In my experience with Cu exploration in NW QLD and Western NSW and with aircore and RC drilling only (so pretty well homogenized samples) I remember seeing a pretty good correlation between the Niton pXRF Cu values and the actual lab assays. In the default calibration our Niton pXRF's were under-reporting Cu by about 10-15% in the 500ppm to 3% range (most interesting range for Cu explorers), its been a while since I have looked at it but I remember that at values above 3-5% Cu the Niton pXRF was less accurate and precise but that didn't concern most as at that level on an XRF you know that it should be going to the lab for a proper ore grade determination assay anyway..,... at the low levels of say 50-500ppm Cu the Niton XRF was also a bit less precise and accurate but that's getting down into the limits of detection of the instrument (and not anything interesting to Cu explorers anyway).
With proper calibration and proper sample preparation (crushing and grinding in a LM3 or 5) I have seen Niton pXRF's used in grade control samples in base metals and they were reported to have a very high level of accuracy and precision (as you'd expect if they are used for grade control decisions on open pit benches). I haven't used Olympus portable XRF much myself, but others have been pretty happy with them.
Interesting that in your experience that you have seen such inaccurate "over reporting" of Cu values by the XRF compared to the lab assay result. Were these discrepancies in drill core or in RC drill samples? Drill core by its nature is always more problematic to get an accurate XRF reading from because of the sample heterogeneity problem the sampling bias problem and the small sample window size of most portable XRF's.
TL;DR version is that pXRF on RC or aircore chips and dust = good accuracy and precision, but pXRF readings on drill core = accurate but not precise enough to be useful.
Explainer for accuracy and precision in visual below for those playing at home:
(20min delay)