I have discussed in another article how you can convert data from one codepage to another.

Also I have described a method to copy unix files to MVS datasets.

This article summarizes a method to copy an MVS dataset to Windows, while keeping records and converting from ebcdic to the utf8 codepage.

This job copies the MVS dataset to a UNIX files, indicating to keep record indicators with the Windows CR character (UNIX uses the CRLF typically as record / line separators). The dataset below ‘YOUR.TEST.PS’ should be a PS – physical sequential – dataset or a PDS(E) member.

//STEP1    EXEC PGM=BPXBATCH                                         
//STDOUT   DD SYSOUT=*                                               
//STDERR   DD SYSOUT=*                                                
//STDIN    DD DUMMY    
//* Values in STDENV below are kept but have no meaning for this function                                              
//STDENV   DD *                                                      
JAVA_HOME=/usr/lpp/java/J8.0_64                                      
PATH=/usr/lpp/mqm/web/bin:/bin:/usr/sbin                      
LIBPATH=/usr/lpp/mqm/java/lib                                 
//STDPARM  DD   *                                                    
SH cp -v -F cr "//'YOUR.TEST.PS'" /your/unixdir/tst.txt 

You can now convert the EBCDIC data to UTF-8 as follows:

iconv -f 37 -t 1208 /your/unixdir/tst.txt  > /your/unixdir/utf8-tst.txt

Now you can transfer the file to Windows. Transfer the file in binary mode, otherwise another unwanted code page conversion will happen.