I promised in the TEA thread to share what I had here... Over on the groups.io HPAK forum I ran into the engineer who was responsible for the analog design of these meters, and... well... pretty soon I found a copy of each and compelled them to join me on the workbench.
One of the interesting things I learned was that even though these two have slightly different specs, the HP engineers considered them pretty much equivalent in terms of long term accuracy. I don't have anything good enough to do proper analysis of them, so that's not something I'm going to worry about. They both present an interesting contrast to the Flukes with their bright LEDs and mechanical front panel switches.
The 3468A, listed as "fair" by the seller, was in fact nearly perfect barring some grime which cleaned up quickly. I don't care about the odd nick and scrape as long as it doesn't impair usability or make the display ugly. The handle was the only thing which had any cosmetic issues anyway.
The 3478A, shown with some "cosmetic" issues, looked as though someone had dropped a boat anchor (figuratively or literally) on it. Nothing was obviously broken, and the dent in the case, though serious, clearly wasn't going to crush anything inside so at the price I took the chance. The only thing I was worried about was a crushed corner of the front bezel which potentially could have damaged the display board.
When I got the 3478A, I tested it immediately and it seemed to work just fine on all modes and ranges. I managed to lever the case off without further mangling anything, and freed the bezel as well (you pretty much have to disassemble the instrument to get the bezel off). The case is beyond my modest metalworking capabilities, so a friend with a shop is going to take a look at it for me. It's not all that bad, but I'd like it squared up and the dent removed so that it will at least slide on/off without too much scraping. As I anticipated, it didn't hit anything inside - there's a lot of free space above the PC board in these units. The inside was very clean.
The bezel: apparently the early units have a metal bezel but mine was the later plastic model. It looked like someone had dropped it (or possibly the boat anchor which landed on it bounced...) but the design prevented any of the damage from even reaching the front panel, let along propagating the force to other parts. It was surprisingly easy to heat the plastic with a paint stripper set on low and gently prod it back into shape. A little tan epoxy and some judicious sanding, and it looks OK. Not perfect, but at a glance it doesn't immediately get your attention.
The 3468A battery is 3.4V, which for an ancient 3V battery is pretty impressive. The 3478A battery is 'only' at 3.2V so it's a little closer to the tipping point. I don't have any way to get data out of the 3468A (in fact, I'm not sure there is a method to talk to an HP-IL meter and retrieve cal data?) but I could get it from the 3478A via HP-IB, I suppose. I'm really not sure I can justify building an interface just to extract a few bytes of data - or pay twice what I spent on the meter for a HPAK clone interface. I may just settle for doing a live swap on the battery.
Overall consensus: I really like these meters. They're fast, quiet, and even the LCD is easy to read in the lighting of my workshop. Unfortunately, now the HP meters are challenging the Flukes for bench space.