Wide (and slow) analyzers are readily available used.
Indeed. I picked up an HP 1660A with 136 channels on Ebay for < $300.00 USD. It's not fast enough for a lot of newer applications, but I doubt that I'm ever - as a hobbyist - going to need more than 136 channels.
I'm also considering picking up a Tektronix TLA 704 or similar at some point, just because A. they're fairly cheap, B. I don't have any Tek kit at the moment, and C. I sort of collect logic analyzers.
And while we're on the topic of "setting up a bench", if anybody cares for any advice from a guy who's been a beginner for 40 years now, here are a couple of random thoughts:
1. A good bench power supply is worth its weight in gold. I use a Rigol DP832A, but any quality supply should be fine. Some key aspects of a real lab power supply (as opposed to re-purposing an old ATX power supply, etc.) - You get adjustable voltage over a fairly wide range, and down to a couple of decimal points of precision. You can also vary the voltage across multiple channels simultaneously, which comes up more often than you might expect. Needing say, a 3.3V rail for logic "stuff" and simultaneously needing a 12V rail for powering an actuator or something. And having a 3rd independent channel really gives you a lot of flexibility. The other big things are a constant view of current on a given channel (how many times have you wanted to know how much current a component or subassembly was
really drawing) AND current limiting. It's nice to be able to preset a current value and know that you can connect something to the supply with no fear of blowing something up by dumping too much current into it.
2. I'm a software guy and a data/analytics guy by trade, so I may care more about this than other people, but I think a
really important factor of test equipment is the ability to interface with it from a computer and log data. I make it a point to always buy gear that has either an Ethernet port with LXI, or a GPIB connector, or worst case, plain-old RS-232. This whole thing is one reason I try to keep the number of brands of gear I use to a minimum (I use almost exclusively Rigol and HP/A/K kit). It's well known that LXI and other protocols aren't always implemented consistently across manufacturers, so I wanted to be able to fiddle with an open source library or what-have-you, get it working for manufacturer X and then mostly not have to worry about it. I have a separate PC that I use for data logging from instruments and I will eventually move that, and all the test gear, to an entirely separate network segment or VLAN within my network, just to keep things neat.
3. I regret waiting as long as I did to buy a proper temperature controlled soldering station. Once you get one, you'll never want to touch one of those cheap Radio Shack "pencil" soldering irons again.
4. I'm not convinced that it's important to use expensive solder, but I will argue that it's important to keep different sizes of solder on hand. I find that using the appropriate gauge solder makes a big difference, and it really does vary depending on what you're doing. Outside of that factor, I haven't found that using 60/40 vs 63/37, using silver bearing or not silver bearing, using multi-core or not, etc. makes a huge difference. I keep solder here from Multicore, Kester, Chipquick, and even some of that exotic stuff that contains silver, copper, and FSM-knows-what-else, and I still find that I reach for an ancient tube of Radio Shack 0.31" silver-bearing solder more often than about anything else. YMMV. *shrug*
5. Additional flux really does seem to help though. I keep a bottle of MG Chemicals flux and an eyedropper handy, and often apply some flux directly to whatever I'm working on. It really helps, even when using the "nice" Multicore solder or whatever.