These test sets came to being when the communications industry needed a good high speed automated test system to deal with the sheer volume of techno-widgets (cell phones an such) streaming off the production line. These test sets can replace a rack of test gear and the problems related to integrating each instrument in the test rack into a automated test system. Considering these test sets can replace all these individual items of instrumentation makes then a very good value for what they are intended to do.
Once these computer driven test systems are taken out of context or their original intent such as a lab / development / research / home hobby environment, their designed in strengths can become a significant dis-advantage.
While as a general rule this is correct, you completely ignore that the CMU200 was actually designed and used as a standalone lab instrument for manual use by a human, not an automated test system. Of course, like other lab test kit intended for manual use (i.e. the HP 8566) it comes with a GPIB port and can be remotely controlled, but that was never the main purpose of the CMU200, which was sitting on a bench and being used by an engineer working on RF designs.
The CRTU (which hardware-wise is mostly the same as the CMU200) actually *was* intended as automated test system, and this was done by adding a Windows 2000 partition which runs a dedicated test environment. However, the CRTU (which is a multi-boot device) comes with the same DOS software as the CMU200, and with that DOS software behaves exactly as a CMU200 (i.e. a standalone lab instrument).
This does not make these test sets un-useable in these environments, their limitations must be considered and well understood before consideration for making meaningful measurements.
True, but that is no different to any other test instrument out there, no matter if automated or manually operated. Every test instrument lies to some extend, and a good engineer takes that into account.
Historically on this web forum there is much prejudice and distain for "old-obsolute and in the way of progress" test gear. Yet there are many historically significant and classic items of instrumentation that is never consider due to their age, lack of computer interface or fashion. These classic pieces of instrumentation can offer many lessons in design and technology and physics... all there for those who are willing to listen, learn of what they have to offer. All this for rather affordable cost.
I disagree. I've been here a while but I have yet to see any "disdain", in fact, there's probably still a strong preference of older, analog instruments over their modern equivalent, probably resulting from the fact that a large part of forum members are hobbyists, and hobbyists tend to have small budgets which usually means older and simpler instruments.
But its true that very old test instruments can often be had for little money these days. However, what you ignore is that in many areas even old high end kit performs worse than modern entry level equipment. In some areas (i.e. scopes) the change in technology is pretty significant, as what once was an instrument which didn't do much else than displaying (parts of) a waveform has now become full signal analyzers. Also, many of the techniques used with old kit no longer make sense or are even counter-productive with modern kit.
You also forget the running costs. It's all fine and dnandy when you can get an old instrument for cheap, however when it's more than two and a half decades old the likelihood of failure increases dramatically, and often repair turns out difficult because parts are no longer available or are generally unobtainium (i.e. some ICs from Tek and HP), with a good chance rendering the investment into a complete loss.
There's also the issue of size, noise and power consumption, all areas where old test instruments don't really shine. Many hobbyists are severely space constrained.
In the world of digital and computing, faster processing, more memory, the ability to crunch information often means greater measurable functionality and what a computer can achieve. The analog-RF-microwave and physics world does not work this way. Nature is what it is and individuals who endeavor into exploiting nature to create a technical widget is forced to make a compromise deal with this hash reality.
Not really, no. Of course the physics behind hasn't changed, but the simple fact remains that analog instruments are severely limited in what you can do with them. For example, have a look at the HP 8566 Series, which by many engineers is considered to be the epithome of analog SAs. But all that huge box does is giving you a display and some markers, but in terms of measurements its pretty dire (even for basic stuff like OBW). You can compensate for some of that by connecting it to a computer, but the box itself is pretty basic. The FFT option is utterly basic, and again pretty useless without an external computer.
It's even worse in other areas, i.e. scopes. Analog scopes give you a waveform display but that's about it. If you're lucky you get some primitive storage capability. Good luck trying to find out what the various frequency components of a square wave signal is, or how a pulse variies over time, or the recurring and intermittend jitter components, or anything else for that matter. The timebase of most analog scopes is pretty inaccurate, and the presence of certain by-products (i.e. noise) might not even be noticed by the operator as it might not even be visible due to insufficient brightness.
You also completely ignore the change of what signals are measured today. In the good old days most hobbyists would probably be playing around with some amplifier, audio equipment, radio kit or other analog stuff. These days it's more like Arduino & Co, fast digital busses, digital transmissions, SDRs and other pretty complex things. Old analog test kit is simply insufficient to extract relevant measurement data from such signal environments.
This is one of the reasons why good spectrum analyzers and other properly designed and built analog instrumentation holds their value on the used market. Consider why the hp 8566 stayed in production for nearly two decades and continues to hold their value to this day.
The reason the 8566 was produced for that long is simply because there was no need for a replacement, Tek was pretty crap in producing SAs, and Tek and HP were essentially the to-go brands for most companies and government agencies around the world. It was also made in the US, which was (and still is) a big thing for the US government, a spender of extraordinary amounts of money for such kit. Also, high end SAs were a pretty small market, and for many manufacturers it simply wasn't worth (or they were simply incapable) to justify investing lots of money into it, as even if they did HP would still outsell them.
That means, without any other major challengers, HP could continue selling the 8566 for a long time without having to invest much money into a product update. It also helped that government customers paid quite a lot of money for this device to be available to them for an extended period of time. HP was merely riding it out.
The other thing is that before the end of the 90's available computing power was limited and simply insuficient to do advanced signal analysis. Only after computing power increased all the functionality that a modern SA is capable of became possible.
The reasons the 8566 is still somewhat expensive as 2nd hand device are that the choice of old lab-grade high-end SAs is pretty limited for the reasons stated before (it was a small market after all), and that it carries the HP label, which is the same reason as to why primitive old Tek DSOs with awful specs still fetch insane prices. The name has a tremendous pull.
However, as good as the 8566 is, these days, if I needed a reasonably priced high-end lab SA I'd rather hunt for a R&S FSIQ than a HP 8566. The FSIQ exists in various variants up to 26GHz and has similarly good specs as the HP, but is faster, smaller, less noisy, comes with a color TFT display instead of a green CRT, and offers tons of analysis capabilities you won't get with the HP. And often enough the 26GHz variant goes for around the same as a 8566B in decent condition. And as a bonus, the FSIQ can emulate various HP SAs including the 8566 on the GPIB bus.
hp is NOT the only manufacture to make spectrum analyzers, others include Avantest, Anritus, Rohde & Schwarz, Tektronix and others. Historically Systron Donner, Ailtech-Eaton, Polarad, Cushman and others also once made SAs.
Older Tek SAs suck (the 492P was the worst SA I've ever had to use). R&S did't produce many SAs before ~1999 and resold Advantek midrange SAs instead. Most of the Anritsu devices offered as "Spectrum Analyzer" on ebay are actually cell phone analyzers, many of which have pretty dire specs (worse than the CMU200/CRTU) while regularly costing much more.
The others mentioned are pretty antique, some even come with valves and mechanical dials. Not sure that is a good investment for a hobby lab, these things are more suitable for collectors.