AC/DC rectifier monitoring in UPS and industrial power

The problem

AC/DC rectifiers feed critical buses in data centers, 24/7 industries, DC UPS and telecom power systems. In N+1 arrangements, reliability depends on the ability to identify which module is degrading before it fully fails — and before a second module failure brings down the parallel. The issue is that many rectifiers expose only binary alarms (ok/fault) and proprietary protocols, which complicates a consolidated view on the NMS. Worse, excessive AC ripple over the DC bus degrades batteries and sensitive loads silently, without triggering any alarm from the internal controller. Operations need an independent supervision layer that measures voltage, current, efficiency and ripple per module, with temporal granularity sufficient to detect early drift.

Golden rule: binary alarm from the controller does not reveal progressive degradation.

Typical topology

A typical industrial rectifier set has 2 to 8 modules in parallel, each with common output on a DC bus. In 48V telecom, modules are typically 50 A; in 380 VDC data center, 30 to 50 A.

Three-phase AC
   |
   v
+--------+   +--------+   +--------+
| Module |   | Module |   | Module |
| AC/DC  |   | AC/DC  |   | AC/DC  |
| #1     |   | #2     |   | #3 (N+1)|
+---+----+   +---+----+   +---+----+
    |            |            |
    +------------+------------+
                 |
                 v
        Common DC bus
                 |
    +------------+-------------+
    |  shunts +  dividers      |
    +--+-------+-------+-------+
       |       |       |
       v       v       v
   +---------------------------+
   | AEM-60DC8 (1 per module)  |
   +-----------+---------------+
               | Modbus RTU
               v
         SCADA Gateway

Each AEM-60DC8 reads 2 channels per module (current and voltage at output), plus aggregate channels for the bus. Typical distance from the DC bus to the gateway is below 20 m.

Golden rule: one AEM-60DC8 per module yields per-unit visibility and safe hot-swap.

What to measure and why

In rectifiers, eight quantities need continuous monitoring:

Quantity	Justification
DC output voltage per module (V)	Detects regulation drift; parallel modules should stay within ±0.5% of each other.
DC output current per module (A)	Identifies load imbalance (load sharing); >15% difference indicates a problem.
Common DC bus voltage (V)	Confirms voltage delivered to loads; reference for all modules.
Total bus current (A)	Sum of loads; input for N+1 margin calculation.
AC ripple on DC (mV RMS and pp)	Above 100 mV pp on 48V indicates capacitor failure; on 380 VDC, limit is typically 1 V pp.
Module internal temperature (deg C)	Via 4–20 mA transducer from the rectifier or external thermocouple; >65 deg C drastically reduces life.
Computed efficiency (%)	DC output power / AC input power ratio; 2% long-term drop indicates wear.
Accumulated operating time (h)	Useful for hours-based maintenance; the AEM-60DC8 keeps the counter in a persistent register.

Efficiency depends on simultaneous AC power reading, which can come from an external Modbus meter chained on the same RS-485 bus — or be estimated from typical power factor.

Golden rule: measuring per-module current is what makes load sharing visible; without it, N+1 is theater.

Sizing — how many AEM-60DC8?

Formula:

N_AEM = ceil( (2 × N_modules + bus_channels) / 8 )

Where 2 channels per module are current and voltage; bus_channels sum total voltage, total current, ripple and ambient temperatures.

Simplified example for 4+1 telecom 48V arrangement (5 modules):

• 5 modules × 2 channels = 10 channels
• Bus: 1 voltage + 1 current + 1 ripple + 1 temperature = 4 channels
• Total: 14 channels → 2 AEM-60DC8 (with 2 free channels for expansion)

Simplified example for 380 VDC data center with 8 modules:

• 8 modules × 2 channels = 16 channels
• Bus: 4 channels
• Total: 20 channels → 3 AEM-60DC8

For safe hot-swap, one dedicated AEM-60DC8 per trio of modules is recommended, keeping fault isolation.

Golden rule: size for the sum of rectifier modules plus 2 reserve channels per unit.

