desham,
You can never completely remove all noise by filtering. White noise is defined as having a uniform spectral density. That means that each (equal-bandwidth) segment of the spectrum has an equal amount of energy. So there will always be some energy within the passband of the filter. Always. Every time.
If you want a noiseless output signal, you will need to generate a new signal which has as its parameters (frequency, amplitude, and phase) the best estimates you can measure and calculate for those parameters from the noisy input signal. Because the input signal is noisy, the parameters will always contain some error. You can estimate the error. You may be able to out some bounds on the size of the error if you know enough about the process and channel which generate and transmit the signal. You can never completely eliminate the error.
To eliminate the DC offset, calculate the mean value and subtract that from the original data. If the offset shifts over time you will need to determine what size segments of data to use to approximate the desired offset removal. A drifting offset is fundamentally like a very low frequency noise component. You can measure and estimate, but probably cannot completely remove it with 100% accuracy.
Here is a cleaned up version of your VI with some extras showing offset removal and filtering of the noise only.
Lynn
