A gateway needs two network relationships, not one, and they’re easy to conflate if you’ve only thought about “the network” as a single thing. One side faces your machines, the plant network where the PLCs and controls live, and how many of them one gateway can reach is its own question. The other side is the uplink, the path the gateway uses to reach Spall’s cloud. This guide is about that second one, because it’s the one decision every install actually has to make, and getting it wrong is the most common reason a new gateway shows up “offline” the first week it’s live.
Why wired is the default, not a preference
A gateway with an Ethernet cable running to your plant switch, and a path from that switch out to the internet, is the simplest, most boring, most reliable uplink there is. It doesn’t need a signal, it doesn’t share bandwidth with anything wireless nearby, and it doesn’t depend on a battery, a modem handshake, or line of sight to anything. If a plant network already has internet access and a spare port near where the gateway sits, wired wins by default, not because it’s fancier, because there’s nothing about it left to go wrong.
The tradeoff, and it’s real, is that it requires exactly what a lot of plant floors don’t have on hand, a data drop in the right spot, or IT willing to run one. That’s the actual reason the other two options exist.
WiFi: fine when it’s actually there, not a thing to force
A gateway can join a plant’s WiFi network as its uplink, and it works exactly the way you’d expect, as well as the WiFi itself does. That’s the whole caveat. A clean, strong signal near the gateway’s mounting spot means WiFi is a perfectly reasonable choice, especially as a step up from cellular when a wired drop isn’t practical. A weak, congested, or metal-shrouded corner of a machine shop is a different story, and it’s worth checking signal strength at the actual mounting location before committing to it, not assuming from where the office WiFi reaches.
The other real cost of WiFi is process, not technology: getting a new device onto a plant’s WiFi usually means a conversation with whoever manages that network, credentials, an approved device list, sometimes a change ticket. That conversation can take longer than running a cable would have.
Cellular: the answer to “what if IT is slow”
Every gateway ships with its own built-in cellular modem, and it’s there specifically to skip the part of a rollout that stalls projects: waiting on a plant network conversation, a firewall exception, or an IT ticket that’s third in line behind two other priorities, exactly the dependency worth surfacing in the questions to ask a vendor before you sign. A gateway on cellular is online the moment it’s powered up near a machine, no coordination with anyone’s network team required. That’s a genuine advantage for a fast pilot, or for a site where the plant network is locked down in ways that make getting a new device onto it a real project of its own.
The tradeoff is that cellular is metered. The gateway is built to be sensible about that, it automatically switches to a leaner, compressed transport preset the moment it detects it’s actually on a metered link, so a temporary failover to cellular doesn’t blow through a data plan the way a naive always-on-fiber assumption would. But cellular signal strength varies by location the same way WiFi does, and a basement or heavily shielded electrical room can be a real weak spot for it, worth checking before assuming it’ll be a smooth primary path rather than an occasional fallback.
It’s not actually a single choice, and that’s the point
The three paths aren’t mutually exclusive, and most real installs end up leaning on that. A gateway with a wired connection and no cellular need never touches its modem. A gateway with a flaky plant network and a cellular modem as backup fails over automatically the moment the wire drops, and fails back just as automatically once it’s healthy again, no reboot, no site visit, no ticket. That failover is priority-ordered by convention: wired beats WiFi beats cellular, whichever paths actually exist on a given box get tried in that order.
The practical upshot: don’t treat this as a single upfront decision that locks in forever. A gateway installed on cellular during a fast pilot, because getting on the plant network would have taken three more weeks, can move to wired later once IT catches up, without touching the machine side of the setup at all. The uplink is the part of the install that’s allowed to change its mind.
What to actually check before install day
Walk to the exact spot the gateway will mount, not the nearest office, and check three things: is there a live Ethernet drop within reach, what does WiFi signal actually look like there if wired isn’t available, and does cellular have a usable signal as a fallback. That fifteen minute walk answers the uplink question more reliably than guessing from the building’s general network reputation, a plant with famously bad WiFi in the front office can still have a perfectly fine signal in the corner where a specific machine sits.
Quick recap
- The uplink and the machine-side network are two different relationships, don’t conflate them
- Wired wins by default when a drop exists, nothing about it depends on a signal or a device policy
- WiFi is a real option when the signal at the actual mounting spot is good, but expect a network-access conversation first
- Cellular ships built in on every gateway, skips the IT wait entirely, and downshifts its own data usage automatically when it’s active
- Failover between whichever paths exist is automatic, wired beats WiFi beats cellular, and failback happens the same way
- Check signal and drop availability at the actual mounting location before deciding, not from the building’s general reputation