Unexplained disconnection of instrument connected via ethernet/VISA

[d.sand]

Hello all, 

My lab is using a Honeywell 2182 UDA to interface with a Durafet pH probe.  The UDA is connected directly to a Windows 7 PC via an ethernet cable, and we view the outputs (Temp, pH) via a Modbus server in Distributed Systems Manager, from which we can pull shared variables into Labview code.  

It's a finicky setup, and we've only had sporadic success.  The most difficult factor is trying to help NI-MAX establish a VISA connection to the device via entry of a raw socket IP address and Port Number.  This, mind you, is darn near impossible.  Finding the right combination of IP, port number, DNS server, etc. is a nightmare.  On the few occasions we've done this, I've been able to get my data to work with my VI in Labview without a hitch.  Until today, that is.  

Tech support from my university was able to help me find the correct IP subnet to use, and then assigned the computer and device to the same subnet while varying the last octet of the IP address (192.168.1.1 for the ethernet adapter on the PC, 192.168.1.254 for the device).  That worked fine: NI-MAX established a connection, I set up the Modbus, imported the variables into my project, and all was well with the world, until after about 45 hours, the connection cut out, suddenly, while I was running my VI.  I wasn't touching the computer at all, just watching the block diagram while the bits flowed through my code.  My PI had checked DSM every few hours since the connection was made, and we were still getting pH and Temp data.  

Pinging the IP address from CMD doesn't do anything, nor does typing the IP in a browser window (another way to view data).  I'm at a loss here, and hoping anybody can help.  

[d.sand]

For those that are interested, more detailed information concerning our connection strategy for Durafet electrodes and the Honeywell UDA can be found here: https://www.frontiersin.org/articles/10.3389/fmars.2017.00321/full#supplementary-material 

[sherlockholm]

d.sand,

If you can't ping the device that is a big problem.

Do you know if the Ethernet cable is bad, or if the network card in your computer is bad?

Are you able to try this instrument on another computer?

NI IO Trace is a helpful troubleshooting tool, but if you can't ping the device I doubt this will be much help.

Also is this how you are setting up your instrument?

[d.sand]

sherlockholm,

Thanks for replying.  After IT at my university took a look at it, one of the things they did was bring in another Ethernet cable, which was included in the eventual (if ephemeral) fix.  Thinking of this, I also tried the same procedure on my lab's other machine: a Windows 10 desktop with the same software.  We didn't establish a connection there either.  

That link describes our process to a "t".  I can try IO Trace when I get into lab on Wednesday, but I suspect that you're right about it if I can't ping it.  

Our IT had some suspicion that the problem is related to the Net Admin's policy of rotating IP addresses every so often, so that even though the network adapter I was using to plug in the Ethernet cable was set to a static IP, some unknown process was interfering with the connection.  I'm far out of my depth when it comes to network administration, so I can only speculate here; the IP address rotation they cited was 8 hours, yet our machine lasted 40 without a problem.  

I began wondering if this could truly be a hardware issue; perhaps in the Ethernet communication card enclosed in the instrument, but an inspection of it found no loose connections, no dust buildup, nothing.  

[d.sand]

For whatever reason, a port reset on the instrument UDA has temporarily fixed the problem; I wonder why, though?  Does anybody have any history of this occurring?

[Minions]

I have seen issues at times when the VISA configuration is different from the port configuration.  If you view it in NI-Max before choosing an option, they can show as a possible conflict.  When faced with this option, we always select the VISA option as the preferred configuration for our LabVIEW systems.

[d.sand · Accepted solution]

That seems like a good point.  I'll incorporate it into the future instrument design if another problem arises.  

Update: The instrument spontaneously began working again after a port reset on the instrument UDA itself.  No settings were changed, so it's a mystery to me as to why this might have changed anything.  Thank you to all contributors for your help.