I used a subquery for the GROUP BY which computes the Sum of Err for each group. Then I added the ConcatRelated function (from Allen Browne) with the fields returned by the subquery. This is the query and the output (based on your sample data in make_table_bp) from the query:
SELECT
sub.[Name],
sub.Cat,
sub.[Desc],
sub.Thresh,
sub.Perc,
sub.SumOfErr,
ConcatRelated("BP",
"make_table_bp",
"[Err] > 0 AND [Name] = '" & sub.[Name]
& "' AND Cat = '"
& sub.Cat & "'",
"BP")
AS concat_BP
FROM
(SELECT
q.[Name],
q.Cat,
q.[Desc],
q.Thresh,
q.Perc,
Sum(q.[Err]) AS SumOfErr
FROM make_table_bp AS q
GROUP BY
q.[Name],
q.Cat,
q.[Desc],
q.Thresh,
q.Perc
) AS sub
ORDER BY
sub.Name,
sub.Cat;
The query outputs this result set:
Name Cat Desc Thresh Perc SumOfErr concat_BP
Bob C1 Inf 7Per 0.05 16 AEC2, BAC2, VBE2
Bob C2 Com 8Per 0.45 4 ADC2, XBC4
Joe C1 Inf 7Per 0.05 3 KSC2, QYC2
Notice I enclosed Name, Desc, and Err with square brackets every place they were referenced in the query. All are reserved words (see Problem names and reserved words in Access). Choose different names for those fields if possible. If not, use the square brackets to avoid confusing the db engine.
But this will not work unless/until your copy of the ConcatRelated function is recognized by your data base engine. I don't understand why it's not; I followed the same steps you listed for storing the function code, and this works fine on my system.
Edit: I tested that query with my version of the table, which has [Err] as a numeric data type. Sounds like yours is text instead. In that case, I'll suggest you change yours to numeric, too. I don't see the benefit of storing numerical values as text instead of actual numbers.
However if you're stuck with [Err] as text, you can adapt the query to deal with it. Change this ...
"[Err] > 0 AND [Name] = '" & sub.[Name]
to this ...
"Val([Err]) > 0 AND [Name] = '" & sub.[Name]
That change prevented the "Data type mismatch in criteria expression" error when I tested with [Err] as text data type. However, I also changed this ...
Sum(q.[Err]) AS SumOfErr
to this ...
Sum(Val(q.[Err])) AS SumOfErr
AFAICT that second change is not strictly necessary. The db engine seems willing to accept numbers as text when you ask it to Sum() them. However I prefer to explicitly transform them to numerical values rather than depend on the db engine to make the right guess on my behalf. The db engine has enough other stuff to deal with, so I try to tell it exactly what I want.
Edit2: If you want only unique values concatenated, you can modify the ConcatRelated() function. Find this section of the code ...
'Build SQL string, and get the records.
strSql = "SELECT " & strField & " FROM " & strTable
and change it to this ...
'Build SQL string, and get the records.
strSql = "SELECT DISTINCT " & strField & " FROM " & strTable
