Slave node ID? Really?
From its origins, the Modbus protocol has used one byte to address 256 slave nodes, reserving address zero as a broadcast (no reply) ID. That means there are millions of devices that use one and only one byte for the slave node ID address.
If that part of the Modbus message was actually altered, then no device using conventional Modbus will ever work on the network, because the added 2nd byte will be misinterpreted as the function code and all other following bytes are offset from they are expected to be. Communication would never happen because no slave would ever recognize its ID, so no slave would ever reply.
Furthermore, for serial RS-485 the practical limitation is 32 devices, due to the electronics. Modbus TCP has a limitation of 256 devices with a 255.255.255.0 subnet mask, but that 255.255.0.0 would get you 254*254 slaves.
So a need to alter the fundamental slave node ID field does not make sense.
Furthermore, it is highly unlikely that a commercial PLC would even offer access to the underlying Modbus protocol, because after all, it's a protocol - a set of rule for the frame format. One doesn't mess with frame formatting and get successful communications.
5 digit or 6 digit addressing?
It is far more likely that the issue is 5 digit vs 6 digit register addressing.
The early Modbus implementation were hampered by limited and costly memory restrictions.
Although the protocol provides for register addressing from by using a 2 byte, 16 bit "Quantity of Registers" field, the implementations of that era frequently limited decimal register addressing from (4)0001 to (4)9999 (known now as "5 digit addressing"), or one less than 10,000 registers by only allowing a 4 digit decimal entry when defining the register addressing. It wasn't the protocol limitation, it was an implementation limitation.
[note: The leading numeral (4) indicates Function Code 3 Holding Register data. The leading numeral is not used in the actual message frame, it used only for humans to identify the address as a FC3 Holding Register address.]
As the general availability of memory increased vendors began implementing the full range of decimal register addressing, for example (4)00001 to (4)99999, or one less the 100,000 registers (known now as "6 digit addressing")
Many slave vendors are now assuming that the master/client can address the full 6 digit register range up (4)99999 and are mapping their data into registers above (4)9999. If a master/client is limited to the older (4)9999 (so called 5 digit addressing), then it has no means of addressing those registers above (4)9999.
I don't know of a means other than empirical trial to determine whether a master/client is capable of 6 digit addressing. I've never seen mention of it in a spec sheet or even a manual, it just depends on the acceptable numerical entry when defining an address.
I would assume that an OPC server's Modbus driver would handle 6 digit addressing, just because those vendors see so many devices and have to write their software to accommodate the full market.
