collects data from ring tables that have filled, there is the possibility of missing records due to the collection process. LoggerNet uses a "round-robin" collection algorithm that collects data from multiple tables in small blocks as it sequences around to all the tables. Collection starts at the oldest data for each table. When a ring table has filled, the oldest data is overwritten by current data. With filled ring tables, as collection begins LoggerNet queries the datalogger for the oldest data starting with the first table. When this data block is returned, LoggerNet goes to the next table and so on until all of the tables are initially collected. By the time LoggerNet makes the second pass requesting more data from the tables, the possibility exists that some of that data may have been overwritten, depending on how fast the datalogger is storing data (that is, data storage rate, number of table values, and number of tables). Normally, LoggerNet gets ahead of the storing datalogger and the remaining data is collected without gaps; however, if the datalogger is storing data fast enough, it is possible to get into an always-behind scenario where LoggerNet never catches up and the datalogger repeatedly overwrites uncollected data. 26