Binary waveform data transfer too short (but ASCII ok) from Agilent scope through GPIB

Hello,

I am using the script below to read data from an Agilent oscilloscope. Data transfer in ASCII format this works fine (but slow), but in BYTE or WORD format, I get only a few of the expected thousands of data points (the actual number of data points varies between script executions).

For instance, using BYTE format, I get waveform data like: "#564000ÿÿþÿþþþ", indicating that there are 64000 data points to follow, but there are only 7 present. Where are the other 63993 data points?

Am I forgetting something?

Thanks for your help, Robert

% script agi1.m
%
% Reads waveforms from an Agilent Infiniium 54830B oscilloscope, used with
% Agilent IO-Library Suite 16 (NetFx_IOLibSuite_16_1_14931.exe), Agilent 
% 82357A USB/GPIB interface and Matlab R2011a.
%
% Reading data in ASCII format works OK (but slow), reading in BYTE or WORD
% format yields much less data points than expected (0 to a few hundred,
% where 64000 or more are expected)
%
% last update 18-08-2011

% set some defaults
SamplingRate = 1e7; % samples per second 

% select data format (uncomment as appropriate)
% DataFormat = 'ASCii';
DataFormat = 'BYTE';
% DataFormat = 'WORD';

switch DataFormat
    case 'ASCii'
        InputBufferSize = 1e7; % good for sampling rates up to 1e8
    case { 'BYTE', 'WORD' }
        InputBufferSize = 1e6;
end

% Find a GPIB object.
obj1 = instrfind('Type', 'gpib', 'BoardIndex', 7, 'PrimaryAddress', 7, 'Tag', '' );

% Create the GPIB object if it does not exist, otherwise use the object that was found.
if isempty(obj1)
    obj1 = gpib('AGILENT', 7, 7);
else
    fclose(obj1);
    obj1 = obj1(1)
end

% Configure instrument object, obj1.
set( obj1, 'Timeout', 25.0); % long timeout for slow ASCII transfer
set( obj1, 'InputBufferSize', InputBufferSize );

% Connect to instrument object, obj1.
fopen( obj1 );

% initialize some parameters
fprintf( obj1, ':SYSTem:HEADer OFF' );
fprintf( obj1, ':TIMebase:REFerence LEFT' ); % set T = 0

% set sampling rate
fprintf( obj1, [ ':ACQuire:SRATe ' num2str( SamplingRate ) ] );

fprintf( obj1, ':TIMebase:RANGe?' ); % full-scale horizontal time in seconds
TimebaseRange = str2num( fscanf( obj1 ) );

% set up data acquisition
fprintf( obj1, ':ACQuire:AVERage OFF' );
fprintf( obj1, ':ACQuire:MODE RTIMe' );
fprintf( obj1, ':ACQuire:POINts AUTO' );

% digitize channel 1 for later read-out
fprintf( obj1, ':DIGitize CHANnel1' );
fprintf( obj1, ':CHANnel1:DISPlay ON' );
% turn channel 1 display back on

fprintf( obj1, ':WAVeform:SOURce CHANnel1' );
fprintf( obj1, [ ':WAVeform:FORMat ' DataFormat ] );

% # data points to read
fprintf( obj1, ':WAVeform:POINts?' );
WaveformPts = str2num( fscanf( obj1 ) );

% read waveform
fprintf( obj1, ':WAVeform:DATA?' );
WaveformData = fscanf( obj1 );

% restart data acquisition by scope
fprintf( obj1, ':RUN' );

% Disconnect all objects.
fclose( obj1 );

% set input buffer back to small size
set( obj1, 'InputBufferSize', 512 );

% Clean up all objects.
delete( obj1 );

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% process data

switch DataFormat
    case 'ASCii'
        Volts = str2num( WaveformData );
    case { 'BYTE', 'WORD' }
        % see page 28_10 of Agilent54830-97014ProgrammersReference.pdf

          disp( WaveformData );

          % start reading after the first '#'
          Idx = findstr( WaveformData, '#' ) + 1 ;
          Idx = Idx( 1 );
          NoHdrBytes = str2num( WaveformData( Idx ) );
          NoDataBytes = str2num( WaveformData( Idx + 1 : Idx + NoHdrBytes ) );

          if WaveformPts ~= NoDataBytes
              disp( '   WaveformPts ~= NoDataBytes' );
          end

          DataBytes = WaveformData( Idx + NoHdrBytes + 1 : end );
          if length( DataBytes ) ~= NoDataBytes
              disp( '   length( DataBytes ) ~= NoDataBytes' );
              disp( [ '   ' num2str( length( DataBytes ) ) ' ~= ' num2str( NoDataBytes ) ] );

          end
          Volts = double( DataBytes );
  end

Answer 1

[Rob Graessle]

Since in the BYTE or WORD case the waveform data is binary format instead of an ASCII string, the FREAD function should be used instead of FSCANF. If you need to interpret the header information, you could use the FSCANF function limited to a pre-defined number of characters, followed by the FREAD function to get the binary data.

However, since the waveform data is in binblock format you could just use the BINBLOCKREAD function to handle the whole process of extracting the header information and waveform data.

This technical solution gives some examples.

[Robert Klein-Douwel]

Thanks a lot! I had indeed overlooked those details, it works fine now.