Join and sum multiple columns - sql-server

Question

Table TB1:
| PaymentID | CashAmount |
--------------------------
| P1 | 3,000|
| P2 | 5,000|
| P3 | 8,000|
Table TB2:
|ChequeID| PaymentID |ChequeAmount|
-----------------------------------
| C1 | P2 | 10,000|
| C2 | P1 | 15,000|
| C3 | P1 | 2,000|
Table TB3:
|TransferID| PaymentID |TransferAmount|
---------------------------------------
| T1 | P2 | 20,000|
| T2 | P2 | 20,000|
| T3 | P1 | 3,000|
Expected Result
| PaymentID | CashAmount |ChequeAmount|TransferAmount|
------------------------------------------------------
| P1 | 3,000| 17,000| 3,000|
| P2 | 5,000| 10,000| 40,000|
| P3 | 8,000| NULL| NULL|
How to write a query JOIN and SUM all these tables together?
I tried writing simple join query but the result was wrong.
Example of incorrect query:
SELECT
TB1.PaymentID, TB1.CashAmount,
SUM(TB2.ChequeAmount) AS ChequeAmount,
SUM(TB3.TransferAmount) AS TransferAmount
FROM
TB1
LEFT JOIN
TB2 ON TB1.PaymentID = TB2.PaymentID
LEFT JOIN
TB3 ON TB1.PaymentID = TB3.PaymentID
GROUP BY
TB1.PaymentID, TB1.CashAmount
Incorrect result:
| PaymentID | CashAmount | ChequeAmount | TransferAmount |
----------------------------------------------------------
| P1 | 3,000 | 17,000 | 6,000 |
| P2 | 5,000 | 20,000 | 40,000 |
| P3 | 8,000 | NULL | NULL |

Answer 1

The problem you have is you're joining the tables that have multiple rows with the same id and it multiplies the numbers. You'll need to do the sum first and then join the data. Assuming the data can be missing from any of the tables, you'll also need to use full outer join. If TB1 has the row always, then left outer join is enough.
This is how you can do it:
SELECT
coalesce(TB1.PaymentID,TB2.PaymentID,TB3.PaymentID),
TB1.CashAmount, TB2.ChequeAmount, TB3.TransferAmount
from (
select PaymentID, SUM(CashAmount) AS CashAmount
from TB1 group by PaymentID
) TB1
full outer join (
select PaymentID, SUM(ChequeAmount) AS ChequeAmount
from TB2 group by PaymentID
) TB2 on TB1.PaymentID = TB2.PaymentID
full outer join (
select PaymentID, SUM(TransferAmount) AS TransferAmount
from TB3 group by PaymentID
) TB3 on isnull(TB1.PaymentID, TB2.PaymentID) = TB3.PaymentID
Example in SQL Fiddle

Answer 2

I would eliminate the duplicates in the minor tables first, then (left) join and sum with the major table, like this
SELECT
TB1.PaymentID,
TB1.CashAmount,
Sum(TB2.ChequeAmount) as ChequeAmount,
Sum(TB3.TransferAmount) as TranAmt
from TB1
left join (
select PaymentID, SUM(ChequeAmount) AS ChequeAmount
from TB2 group by PaymentID
) TB2 on TB1.PaymentID = TB2.PaymentID
left join (
select PaymentID, SUM(TB3.TransferAmount) AS TransferAmount
from TB3 group by PaymentID
) TB3 on isnull(TB1.PaymentID, TB2.PaymentID) = TB3.PaymentID
group by
tb1.PaymentId, tb1.CashAmount

Answer 3

WITH t2
AS (
SELECT [PaymentId], SUM([ChequeAmount]) AS ChequeAmount
FROM TB2
GROUP BY [PaymentId]
),
t3
AS (
SELECT [PaymentId], SUM([TransferAmount]) AS TransferAmount
FROM TB3
GROUP BY [PaymentId]
)
SELECT t.PaymentId, SUM(CashAmount) AS CashAmount,
SUM(t2.[ChequeAmount]) AS [ChequeAmount],
SUM(t3.[TransferAmount]) AS [TransferAmount]
FROM TB1 AS [t]
LEFT JOIN [t2] ON [t2].[PaymentId] = [t].[PaymentId]
LEFT JOIN [t3] ON [t3].[PaymentId] = [t].[PaymentId]
GROUP BY [t].[PaymentId];