reading single-precision floating-point number from binary .dat file written by vb6

I have a old program written in vb6 that writes test data into a .dat file using the binary file writer in Visual Basic. I'm trying to read the binary file back into MATLAB, but it seems like VB6 writes numbers in "single" format differently than MATLAB reads it.

here is the code as follows:

fid = fopen('C:\Users\mvandyke\Downloads\csdm000111.dat');
fom = double(fread(fid, 1, 'single'));
mName = char(fread(fid, 60)');
mDateTime = char(fread(fid, 18)');
rating = fread(fid, 1, 'int16');
rpx = fread(fid, 2, 'single');
tracked = fread(fid, 2, 'int8');
numcomps = fread(fid, 1, 'int16');
maxlvl = fread(fid, 1, 'single');
maxexc = fread(fid, 1, 'single');
maxloc = fread(fid, 2, 'int8');
maxord = fread(fid, 1, 'single');
maxlbl = char(fread(fid, 2)');
avgSamples = fread(fid, 1, 'int16');
avgRpm = fread(fid, 1, 'single');
nontriax = fread(fid, 1);
comps = 1:10;
PU = int16(1:10);
ORIENTS = int8(1:30);
CMPNUMS = int16(1:10);
numcols = int16(1:8);

for i = 1:numcomps
    comps(i) = fread(fid, 1, 'single');
    PU(i) = fread(fid, 1, 'int16');
    ORIENTS(3 * (i - 1) + 1) = fread(fid, 1, 'int8');
    ORIENTS(3 * (i - 1) + 2) = fread(fid, 1, 'int8');
    ORIENTS(3 * (i - 1) + 3) = fread(fid, 1, 'int8');
    CMPNUMS(i) = fread(fid, 1, 'int16');
    numcols(i) = fread(fid, 1, 'int16');
end
fread(fid, 1);

TBLVALS = 8:100:84;

for k = 1:84
    for j = 1:100
        for i = 1:8            
            TBLVALS(i,j,k) = fread(fid,1,'single');
        end
    end
end

comp1 = permute(TBLVALS(1,:,:), [2 3 1])';
comp2 = permute(TBLVALS(2,:,:), [2 3 1])';
comp3 = permute(TBLVALS(3,:,:), [2 3 1])';

clear i j k fid ans TBLVALS;

it seems to write everything else correctly, such as the char arrays, but the single-precision numbers come out as very high numbers in the positive or negative range. Any advice?

Answer 1

[Walter Roberson]

Experiment with using swapbytes() on the singles. It might be an endian issue.

When reading binary files, it is recommended that you pass the endian information to fopen() if it is known ahead of time; if you need it to be different for some reads then you can pass the endian information on an fread() call.

Answer 2

[Tony Scarcia]

I use the following to read binary files into matlab. The function below reads the first 10 bytes of the file called 'filename.bin' into an array called sp.

function[sp] = read_cf_file('filename.bin')
format short g;
fileID = fopen('filename.bin', 'r');
sp = fread(fileID,[10],'single');
fclose(fileID);
end