how can I dynamically change the number of records that needs to be filtered out?

This works coming from SQL. It does have the drawback of always trying to pull the missing rows from table a first, but if that is not an issue this should suffice.

EDIT: Removed extraneous WHILE bits and just used TOP to try and get 10 rows from each table to start.

DECLARE @a TABLE (id INT IDENTITY, adata VARCHAR(10))
DECLARE @b TABLE (id INT IDENTITY, bdata VARCHAR(10))
DECLARE @c TABLE (id INT IDENTITY, cdata VARCHAR(10))

INSERT INTO @a
VALUES
('a1'),
('a2'),
('a3'),
('a4'),
('a5'),
('a6'),
('a7'),
('a8'),
('a9'),
('a10'),
('a11'),
('a12'),
('a13'),
('a14'),
('a15')

INSERT INTO @b
VALUES
('b1'),
('b2'),
('b3'),
('b4'),
('b5'),
('b6'),
('b7'),
('b8'),
('b9'),
('b10'),
('b11'),
('b12'),
('b13'),
('b14'),
('b15')

INSERT INTO @c
VALUES
('c1'),
('c2'),
('c3'),
('c4'),
('c5')

DECLARE @dataset TABLE (id INT IDENTITY, datasource VARCHAR(10), mdata VARCHAR(10), sourceid INT)

DECLARE @countert INT = 0

--try to get 10 rows from table a
INSERT INTO @dataset (datasource, mdata, sourceid)
SELECT TOP (10) 'a' AS datasource
    , adata AS mdata
    , id AS sourceid
FROM @a a
ORDER BY id

--try to get 10 rows from table b
INSERT INTO @dataset (datasource, mdata, sourceid)
SELECT TOP (10) 'b' AS datasource
    , bdata AS mdata
    , id AS sourceid
FROM @b b
ORDER BY id

--try to get 10 rows from table c
INSERT INTO @dataset (datasource, mdata, sourceid)
SELECT TOP (10) 'c' AS datasource
    , cdata AS mdata
    , id AS sourceid
FROM @c c
ORDER BY id

--set the total counter to the current number of rows
SET @countert = @countert + (SELECT COUNT(id) FROM @dataset)

--try to finish off from table a
WHILE @countert < 30
BEGIN
INSERT INTO @dataset (datasource, mdata, sourceid)
SELECT TOP (1) 'a' AS datasource
    , adata AS mdata
    , id AS sourceid
FROM @a a
WHERE NOT EXISTS (SELECT 1 FROM @dataset d WHERE datasource = 'a' AND d.sourceid = a.id)
ORDER BY id
SET @countert = @countert +1
END

--try to finish off from table b
WHILE @countert < 30
BEGIN
INSERT INTO @dataset (datasource, mdata, sourceid)
SELECT TOP (1) 'b' AS datasource
    , bdata AS mdata
    , id AS sourceid
FROM @b b
WHERE NOT EXISTS (SELECT 1 FROM @dataset d WHERE datasource = 'b' AND d.sourceid = b.id)
ORDER BY id
SET @countert = @countert +1
END

--try to finish off from table c
WHILE @countert < 30
BEGIN
INSERT INTO @dataset (datasource, mdata, sourceid)
SELECT TOP (1) 'c' AS datasource
    , cdata AS mdata
    , id AS sourceid
FROM @c c
WHERE NOT EXISTS (SELECT 1 FROM @dataset d WHERE datasource = 'c' AND d.sourceid = c.id)
ORDER BY id
SET @countert = @countert +1
END

SELECT * FROM @dataset