Tom Burke has had enough with GPIB, calling it as finicky as the ISA bus. Do you agree, or should GPIB continue its long life?
Let's talk about addressing devices, and communications in general. If you're like me, and you're old enough to remember the Industry Standard Architecture (ISA) bus in PCs, you'll understand. In those days, you'd get your shiny new computer expansion card, flip some DIP switches for the base address and maybe an interrupt request line, stick it in an ISA slot, and then find out it wouldn't work. You'd go through the painful exercise of moving cards from slot to slot, changing their addresses, and generally just experimenting until you could communicate with the card.
Adding GPIB to a desktop computer used to mean installing a card. Now we can at least use an adapter from USB or Ethernet.
In my experience, GPIB is finicky this way. Once I got a chain of devices working and decided to add a device, there was a reasonably good chance that I would have to go through this exercise again, jockeying the position of devices on the chain until all of them were happy.
This leads to the second big issue: cabling. Now, these cables aren't small. They're about 0.4 inches in diameter, which makes them hard to bend, twist, or do any of the other bad things that you have to do to cables to get them to fit inside your test rack. This means you can't run a pure star topology like modern Ethernet, and because of the way the cables get stacked and snaked around, you usually don't end up with a purely serially connected daisy chain, either. Instead, you wind up with some crazy topology that looks like a tree with a couple of branches.
So this is a threefold hit. First, all those twisted cables have a tendency to wear out after a while, even if you're not moving them much. Next, some of those crazy topologies just won't work (even though they are supposed to). Last, three-deep stacks of connectors are a real pain to pull, loosen, move, and reconnect when a piece of equipment needs to go out for calibration.
Then there's the ever-present battle of compatibility versus compliance. Granted, this is something we see in a lot of places, but it seems a bit more prevalent in the GPIB equipment I've used, particularly in older equipment. Basically, you get a vendor that makes something "compatible" with your system, but not 100% "compliant." Then it starts responding and doing things when you're actually talking to other devices on the bus -- not fun at all.
I'm not saying that all (or even any) of these things are something you'll run into if you decide to use GPIB equipment, but that's been my regular experience. I'm pretty sure the only reason GPIB is still around is because of legacy equipment and even legacy engineers, the kind who say, "We have this, it works (sort of), and we know how to use it." This is a good reason to use something, but it's not an honest comparison to other methods. Modern equipment is much easier to use. The software interface is much cleaner, and cabling is smaller and easier. There's a whole litany of good reasons not to use GPIB if you're not already stuck with it. Even if you are stuck with it, I'd recommend you take every opportunity you can find to replace it.
Though GPIB might still be lingering on, I'd be more than happy to celebrate its demise.