For the wider instrument selection context, see the three-phase electric power meter guide<\/a>.<\/p>\n Ask for the register definition for the exact model and firmware. Determine the minimum value, maximum value, increment, decimal scale, sign convention and behavior at the limit. Some counters wrap, some saturate, some reset under defined actions, and some use a representation with a range that differs from the front-panel display.<\/p>\n Do not infer the communication register limit from the number of visible display digits. A display may change resolution, show a suffix or present only part of a larger internal value. Conversely, a gateway might store a smaller integer than the instrument provides. The narrowest representation in the data path can become the practical failure point.<\/p>\n The Modbus setup guide<\/a> addresses decoding, units and word ordering. Confirm those first. A signed\/unsigned error or an inconsistent read across multiple words can resemble a wrap without representing a real energy-counter event.<\/p>\n For a documented zero-based counter with a known modulus M, let C\u2081 be the earlier raw count and C\u2082 the later raw count. If no wrap occurred, the count difference is C\u2082 \u2212 C\u2081. If exactly one wrap occurred, the difference is:<\/p>\n \u0394C = (M \u2212 C\u2081) + C\u2082<\/strong><\/p>\n The modulus is the number of discrete count states, not always the highest displayed value. A hypothetical integer counter that runs from 0 through 999,999 has M = 1,000,000. If it moves from 999,980 to 25 with exactly one confirmed wrap, the difference is 45 counts. At an illustrative scale of 0.1 kWh per count, that equals 4.5 kWh.<\/p>\n Those values are arithmetic examples, not BY194 specifications. If documentation supplies a maximum engineering value rather than a raw-count range, also obtain the increment and reset endpoint. A floating-point energy register cannot automatically be treated as a fixed-width unsigned integer counter.<\/p>\n The formula becomes unsafe when the event could be a reset or when multiple wraps are possible. Adding one modulus may produce a positive number while still substantially undercounting the true interval. A plausible result is insufficient evidence that the selected explanation is correct.<\/p>\n A wrap normally has evidence compatible with the documented limit: the earlier reading is near the upper end, the next is near the start, and the resulting interval is consistent with known operating conditions. Even that pattern should be checked against settings changes, event logs and device identity.<\/p>\n A reset may follow an authorized command, commissioning action or a software restart. A replacement changes the physical instrument or data source. A communication error may produce zero, a stale value or a sentinel. Each requires different treatment. Never convert every negative difference into a wrap merely to keep a chart nonnegative.<\/p>\n The Prometheus documentation on counter functions<\/a> describes handling counter decreases as resets in its software monitoring model. That illustrates why a data system needs explicit counter semantics. Its reset interpretation does not establish the modulus or physical behavior of an industrial electricity meter; a documented meter wrap requires its own treatment.<\/p>\n Use the table as an investigation framework. Acceptance depends on the actual register documentation and retained evidence.<\/p>\n A cumulative counter does not preserve the number of times it has wrapped unless the device supplies additional information. Bound the possible energy between observations using a defensible maximum for the measurement point. If that bound is comparable to or greater than the counter’s energy span, more than one wrap may be possible and two endpoint readings are insufficient.<\/p>\n For a constant upper-bound power Pmax in kW and an elapsed time t in hours, the maximum energy bound is Pmax \u00d7 t in kWh. Use consistent units and justify the bound from the installation. A nameplate total that excludes generation, bidirectional flow or a changed feeder arrangement may not support the intended assumption.<\/p>\n An illustrative 100 kW bound over two hours gives no more than 200 kWh of accumulated energy in a unidirectional counter under those assumptions. This is a planning check, not a prescribed polling interval. Select a practical interval with margin, account for communication outages, and verify that the collector retains original timestamps and quality flags.<\/p>\n Useful corroboration may include the instrument’s local cumulative reading, a documented interval log, another meter at an appropriate boundary or a compatible pulse count. Each comparison has limitations. Two meters at different feeder boundaries can legitimately disagree because they include different loads or losses.<\/p>\n OpenEnergyMonitor’s pulse-counting explanation<\/a> describes the relationship between a defined pulse quantity and accumulated energy. Use the actual instrument’s documented pulse constant when comparing records; do not copy another meter’s example constant. Confirm whether transformer ratios are already included and whether the pulse collector itself reset or missed events.<\/p>\nObtain the range from the actual documentation<\/h2>\n
![\"Disconnected](\"https:\/\/cnbygele.com\/wp-content\/uploads\/2026\/10\/power-meter-energy-counter-rollover-inspection-20261024.png\")
Use modular subtraction only under explicit conditions<\/h2>\n
Distinguish wrap, reset and replacement<\/h2>\n
Build an event classification table<\/h2>\n
\n\n
\n \nObserved change<\/th>\n Evidence needed<\/th>\n Treatment before reporting consumption<\/th>\n<\/tr>\n<\/thead>\n \n High value followed by low value<\/td>\n Documented modulus, timestamps and operating bound<\/td>\n Apply one-wrap arithmetic only when the assumptions hold<\/td>\n<\/tr>\n \n Value becomes zero after maintenance<\/td>\n Change log, reset command and meter identity<\/td>\n Segment the record at the confirmed reset<\/td>\n<\/tr>\n \n Serial number or data-source identity changes<\/td>\n Replacement and mapping records<\/td>\n Start a new counter segment with a documented opening value<\/td>\n<\/tr>\n \n One impossible sample between normal readings<\/td>\n Raw response and communication status<\/td>\n Quarantine the suspect sample and investigate decoding<\/td>\n<\/tr>\n \n Negative net-energy change<\/td>\n Signed register definition and power direction<\/td>\n Interpret the documented net quantity rather than assuming a wrap<\/td>\n<\/tr>\n \n Long gap followed by a lower value<\/td>\n Maximum plausible energy and counter range<\/td>\n Retain an unresolved interval when wrap count is ambiguous<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n Choose a reading interval that limits ambiguity<\/h2>\n
Reconcile with an independent record when available<\/h2>\n
![\"Disconnected](\"https:\/\/cnbygele.com\/wp-content\/uploads\/2026\/10\/power-meter-energy-counter-rollover-records-20261024.png\")