I am investigating three posible causes:
1. The 1038 is just more efficient than the Alpha 10-A. This can be corrected with just a cap in series with the piezo--as Wayne already explained below.
2. The plastic housing of the 1038 is resonating. This might be cured by applying a mixture of silicone calking and sand to the extrior of the back side of the 1030 for damping.
3. The Alpha 10-A and the 1038 may overlap a bit too much. The Alpha 10-A has fairly flat response up to 5KHz while the 1038 is prety flat from 3.5KHz up. I make this comment based only on the published response curves of the drivers and my intepretation of them. Any feed back on this posibility will be most appreciated.

Items one and two are simple to experiment on. Item three, however would be difficult to "fix" or even verify. I'll borrow a real time analyzer to check it out.