Aspheric surface testing always needs null lens, but the accuracy of the null lens are decided by the optical elements and the assemblage, and the accuracy and the feasibility of the null lens are unknown, and most of null lens need to be calibrated by tilted computer-generated-hologram (CGH). We introduce the principle of the null lens, and designe null lens to test aspherical surface, getting the simulation accuracy of the null lens which is 2.91 nm [root mean square (RMS)], and then design CGH to calibrate the null lens. Our initial results show that the accuracy of null lens is limited by the accuracy of optical elements and the assemblage. The simulation accuracy of null lens is 2.91 nm (RMS) and the actual accuracy is unknown. Though the calibrating of null lens, we can test the aspheric surface to 1.79 nm (RMS) surface errors, and also make sure the accuracy of null lens to test aspheric surface.