I bought a Fluke 720A more than 20 years ago locally from a surplus dealer for around $400 (as far as I can remember). The first thing I did was try the "Self Cal" procedure as shown in the manual using a HP3456A as the null meter. Several of the adjustments in the first decade and the second decade were beyond what the 20 turn adjustment pots would cover. The first thing I did was to clean the decade switches with FREON TF and then apply Cramolin "R" to the switch contacts. This did very little or nothing to correct the problem. If I were to apply a contact cleaner today I would used CAIG LABS "DeoxIT Gold G5".
After reviewing the schematics I saw that Fluke had "factory selected" resistors in series with each of the main resistors in the first and second decades. So I examined those "factory selected" resistors and found that they were a very low value, under 10 ohms. So I reasoned that the main resistors inside of the oiled filled tank (nominal 9.836k) and the second decade resistors (nominal 9.836k) had drifted outside of their original values when Fluke "factory selected" those resistors so that the 20 turn adjustment pots would be in the center of their range. I also noted that most were only slightly outside of their adjustment range. So I simply added a very small resistor in series (if it had to be higher) with the "factory selected" resistor, or placed a very large resistor in parallel (if it had to be lower) with the "factory selected" resistor (depending upon which way the main resistor had drifted). Both series or parallel trim resistors were the best tolerance I could locate locally, figuring that a 0.01% or even 0.1% tolerance resistor would have the best overall TC characteristics. In addition by either paralleling or series insertion of my resistors I would keep the original "factory selected" resistors and my resistors would have a very minimal TC effect on the adjustments. So far this has worked extremely well and I cannot detect an effects of my "home selections" vs. what room temp variations would happen when the 720A came from the factory.
Some people on this BLOG have said "Send it back to Fluke for repairs". I am a home hobbyist and NOT an accredited cal lab I am happy getting the results I have. I believe I saw someone on this BLOG who sent their 720A back for repairs at Fluke and it was around $7,000 or so for the "repairs". I would doubt that Fluke did much more than change the "factory select" resistors to bring the main resistors back into adjustment range. IF one of the main resistors in either the oil filled tank or in the second decade had been damaged then I am sure that Fluke would have replace the oil filled tank at some cost that probably would have put that cost well over $15,000 or more. No telling how much that oil filled tank costs these days IF they can even be obtained anymore. It does look like Fluke still offers the 720A for sale so I would also guess that repair parts can be obtained.
Also since the 720A is "self calibrating" and only uses an internal bridge to adjust the values of the nominal 10K steps of the divider to be equal, t he 720A is by definition calibrated. About the only way I can think of to verify the linearity of the 720A would be to compare it to another Fluke 720A, or the discrete steps of a Josephson Junction array. This is outlined in a Fluke app note which I have attached to this message.
Either way have fun.
Bill