Observations at X/Ka-band are motivated by their ability to access more compact source mor- phology and reduced core shift relative to observations at the historically standard S/X-band. In addition, the factor of four increase in interferometer resolution at Ka-band should resolve out some extended source structure. Given these motivations, an X/Ka-band (8.4/32 GHz) celestial reference frame has been constructed using a combined NASA and ESA Deep Space Network. In 176 observing sessions we detected 678 sources covering the full 24 hours of right ascension, α, and the full range of declinations, δ. The resulting XKa median precision is now 67 μas in αcosδ and 95 μas in δ.
Comparison of 529 X/Ka sources in common with the S/X-band (2.3/8.4 GHz) ICRF3 produced wRMS agreement of better than 165 μas. There is evidence for systematic errors at the 100 μas level. Known errors include limited SNR, lack of phase calibration, troposphere mismodelling, and terrestrial frame distortions. Actions are underway to reduce all of these errors. In particu- lar, a collaboration between NASA and the ESA deep space antenna in Malargüe, Argentina is reducing weaknesses in the southern hemisphere. By comparing coordinate estimates, we probe the accuracy limits of current celestial frames in an effort to understand the advantages of each frame.