R&S FSP3/7. After Wuerstchenhund tried to convince me (on an earlier thread) that these little analysers were better than the HP8566/8 I went over to the test/ATE area where we had several FSP7 models that were sat unloved on the shelves. It's not difficult to see why these analysers are not popular. They take ages to boot up, the display is gloomy and the far out phase noise performance is really poor. The noise floor performance is poor at low frequencies and the linearity suffers badly down here too.
Sounds like you have the older generation FSPs then. Like with many R&S gear you have to careful with what model you have exactly. In regards to the FSP, there are two variants: the older one (1064.4495.xx) and the later one (1164.4391.xx). The old version runs Windows NT, has a very slow processor (usually FMR5/6, i.e. some AMD K6 or intel Pentium3 Celeron w. 256MB RAM), and it runs WindowsNT 4. Boot times of these older FSPs are indeed dreadful with the slow standard EIDE hard drive, as is the UI.
The newer generation has an improved RF section, a much faster processor (FMR7 with Pentium-M and 1/1.5GB RAM, although a few early models were sold with FMR6+ with P3 900MHz), and it runs WindowsXP Embedded. The FSP7 models with FMR7 come with a SATA hard drive and boot noticeably faster (approx 40s), plus they can easily be retrofitted with a bog standard SATA SSD which reduces boot times to less than 30s.
I'd generally stay away from the older 1064.4495.xx variants unless its really cheap and you can live with its limitations. They are still better than the Rigol DSA800 Series, though. Like, a lot better.
Not sure why you say the display is "gloomy", it's a bog standard industrial VGA TFT display made by Toshiba that's in use in a dozen or so other instruments from various manufacturers. It's not a CRT. Maybe the backlights on your units are worn out.
You also have to check the firmware level, as there were various improvements in later versions. In my experience, equipment that is sitting in some corner often tends to not get updated, so that's something to pay attention to.
As to the FSP not being popular, well there seem to be really an awful lot around of them. In fact, back then the FSP was one of the best selling R&S SAs, and R&S does sell quite a lot of spectrum analyzers. Show me anything from Agilent (or anyone else) back then (2004-2006) in the same price bracket that offered a better performance and went up to 40Ghz. Anything.
Another bonus of these R&S analyzers is that they can emulate a lot of the old HP instruments, i.e. the 8566/8568/8594 etc.
They also offer some interesting options, like External Generator Control (option B10 on the FSP) which essentially allows to use an external RF generator to be used as Tracking Generator. This works not only with R&S generators but also with many HP and Agilent generators like the E44xx Series (although the R&S generators are faster due to their dedicated interface while the other generators connect via GPIB).
I do find it puzzling why people on Eevblog are often hyper critical (snobby?) about scopes and DMMs yet they are happy to spend big money on mediocre (often new or nearly new) spectrum analysers. I can understand the popularity of the Rigol 815 with its modern display and modern features but I can't understand why people buy the HP859x series or the test sets.
The simple reason people buy them is that not everyone needs high-end top-notch RF performance, for which you pay one way or the other (i.e. with lots of money for something modern, or for less money having to live with some 30yrs old huge, noisy and power hungry monster with CRT which due to its age alone is likely to die on you any day, where no new spares are available and where many defects pretty much equal a total loss). You can very often get enough relevant information from a device with less than top-end performance, and as a bonus you often get a much more modern instrument that takes less room and comes with conveniences like LAN or even just the ability to store screen shots and data directly to a USB drive. Believe it or not, these things can often be even more important than the best ever RF performance.
I also disagree that people here are snobbish re. DVMs or scopes, in fact the majority of threads in this forum revolve around cheap beginner's scopes (mostly Rigols) and the various bugs the Chinese B-brand scopes come with as standard. And it's pretty clear that a beginner doesn't need a 30+Ghz high-end scope, for most even the scruffy Rigol DS1054z does perfectly fine. The same is true for an SA. I seriously hope you don't want to suggest that some starter who wants to learn more about RF should buy an antique HP 856x boat anchor plus an (long unsupported) E4406A (which is only really useful with the Agilent 89600A software, a $10k+ piece that can't even be bought anymore and unless you happen to own a license you pretty much have to crack it for using it for more than 30days which isn't just a grey area, it's borderline illegal). You seriously suggest that is the path a beginner should go?
It's also somewhat funny that you bash pretty much anything that you don't consider equal to your old 856x boat anchors, but still say the Rigol DSA800 is OK, a device with aside from DANL performs worse than the SA in comms testers like the CMU200/CRTU (which also have a larger bandwidth), plus it suffered from a number of annoying bugs, and it really sees no light even against the old FSP variant. It's only attractive because it's really cheap.
I understand that you a RF guy by heart, and it's pretty clear that you have very high expectations from your test gear and a certain fondness for HP gear. But you're also often pretty blind to other people's requirements, and the dismissal of anything which isn't top-notch RF gear is a bit silly, really. There's a reason why the SA market not only offers high end instruments, and a lot of useful work is done every day in labs around the world with mid-range and even entry-level devices. You should really take the blinkers off from time to time.