I know you're probably the wrong person to present this argument to, but if you're interfacing on a Windows system to a GPIB instrument via Ethernet, I would strongly argue for the GPIB-ENET/100 solution. I realize that being an NI employee may weaken my argument, but I assure you I'm not just trying to "sell you something".
The GPIB-ENET/100 is 5+ years old. It has been designed, tested, released, tweaked, fixed, re-released, a number of times in its life. There was even a predecessor product called the GPIB-ENET. Lessons learned on the GPIB-ENET were of course incorporated into our GPIB-ENET/100 design. The driver software, NI-488.2 v2.4, has gone through a similar number of tweakings, huge amounts of internal testing, and massive external testing by GPIB customers all over the world. My point is that the GPIB-ENET/100 and its corresponding driver software are nearly rock-solid, and they didn't get that way overnight.
The path on which you are embarking by designing one from scratch will produce a fine product, but like any new product (and like the GPIB-ENET/100's first iterations), it will be frought with bugs. Some bugs that you will have to fix in hardware, some bugs that you will fix in software, and some that the users of your product will have to work around because "it just works that way". I'm not sure that the cost in man-hours of 1) hardware design 2) firmware design 3) software design 4) testing 5) end users working around issues has fully been reconciled against the cost of 50 GPIB-ENET/100's. We do have customers who purchase hundreds of GPIB-ENET/100's because they find it a more reasonable choice than building their own from scratch.
In any case, those are my thoughts. I apologize for the rant. If your management still decides to go down the custom path, we will of course continue to try to help you on the forum as best we can. Good luck!
Scott B.
GPIB Software
National Instruments
