EN
When a Pump Stops Without a Mechanical Reason
A hydraulic power pack sitting in a chemical plant in the southern United States kept tripping its control relay. Pressure readings drifted, the directional valve would occasionally stay energized after the command was removed, and the PLC logged communication faults that no one could reproduce on a bench test. The pump itself checked out. The motor windings were fine. The relief valve was clean. What finally explained the behavior was not inside the hydraulic circuit at all. A newly installed variable frequency drive, mounted in the same enclosure as the pressure transducer wiring, was injecting high-frequency noise into the 4–20 mA signal loop.
That scene repeats across industries more often than most maintenance teams expect. Electric hydraulic pumps are electromechanical assemblies, and the electrical half can undermine the hydraulic half long before any seal leaks or piston wears. Minimizing electromagnetic interference is not a compliance checkbox. It is a reliability practice.
Where the Noise Actually Comes From
A modern electric hydraulic pump rarely runs directly across the line. Speed control, soft starting, and pressure regulation increasingly rely on PWM drives and switch-mode power supplies. Those devices switch at kilohertz frequencies with fast voltage edges, and that is exactly what generates broadband emissions.
Two coupling paths matter most. The first is conducted emission along motor power cables: PWM energy travels down the cable, and without a low-impedance return path, it finds its way into sensor commons, communication shields, and control grounds. The second is capacitive coupling, where long motor leads behave like a capacitor and inject current into adjacent wiring. Neither path requires a catastrophic fault. A few volts of common-mode noise riding on a pressure feedback line is enough to push a PLC analog input out of range.
Inductive loads add another layer. Solenoid valve coils, contactor coils, and relay coils release stored energy when de-energized, producing voltage spikes that can travel back through shared DC commons. A pump with several solenoid valves cycling under pressure control will generate repetitive transients even when the VFD is operating normally.
Shielding and Bonding Decide Whether Mitigation Works
Cable selection gets attention, but termination is where many installations fail. A braided shield with 85% coverage around motor power leads is a common baseline for managing PWM noise. The shield only does its job when it is bonded to the drive enclosure and, ideally, to the motor frame with a low-impedance connection. A pigtail wire pulled out of the shield and landed on a terminal block raises impedance at high frequency and defeats much of the shielding effect.
Grounding configuration matters just as much. A star topology, with each device bonded back to a single earth reference using short, wide conductors, avoids the ground loops that daisy-chained grounds create. In practice, this means removing paint from the back panel behind drives and pumps, using braided straps or busbars rather than relying on mounting screws, and keeping bonding runs short. For remote pump skids, a local earth reference bonded to the main enclosure usually performs better than a long, thin ground wire run across the machine.
Signal wiring deserves separate treatment. Analog pressure and position feedback should use shielded twisted pairs, and the shield should be terminated according to the signal type. Power and signal cables should not share a conduit or a cable tray without physical separation. A 30 cm separation is a working minimum for low-voltage signal lines running near motor leads; more distance is better where space allows.
Matching the Filter to the Failure Mode
Not every filter solves every problem. A common mistake is adding an AC line filter on the drive input and expecting pressure signal noise to disappear. Input filters reduce conducted emissions back toward the supply, but they do not address PWM energy traveling from drive to motor. Output reactors and output filters act on the motor side, and their effects differ enough that the choice should follow the symptom.
| Configuration | What it addresses | Typical effect on PWM content | Best fit for pump systems |
|---|---|---|---|
| No output filter | Nothing | Full PWM edge reaches motor leads | Only very short motor cables in low-noise environments |
| 5% output reactor | dV/dt stress and cable reflection | Reduces peak voltage, PWM carrier remains | Moderate cable runs, basic motor protection |
| Sine wave output filter | PWM carrier and high-frequency content | Converts PWM to near-sinusoidal output, THD typically below 3% | Sensitive electronics sharing an enclosure, long motor leads |
The table is not a ranking. A short-coupled pump with no sensitive instrumentation nearby may run reliably with a reactor alone. A skid-mounted pump sharing an enclosure with a PLC, a pressure transmitter, and a communication gateway is a different risk profile. Sine wave filters cost more and take up space, but they change the output waveform itself rather than just softening its edges. That distinction matters when the failure mode is corrupted sensor data rather than motor insulation stress.
Protecting the Low-Voltage Side Without Overengineering
The control side of an electric hydraulic pump is where EMI becomes visible. Pressure transducers, position sensors, and PLC analog inputs operate at low signal levels, and a few hundred millivolts of noise is significant. Isolation is often more effective than filtration at this stage. Isolated transmitters on analog inputs, optocouplers on digital signals, and a dedicated 24 V supply for control electronics separate from the power circuit reduce the chance that motor-side transients reach the control common.
There is a limit to what isolation can fix. If the noise is radiating from an improperly shielded motor cable and coupling directly into a sensor housing, no amount of signal isolation will clean up the reading. Field experience with hydraulic control circuits in high-EMI environments shows that corrupted low-voltage solenoid signals account for a substantial share of hydraulic failures, which is why the physical layer deserves attention before adding more electronic correction.
A skid in a steel mill or near a substation may need the full treatment: shielded VFD cable, sine wave filtering, star grounding, isolated I/O, and physical separation between power and signal routing. A compact pump in a clean workshop may need little more than a properly bonded shield and a short motor cable. Treating every installation as identical either wastes money or leaves the difficult cases underprotected.
The Manufacturing Side of Interference Control
Mitigation does not end at the panel drawing. Consistency in assembly, bonding, and cable routing determines whether the first unit and the five-hundredth unit behave the same way. A pump built with a documented grounding scheme, controlled shield termination, and verified component placement reduces the variables that field technicians otherwise have to chase.
Ruiqi has been manufacturing hydraulic tools and electric pump systems since 1997, with production and quality processes built around repeatable assembly rather than one-off builds. That kind of manufacturing discipline matters for EMI because interference problems often come from small deviations: a missing bonding strap, a shield landed on the wrong terminal, a signal cable routed too close to a motor lead. Supply chain and production consistency do not make a pump immune to noise, but they make the electrical behavior predictable enough that mitigation measures designed for the platform actually hold up in the field.
Reliable operation under electromagnetic stress is not achieved by a single component. It comes from treating the pump as an electrical system as much as a hydraulic one, and from making the quiet path the default path.