Recommended setpoints and alarms

Variable	Warning	Critical	Automatic action
48V module output voltage (V)	<53.0 or >55.5	<51.5 or >57.0	Notify NMS; block new parallel
380V module output voltage (V)	<370 or >390	<360 or >400	Notify NMS; shut down module
Current imbalance between modules (%)	>15	>25	Flag faulty load share
48V AC ripple (mV pp)	>100	>200	Inspect rectifier
380V AC ripple (mV pp)	>800	>1500	Replace capacitor bank
Module internal temperature (deg C)	>55	>70	Reduce requested current
Efficiency (%)	<89	<85	Predictive maintenance
Total load / nominal capacity (%)	>85	>95	Evaluate adding a module

Limits occupy range 0x40–0x6F of the 147 holding registers, with persistent anti-rollback ensured by firmware v1.03.

Golden rule: a current imbalance alarm is the earliest sensor of N+1 failure.

SCADA / PLC integration

The AEM-60DC8 operates as a Modbus RTU Slave over RS-485, with baudrates 4800/9600/19200/38400/57600/115200 bps. In rectifier environments, 38400 bps is recommended to reduce polling latency in 8+ module arrangements.

Ignition: configure each AEM-60DC8 as a separate Device. Use a UDT (User Defined Type) named "RectifierModule" grouping voltage, current, ripple and flags. Polling of 1 s for current (load sharing), 5 s for ripple and 30 s for efficiency.

Elipse E3 / Elipse Power: leverage the XObject structure to represent each module. Alarm scripts can compare current between modules and generate "load share off-spec" automatically. Elipse Power has direct support for parallel bus topology.

AVEVA System Platform: use $UserDefined objects with voltage, current and ripple attributes. In parallel, configure ApplicationServer Galaxy with hierarchical Site → BusbarDC → Module model.

For PLCs (Siemens S7-1500, Allen-Bradley CompactLogix), an HMS Anybus or ProSoft Modbus RTU → Profinet/EtherNet-IP gateway enables reading. Keep the AEM-60DC8 isolated from the corporate network.

See whitepaper WP01 about Modbus RTU on Secure by Design.

Golden rule: group modules as objects in SCADA — analyzing loose variables in N+1 arrangements is wasted work.

Regulatory compliance

Industrial and telecom rectifiers follow distinct standards depending on application:

• IEC 62040-1 (UPS — safety) for systems with associated inverter
• IEC 62040-3 (UPS — methods of measurement and performance requirements)
• IEEE 519-2014 (Recommended Practices for Harmonic Control), applicable to AC rectifier distortion
• ETSI EN 300 132-2 (Power supply interface at the input to telecommunications equipment) for -48 VDC rectifiers
• ETSI EN 300 132-3-1 for 380 VDC rectifiers in data centers
• IEC 62443-4-2 SL2 as target/in progress for the cybersecurity layer

The AEM-60DC8 provides forensic telemetry required in IEC 62443 audits and in ISO 9001 traceability of critical events.

Golden rule: UPS standards are class-specific — confirm before setting limits.

Illustrative case

Simplified example: a colocation data center in southeast Brazil operated 6 rectifier modules of 380 VDC in N+1 arrangement. Original monitoring was via OEM controller with binary alarms. Over six months, the team noticed three "module fault" alarms that, after module replacement, returned in subsequent weeks — typical symptom of thermal stress on another module of the parallel.

Implementation: 2 AEM-60DC8 units on a dedicated Modbus RTU bus, reading current and voltage of each of the 6 modules. The NMS received UDT per module in Ignition.

Result: analysis of the first 60 days revealed that module #3 carried 23% more current than the average — consequence of divergent offset calibration after an OEM firmware update. The fix was recalibration, not replacement. In parallel, bus ripple was at 1.2 V pp during night periods, indicating a degrading capacitor on module #5, identified preventively. Projected MTBF gain was approximately 40% for the set.

Golden rule: measuring and comparing all modules reveals problems the internal controller does not see.

Deployment checklist

1. Survey quantity, model and topology (N+1, 2N) of rectifier modules.
2. Define measurement points (output per module, common bus).
3. Specify shunts and resistive dividers compatible with nominal range.
4. Compute N_AEM and provide 2 reserve channels per unit.
5. Place AEM-60DC8 close to rectifier panel, less than 20 m from the gateway.
6. 22 or 24 AWG shielded twisted-pair cable, shield grounded at one point.
7. Configure baudrate 38400 bps and unique Slave ID.
8. Load warning/critical setpoints in holding registers 0x40–0x6F.
9. Validate reading of 147 holding registers and enable forensic telemetry.
10. Document ripple and efficiency baseline at commissioning.

Golden rule: commissioning without a ripple baseline makes future diagnosis impossible.

FAQ

See the FAQ block in the frontmatter of this page.