Weak magnetism at low temperature: The diffuser is weakly magnetic. For the Knight Shift insert below 100 K the error becomes significant and is about 140 ppm at base temperature. The best solution at this time is to place a 9.2 mm spacer between the diffuser and sample/reference, this will reduce the error to about 20 ppm. Also, if only the high temperature reference frequency is used the error is reduced to 10 ppm.
Check out the data at: http://trshare.triumf.ca/~hitti/Cryostats%20and%20Thermometry/Cryostats/...