Rigol RSA3000/5000 spectrum analyzers: question for owners or operators

[VK6HP]

Hello everyone,

I have a Rigol RSA3015E-TG analyzer and have been largely satisfied with it in the 15 months or so I've owned it.  A couple of months ago I noticed an anomaly with the noise floor and calibration in the wideband real-time (RSA) mode, whereby the noise floor of the instrument jumps many tens of dB once spans (FFT total bandwidth) exceed exactly 3 MHz, sensitivity is much reduced, calibration of levels is wildly inaccurate and large baseline curvature is evident over e.g. 5 or 10 MHz spans (spectra unable to be aligned or flattened).  RSA behaviour below 3 MHz is pretty much normal, as is GPSA (sweeping analyzer) operation.

The analyzer is still under warranty and I just received it back from Rigol after the Australian distributor arranged for it to be sent back to China for repair.  Unhappily, the bad behaviour persists.  With the holiday season now upon us, nothing much further can happen until early January.  In the meantime, I'd be interested to hear from any other owners who might be able to report their RSA noise floors at (say) 100 MHz, with spans of 100 kHz, 1 MHz, 5 MHz and 10 MHz.  (If you have a fancier analyzer version with more bandwidth, feel free to go further).  No need for much accuracy: the effect I see is a huge one.

Most likely this RSA issue is simply a fault in my instrument but I've been surprised at what's gone unnoticed (or unadvised) in the past - for example, my report of a ~12 MHz noise floor step seemingly triggered a Rigol web FAQ to re-assure RSA3000/5000 owners that calibration is not affected. A quick look at the just-returned analyzer leads me to think that Rigol have now upgraded the firmware/hardware to mitigate this previously-noted effect.

Thanks, and regards,
Peter.