A register change should be invisible behind the hatch: the same stations, the same tickets, printed by something new at the front.
A decision to change point of sale starts at the front of the house, with hardware, card readers and reporting. The kitchen hears about it late, and what it hears is that everything behind the hatch is about to be rearranged too. It does not have to be. Which station makes what, where the paper comes out, how a ticket reads at arm's length, who calls a collection - none of that needs to change because the register did. A switch done well is invisible from the pass.
So approach it backwards. Rather than starting from what a new system can do, write down what your kitchen already does and treat that as the specification the new register has to satisfy. Anything it cannot reproduce is a change to service you are choosing, not a detail to discover on a busy night.
With that description in hand, most of the switch is configuration rather than construction: rebuild the station list, assign every item to its station, and lay the ticket out to match the one you photographed. Where two counters sell the same thing, each gets its own copy of the item pointed at its own station. An item with no station is the line that quietly never prints, so walk the finished menu once looking for exactly that.
Keep the metal if you can. Ordinary thermal station printers speak a common standard, and a register that insists on replacing every one of them is asking you to re-cable the building to solve a problem at the till. Ask instead about the three things that decide whether tickets survive a bad night: a paid order is recorded somewhere durable before anything tries to print it, a ticket that could not print is retried on its own as soon as that station is back, and every ticket carries an identifier tied to its station so a reconnection produces one ticket rather than three. Nobody demonstrates those unless you ask.
Do the changeover on the slowest service you have, at one counter, with the old system still reachable. Set two printers on a table the day before and ring real orders at them: one item for each station, then a mixed order across three stations, checking each ticket carries only its own lines under the same order number. Ring an item with modifiers and read the ticket from arm's length rather than from above it.
Then break things deliberately. Pull the paper out mid-order, finish the order, reload, and watch the missing ticket arrive by itself. Send the same order twice and confirm one ticket at each station. Switch a station off completely and see whether anybody is told, because a silent counter is the fault that costs you orders. Compare the end-of-night totals per station on both systems before the second counter moves across. Done this way, the kitchen's share of the training is one sentence: the tickets look the same and they come out of the same printer.