Carlo Gavazzi EM530 / EM540 Modbus register map

Device identity

What a tool can read to check that this map fits the connected device (SPEC.md section 4.7). Registers with a role are read like any other register.

Sources and notes

Source: Carlo Gavazzi, "EM530/EM540 Modbus communication protocol", public version 1.3 (2024-02-13), tables 4.1-4.24; https://www.gavazziautomation.com/fileadmin/images/PIM/OTHERSTUFF/COMPRO/EM500_CPP_Mod_V1.3_13022024.pdf

• Carlo Gavazzi EM530 and EM540 energy analysers (RS485).
• FC03 and FC04 return the same values; this map reads everything as holding registers.
• Tables 4.1 and 4.2 of the protocol hold the same values at two addresses; the second copy is in the group "By phase (copy)".
• Tested on EM530DINAV53XS1X (EM530 X), firmware 1.3.3, October 2026, over RS485. Checked: voltage, frequency, device and setup registers, the "By phase (copy)" group and the 32-bit word order. Current, power, power factor and energy were read with no current flowing, and the per-phase registers of L2 and L3 were not tested under load.
• A second unit was tested under load (about 15.5 A per phase) over RS485 at 9600 baud, November 2026: 252 of 257 registers read. Checked: voltages, currents, power, power factor, frequency, THD, demand, energies, the "By phase (copy)" group, setup and the 32-bit word order. Apparent power matches active and reactive power on every phase (L2: 2628 W and 2628 var give 3715.8 VA), the per-phase kWh add up to the total (20111.5 against 20111.7 kWh) and the 64-bit Wh copies agree with the 0.1 kWh values.
• Digital input status (768), Active tariff (769) and the firmware word (770, firmware 1.3.3) answered illegal address when read together as one block of 3 registers, but each read fine on its own (768 and 769 as a block of 2, 770 alone). The meter refuses a read that runs from 0301h into 0302h, so the map reads 770 alone (read_alone) and 768 and 769 together (read_group). The identification code (11) read 0, which is not in the model list, and FC17 got no reply.