From the X and Ku band data, it is clear that:
ea05 is giving unreliable Psum and PDif values at X-band only in all
four IFs. The visibility data are fine, but the switched power values
have little relation to actual gain changes.
ea17, in LCP, is giving poor switched power values in both X and Ku
bands, in IF 'C' only. The values are extremely noisy.