How to fix ODBC Driver Error "Invalid Character Conversion"

Question

How to fix ODBC Driver Error "Invalid Character Conversion"

Hartman, Richard 0

We are transferring data from sybase to mssql server, and the migration assistant is throwing "invalid character conversion" errors for some fields. These fields contain things like a degree symbol (alt-0176).

I tried getting at the data directly using a python script and pyodbc, and it threw the same error:
U:\scratch>python pyquery.py

Error: ('HY000', '[HY000] [SAP][ASE ODBC Driver]Invalid character conversion. (30062) (SQLGetData); [HY000] [SAP][ASE ODBC Driver]Invalid character conversion. (30062)')

So I can't script a data substitution that way.

The column type in both the source (sybase) and destination (mssql server) is VARCHAR.

There are other non-standard characters, like the em-dash. They are not all just degree signs, so I can't just make a substitution assuming all invalid characters are one thing.

Is there any way to disable this error? Or establish a substitution table for the migration assistant?

Erland Sommarskog 121.4K Reputation points MVP Volunteer Moderator

2025-06-06T20:58:44.97+00:00

You first need to get an understanding of why you get that error. Since you are able to reproduce this with a Python client, you are in a better position to do this.

You need to check what is the character set of the varchar data. Since you say that 176 is a degree symbol, it sounds like the data is in Latin-1. But what is the actual collation of the data? "Incorrect character conversion" sounds like there is an attempt to convert from UTF-8, which will fail if that is a lone byte with the high bit set. But why would the ODBC driver get the idea of converting from UTF-8 is the data is Latin-1? Is something set incorrectly?

You need to dig into this.

Your answer

Erland Sommarskog 121.4K Reputation points MVP Volunteer Moderator

2025-06-06T20:58:44.97+00:00

You first need to get an understanding of why you get that error. Since you are able to reproduce this with a Python client, you are in a better position to do this.

You need to check what is the character set of the varchar data. Since you say that 176 is a degree symbol, it sounds like the data is in Latin-1. But what is the actual collation of the data? "Incorrect character conversion" sounds like there is an attempt to convert from UTF-8, which will fail if that is a lone byte with the high bit set. But why would the ODBC driver get the idea of converting from UTF-8 is the data is Latin-1? Is something set incorrectly?

You need to dig into this.

Share via

How to fix ODBC Driver Error "Invalid Character Conversion"

Your answer