VAN said:
I'm kind of late to the party, but any upgrade I do I try and rip out all old networks.
RIO? GONE
DeviceNet, hell no, gone.
etc. etc.
OK, they're ripped out. That's half the job done. But what are you replacing them with, Ethernet?
I'm always reluctant to advise the "just do what I do" option when someone is choosing a migration path. It, in my opinion and experience, and once more to the Forum, should be an application driven decision. What is the best fit for the existing machine or process going forward, to achieve the required goal, while also considering the general install base and end user knowledge at the installation, followed up lastly by the cost factors. If the cost for the recommended best option is prohibitive then compromises or alternative migration paths can be considered. This, depending on which option was deemed the best, could drive the migration path towards either all new or away from it to an alternative interim path, such as the one being implemented here.
It is very difficult for any of us, without more intimate knowledge of this existing installation and its users, to blanket advise a "rip & replace" based on our previous and possibly biased project work.
I would also certainly respect ASF's apparent experience portrayed on the Forum as being good enough to have already considered what is the best migration option here.
As for ripping out RIO or DeviceNet networks just for apparently simply existing?...
I could agree, to some degree, that RIO can be troublesome and has had its fair share of niggles over the years for various users, including myself, and the added complexities of configuring block transfers and I/O rack slot addressing has driven some to dementia. The Flex I/O point density v module size is also a negative but I could argue that POINT I/O is similarly lacking. However, the latter's footprint is quite small and they are more intended for moderately smaller installations, such as "on machine" distributed I/O. The use of RIO with PanelView terminals was also one of their more torturous offerings.
But I'll not have DeviceNet tarred with the same messy brush. It is, once properly installed and configured, one of the most robust networking architectures I have ever had the pleasure of working with. Granted, EtherNet/IP is by far the modern preference for CIP based integrated architectures and is indeed the way forward, for now, within the exploding IIoT revolution. But it will take something exceptional that the EtherNet/IP protocol and/or architecture provides, and that an existing DeviceNet architecture requires, before I would recommend replacing a working DeviceNet system with an EtherNet one, just because it exists. Again, if the existing application dependencies do not require it, I would not recommend such a migration path.
Also note how DeviceNet compatibility and functionality is still being maintained through to certain newer AB platforms. It's not going anywhere just yet. Granted, for new installs I would not be readily recommending it by today's standards, but for adding to an existing DeviceNet network, these options are still very important to have at our disposal.
Some installers (not necessarily you VAN) might just be preferring the "newer" Ethernet over "anything" else because they feel they have the necessary knowledge or experience to get and keep them up and running, and perhaps don't have as much of the proprietary knowledge or experienced required to do likewise with the likes of the somewhat more obscure DeviceNet systems. Standardisation may also be a driving factor here. But again, these desicion should not be just based on what suits the installer, with less, little or no regard to what best suits the application, customer or end users.
There is nothing essentially wrong with what each of us normally tends to do in these cases, but that doesn't necessarily mean it's also right for everyone else in a similar situation, that's all.
Regards,
George