Our HP8594EM does the same thing if you are spanning too wide a frequency range. I suggest you break the CISPR 0.15 to 30 MHz band into three subbands as: 0.15 to 1 MHz, 1 MHz to 10 MHz, then 10 MHz to 30 MHz. When measuring a signal very close to 0.15 MHz, I further reduce the sweep range to near zero to keep the marker software from wandering out-of-band.
My supposition is that the frequency resolution is too rough if you span the whole CISPR-22/11 conducted spectrum, the analyzer firmware can't handle it well enough (and evidently doesn't do any bounds-checking against the span criteria) and grabs the reference oscillator or some other strong signal just outside the desired frequency range. When I'm done taking all the individual data points, I return to full 0.15 to 30 MHz span, max hold for a few sweeps, and plot the tabular data along with a full sweep; unless I have too many data points to plot, then I plot the subbands with associated tabular data. The HP8594EM is a darn handy instrument, but I've bumped into other silly bugs (I never did catalog them all), which suggests that the shipping schedule won out over the QA function. Maybe I should switch to LabView to run it... (Just *who* makes that nifty LabView anyway?) ;) Eric Lifsey National Instruments www.natinst.com [email protected] on 09/24/97 06:23:19 PM Please respond to [email protected] To: [email protected] cc: (bcc: Eric Lifsey/AUS/NIC) Subject: H/P 8590EM spectrum analyzer: marker & list measurements We have a Hewlett-Packard 8590EM spectrum analyzer used for conducted emissions testing in the 0.15-30MHz range. The analyzer has a feature that lets you adjust a marker to a given frequency, then press a single button to display the signal amplitude using peak, quasi-peak, and average mode detectors. Unfortunately, there are times when the analyzer 'jumps' frequency during this measurement, and gives the amplitude at 1Hz! Has anyone else seen this behavior? Pat Lawler [email protected]
