Six identical filler stations on a 1756-L83E and the tags for them grew a station at a time over about four years. Some are Stn1_Count, some are FILLER_2_PARTS, one of them is just PartsB. Forty odd of them scattered through three programs, none of it done by the same person twice.
Finance want a shift report per station now and every attempt I make at pulling them ends up missing a station or picking up the wrong one, because there is no pattern to key off. I tried aliasing the lot into one program to tidy the front of it, which helped the panel and did nothing at all for the report.
I'm stuck between doing it properly and just renaming the forty and hoping. Before I rip it apart, is a UDT with an array of six the right shape here, or is that overkill for six stations and a counter each?
How does the report actually get at the data. Is something browsing tags by name, or is it a query against a historian that is already logging them?
FactoryTalk Historian SE, and the point list in it was typed in by hand by whoever set it up. That's half the reason it's a mess, nobody has touched the list since 2023 and I'm not sure anyone knows what half of it maps to.
On our side we would put all six stations in one DB as an array and let the report take them by index, station 1 is index 0 and so on. Far less to maintain than six named sets, and adding a station is one number.
Does Historian let you pull an array element straight, or do you have to give it a point name per member anyway?
Index works right up until somebody adds station seven in the middle or pulls four out for a rebuild, and then every report older than that day is reading the wrong machine and nothing anywhere complains about it. Seen that twice, once on a bottling line where six months of data was quietly off by one.
Build a UDT with the count, the rejects, the running minutes and a name string in it, make an array of six, and write the station name into the string member on first scan. The report keys off the name and not the position. You pay for it once and I reckon that is the cheap end of it.
Put a timestamp member in while you are in there. When the link to the report drops for an hour you want to know when the number was true, not when it turned up.
Fair enough, I guess the name member answers my objection. I would still keep it as one array rather than six separate sets, the maintenance is much the same either way once the name is in there.
If you have not built one before the layout matters more than the syntax does, someone did a decent article on it: https://plctr.com/user-defined-datatype-udt-usage-examples-in-rslogix-5000-plc/
Built it over the weekend. One UDT called Stn_Data with Count, Rejects, RunMinutes, a 20 character name string and a DINT for the last stamp, then Stations[6] of it. Names get written on first scan from one rung nobody else is allowed near. Historian points are per member now and the report picks the station off the name string, so a seventh is an array resize and one line.
Six of the old tags are staying exactly where they are because the PanelView references them by name and I am not opening that today. Cheers rsdave, kiran_s and wandak, the name member was the bit I would have missed.
Good. Resize is the easy half, the name string is what saves you.