To understand how to use your instruments and for basic surveying techniques, the first port of call should be to refer to the user manuals which came with your instrument. Rather than give yet another set of instructions on how to set out a grid and conduct a survey, this page will direct you to other websites which have already covered this area admirably.
For how to go about planning a survey and setting up a grid, you could do a lot worse than visit Sussex archaeological geophysics page, a web page set up by David Staveley. There are further links here on how to carry out resistivity surveys as well as a free (for non commercial use) resistivity processing software download called "Snuffler".
For a more in depth introduction, visit the Geophysical data in archaeology: a guide to good practice by Armin Schmidt et al, which is available on the Archaeological Data Service website.
It is always best to collect data in the field which requires the minimum processing after the survey is completed. Trying to avoid the main operator errors is a good start.
It is also important that you find a good zero reference point, otherwise your data will require more post-processing then should have been necessary. When using a fluxgate gradiometer system, first it is vital to allow the sensors to establish local temperature equilibrium (taking between 20-30 minutes, depending on the array) before establishing a zero point. Take the machine out of its box and assemble it as soon as you arrive on site. You can use this time to begin setting out your grid, if required. The zero point is a reference area which is found by walking around with the gradiometer to establish an average magnetic value, and then finding an area where a stable average reading can be obtained. The gradiometer is then set to zero over this point. This allows both negative and positive magnetic values (relative to the assigned zero) to be recorded. When necessary due to instrument drift, the gradiometer must be returned to this point to recalibrate the sensors. If this is not done regularly, then differences in the survey data will begin to occur.