In Excel, how can I ensure that N randomly generated columns from a named range have unique values across each row?

Background

I am able to select values at random from a named range and populate a column called: MyColumn using =INDEX(aNamedRange,RANDBETWEEN(1,ROWS(aNamedRange))).

It looks like this:

|            MyColumn           |
|-------------------------------|
| RandomlySelectedValue_1, Col1 |
| RandomlySelectedValue_2, Col1 |
|              ...              |
| RandomlySelectedValue_N, Col1 |

I have made two other columns: MyColumn2 and MyColumn3.

Question

As RandBetween executes upon refresh is there a way to ensure that RandomlySelectedValue_1, Col1 does not equal RandomlySelectedValue_1, Col2?.

In other words:

• This is the expected behavior, because all the randomly selected values in the columns are distinct across the same row.
• This is what's happening, there is no check to prevent any randomly selected duplicates in the same row.

What I've tried

I made this VBA Script:

Sub UniqueRandoms()
Do
    Worksheets("Sheet1").Range("AB7").Value = WorksheetFunction.Index(Range("FileName.xls!aNamedRange"), WorksheetFunction.RandBetween(1, Sheets("Sheet2").Rows(Range("FileName.xls!aNamedRange"))))
    Worksheets("Sheet1").Range("AC7").Value = WorksheetFunction.Index(Range("FileName.xls!aNamedRange"), WorksheetFunction.RandBetween(1, Sheets("Sheet2").Rows(Range("FileName.xls!aNamedRange"))))
    Worksheets("Sheet1").Range("AD7").Value = WorksheetFunction.Index(Range("FileName.xls!aNamedRange"), WorksheetFunction.RandBetween(1, Sheets("Sheet2").Rows(Range("FileName.xls!aNamedRange"))))
Exit Do

Loop Until Worksheets("Sheet1").Range("AB7").Value <> Worksheets("Sheet1").Range("AC7").Value And Worksheets("Sheet1").Range("AD7").Value <> Worksheets("Sheet1").Range("AC7").Value And Worksheets("Sheet1").Range("AB7").Value <> Worksheets("Sheet1").Range("AD7").Value

End Sub

but I'm receiving a Run-time error '1004': Application-defined or object-defined error.

Is the above script the best way to solve the problem or is there a simpler solution (that is potentially generalizable to any number of columns)?