libpq - characterset encoding error from selecting BYTEA

My understanding of this part of documentation

"operations on binary strings process the actual bytes, whereas the
processing of character strings depends on locale settings."

is that reading BYTEA column does not involve character set conversion
especially when "1", indicating binary, is passed to parameter
"resultFormat" by calling libpq function PQexecParams().

Given

CREATE TABLE t1 (
PRIMARY KEY (c1,c2),
c1 SMALLINT,
c2 SMALLINT,
c3 BYTEA
);

PGconn* first switches to "BIG5" client encoding:
PQexec(conn,"SET CLIENT_ENCODING TO BIG5");

version 1:

const char *pValues[]={"5","6"};
PGresult *r=PQexecParams(conn,"SELECT c3::BYTEA FROM t1 WHERE c1=$1 AND
c2=$2",2,NULL,pValues,NULL,NULL,1);

What is strange is that above PQexecParams() works in one portion of my
program but in another portion it yields the following error:

ERROR: character with byte sequence 0x98 0xe1 in encoding "BIG5" has no
equivalent in encoding "UTF8"

version 2:

PQgetvalue(r,0,0) returns the correct BYTEA value from this version:
PGresult *r=PQexecParams(conn,"SELECT c3::BYTEA FROM t1 WHERE c1=5 AND
c2=6",0,NULL,NULL,NULL,NULL,1);

I simply can not find any intrinsic difference between the above two
versions. Why the first version is "unstable"?

Best Regards,
CN

--
http://www.fastmail.com - Or how I learned to stop worrying and
love email again