I know enough about DB design to know that I am not the best resource to answer this question.
I'm thinking based on your observation that the x-data quite freq the same. Cound you add additional field to the Y-data table that ID's the sample rate, start time, and sample count?This may eliminate the need forthe x-table and the X-reference.
Otherwise I think if you add a unique key filed to your x-table, you could bind to an equivelent filed in the y-table, thereby tossing the X-reference.
Just thinking,
Ben