(1) BPMS

Prior to data collection, check the bpm calibration. Consult with OPs on calibrating bpms, particularly if a fair number are reading "bad cal" or show discrepant TMITs. Keep in mind that every bad bpm will cost you about 1/2 hour to find via RESOLVE; it takes this much time because you are at the same time searching for an error in your lattice, and therefore some bpms will be giving good readings that are nonetheless discrepant with the fitting. The only bad bpms that are easy to remove are those which read zero in both planes on multiple tracks, or otherwise fail to respond to beam motions on multiple tracks. A listing of "bad bpms" will be the first useful result of your RESOLVE analysis.

In the course of this you should come to understand your bpms:

• Are bpms multiplexed?
• If so, are they revealing shot-to-shot jitter in beam parameters (e.g. energy)
• What kind of averaging is available from the control system---and is it useful for your analysis?

(2) REFERENCE ORBIT

Check the current absolute orbit. Your initial RESOLVE studies will employ linear optics about this orbit. You must decide if this is a good choice of reference orbit for the study you wish to perform. For example, if the orbit includes a large bump of some kind, find out why and decide if this is the orbit you want. Avoid a large bump in a known nonlinear element, unless you wish to take special account of this in setting up your .RESOLVE beamline specification. Anything but a largely "flat" orbit should be questioned. Record corrector strengths and bpm readings for this orbit (.e.g from the STEER panel). Be sure this data is backed up.

(3) LINEAR OPTICS

Next you will need to calibrate the model beamline you are using, i.e. remove errors in known linear elements. This involves calibrating corrector strengths, power supplies, and quad trims. We distinguish between these errors, and errors of higher order or coupling errors. This involves taking a track or two of bpm & corrector data for each corrector you wish to calibrate (you should do all of them). In analyzing these data you will find certain "good regions", regions where orbits on multiple tracks (of different phases) can be fit for the same lattice parameters (power supply group settings and individual trim, corrector settings). This step establishes your linear optics model for the lattice. Tracks may not fit well throughout the lattice, at this point, and the presumption is, that this due to a higher-order error or errors in the lattice. A table of corrector calibrations is the second useful result of your RESOLVE analysis.

(4) At this point, having fixed the linear optics parameters, you are ready to begin searching for higher-order errors in the lattice. A brute force approach would involve iterating over the following cycle:

• set a single corrector to a large positive value,
  while avoiding beam loss,
• scan in bpm & corrector data,
• set the corrector to a large negative value,
  while avoiding beam loss,
• scan in bpm & corrector data,
• restore the corrector to its original value
• proceed to the next corrector.

In practice it will not be necessary to do all correctors, however, this step can in principle be combined with step (3), so it is in fact useful to do all correctors.

(5) CLOSE BUMPS

To isolate errors it will be quite helpful to follow procedure (4), except via closed bumps that probe localized regions of the machine. For the NDR, 3-corrector closed bumps are available via multi-knob. Many of these actually do produce closed bumps.

(6) ERRORS

At this point you are ready to begin analyzing your tracks to identify errors (to be cont'd)