Filter Implementation

I am trying to develop my own IP core for a second order low pass filter implementation .

The coefficients of the designed filter are floating point values . How can i implement the floating point precision values in the IP?

I tried with floating point hardware in Nios platform .It is only supporting single precision. As it is only single precision support the results are not meeting the expected due to lose of precision in the coefficients.

How to implement the coefficients of the filter in floating or fixed format without loosing accuracy?

Is there any procedure to implement the same?

Thanks in advance