Mam uz dlhsiu dobu ulozeny tento komentar z redditu z r/PLC.
Zemnim svedomite obe strany a vsetko je ako ma byt.
No this from a time when the noise in a plant was lower frequencies and you could get away with this. Typically 3rd harmonic of power frequency was your biggest concern. Now, all the constant current LED lighting etc chopping up the line to make constant current buzz out to 10 Meg's and beyond.
A wire earthed at one end is an antenna, and with EMC up to 10MHz and beyond these days, you can easily get a quarter wavelength resonance going - around 7.5 metres at 10 MHz.
Grounding both ends makes a Faraday shield, that will protect you from the E field, but twisted pair is what protects you from the B field.
But there is more to it, there's lots you need to do right to fully avoid EMC/RFI.
And I know what you mean about lifting or clearing far end screen can often improve things, but probably not in fault or surge events. I'm electrical engineer first, but I've been doing control systems over 30 years and seen a lot of different shit go down. Most installations I have seen have shortcomings, often serious, when it comes to earthing, bonding and shielding.
Look up Siemens document "S7-1500 and ET-200 Designing Interference Free Systems" to see contemporary thinking.
And if you are getting bad earth loops because you are earthing at both ends, you probably have bigger problems you might not find about until you get a fault. But in the worst case, you can run a parallel earth conductor out for the earth at the far end, and that will also reduce your exposure to lightning events tenfold as well.