Guralp Systems CMG-DM24 - Manual (Page 61)

Operator’s Guide DM-24 Digitiser
Issue G January 2003 59
The next byte contains the sample rate (s/sec in binary). If this field is zero this indicates
that the block contains status (text) information.
The third byte is used to specify the compression format for all the data in the block.
The 3 least significant bits are used to indicate whether the data elements (32 bits)
contain 1, 2 or 4 sample points - i.e. a value 4 indicates that the data records should be
treated as 4 8 bit differences, a value 2 means the data are 2 16 bit differences and the
value 1 the data are a single 32 bit difference.
The fourth byte contains the count of the number of data records (32 bits).
The product of these last 2 bytes can be used to calculate the total number of sample
points in the block.
Block Data.
First absolute value
32 bits
First 1, 2 or 4 differences 1*32bit, 2*16bit or 4*8bit values
....
....
Final 1, 2 or 4 differences same format 1, 2 or 4 differences
Final absolute value 32 bits
The first data record in the block, following this format record, is the initial 32 bit
absolute value (forward integration constant). The last record is the final 32 bit absolute
value (reverse integration constant). Between these are the specified number of data
records. Each data record contains the specified number of 8, 16 or 32 bit differences
from the previous value. The first difference is always zero as it corresponds to the first
sample.
By definition each data block starts on an integral second and contains an integral
number of seconds of data. The data block has a maximum size of 1024 bytes (16 byte
header + 8 bytes for start and end value leaves 1000 bytes for data differences e.g. at 8
bit compression the block will contain 1000 sample points i.e. 10 seconds of data at 100
s/s - using minimum compression (32 bit) only 2 complete seconds of data can be fitted
in a block at this sample rate - 2*100*4 = 800 bytes). Changes in signal level will result
in the compression algorithm having to change the format so blocks are not necessarily
filled to the maximum specified capacity, when the next second of data requires more
dynamic range.
The data has this format when processed and stored in the DM and SAM units
(maintaining data values on word boundaries in the processor), but when data blocks are
transmitted via the serial ports, blocks of 32 bit differences have the redundant most
significant byte (of the difference values) omitted to optimise the efficiency of the serial
transmission.