Convert double to float for Vector3

I get a "Cannot implicitly convert 'double' to 'float'. An explicit conversion exists (are you missing a cast?)" message for

the following formula to be used as argument in a Vector3:

float sph_x = Input.mousePosition.x/1280*7.7-3.85f; //sph_x would be 0,22859374999999900000

Surprisingly it works for this formula : float sph_x = Input.mousePosition.x/1280*7-3.85f; //sph_x would be -0,14218750000000000000

I need only 2 or 3 decimals of precision. Is there no conversion from double to float? I found only the Mathf.RoundToInt.

Not even if I multiply by 1000, Round to int and then divide back by 1000 doesn't get a float:

float sph_y = ( Mathf.RoundToInt (1000* (Input.mousePosition.x/1280*7.65-3.85f) ) )/1000;

Now it gets 2 errors: the "cannot implicitly ..." and another one "best overloaded method..." Dividing an integer by 1000 results in 3 decimals ONLY so a float should be OK.

Only it isn't.

There has to be a nice simple way to do this...

Thanks.