First of all, this is very small amount of information that you have provided. More information, maybe even a drawing, would give better answers.
From experience I can say that a peek seat seals best at low pressures when matched (lapped) to the ball, similar to a metal seat. Or the spring force must be increased significantly, which causes in a very high torque.

Non leakage at high pressure, and leakage at low pressure, all else equal, gives a clear indication that 'something' moves physically by higher pressure to provide a better sealing. It could either be a movement of parts of the sealing (lips?), or the sealing itself.

Anyway, a perfect sealing is obtained by exact, precise machining, within rigid microscopic tolerences. Control measurements of all parts? A 'hair siced' deviation can give a leak.

First step is to determine where it is leaking. Between ball and seat? Around OD of seat? Or both. You can determine this by changing the depth of water on the downstream side and slowly increase pressure until the bubbles start to flow. Using rubbing alcohol instead of water can make it easier to see the bubbles. Once you know the location(s) of the leak, then you know where to focus your effort.

You said you increased and decreases spring force with no success. Did the amount of leak change? You should be measuring the leak rate so you can tell if any changes you make are making the leakage lower or higher. Otherwise you will keep guessing and hoping you can solve the problem and not make any progress. Time to be an Engineer and investigate, evaluate and solve a problem.

Thank you everyone for your valuable feedback.
I have identified the problem. The force exerted by the spring were still less. Hence increased the no.of spring and got the positive result