Front-End vs Back-End Isolation in MV Test Systems

In brief

An MV test system needs defined isolation, earthing and protection boundaries, but not every implementation uses or places transformers identically. In the two transformer-isolated architectures compared here, back-end isolation places a step-up transformer between the converter and the device under test, while front-end isolation places the transformer upstream of a direct-MV converter. Placement changes which magnetic effects are in the controlled output path, but converter bandwidth, filtering, cabling, earthing and protection also determine what a test can prove.

Key takeaways

Every MV bench needs documented isolation, earthing and protection boundaries; whether a dedicated transformer is required, and where it sits, depends on the released topology and facility design.
An output-side transformer is part of the test circuit, so its measured frequency response, saturation behaviour, vector group, impedance and earthing connection must be included in the acceptance method.
Front-end isolation with direct-MV output can remove output-transformer effects from the DUT path, but it does not remove converter, filter, cable, protection or measurement limits.
Direct-MV architectures avoid recovery associated with an output transformer core; event repetition still depends on the converter's thermal, current, DC-link and protection envelope.
The back-end route remains legitimate for modest-fidelity programmes and where an existing LV source must be reused — provided its limitations are modelled and accepted, not ignored.

Why does an MV test system need isolation at all?

An MV test plan must define its isolation, earthing and protection boundaries before equipment is selected. A galvanic-isolation transformer is a common part of that design, but whether a dedicated transformer is required, how it is connected and where it sits depend on the facility feed, converter topology, DUT and applicable safety rules.

When transformer isolation is used, it serves several related objectives. A defined galvanic boundary can help separate the facility supply from the test circuit and give protection engineers a clear coordination zone. It can also provide a neutral and earthing boundary that is configured for the planned unbalance, leakage-current or phase-to-earth scenario. Finally, the transformer impedance and the surrounding switchgear and protection contribute to limiting and clearing fault energy. None of those functions replaces insulation coordination, interlocks, access control, emergency stops, discharge verification or a site fault study.

This article therefore compares two common transformer-isolated arrangements rather than claiming that every MV bench must use the same isolation method. Placement matters because a transformer has measurable frequency response, leakage impedance, saturation behaviour, vector-group phase displacement and winding capacitance. It is not the only determinant of test fidelity, but any transformer in the controlled output path must be included in the evidence.

What are the two architectures?

The simplest distinction is the order of the transformer and controlled converter relative to the DUT.

Architecture	Power-path order	Consequence for the programmed test
Back-end isolation	Facility supply → low-voltage converter → output step-up transformer → DUT	The transformer is inside the controlled test path, so its measured response, impedance, vector group and earthing connection contribute to the DUT-terminal conditions.
Front-end isolation	Facility supply → input isolation transformer → direct-MV converter → DUT	The transformer is upstream of the controlled output; the DUT path is instead defined by the converter, output filter, protection, cabling and earthing design.

Single-line comparison of back-end isolation, where a low-voltage converter feeds an output step-up transformer before the DUT, and front-end isolation, where an input isolation transformer feeds a direct-medium-voltage converter before the DUT

Figure 1. Simplified architecture comparison. Actual systems also include switchgear, filters, protection, earthing, measurement and auxiliary supplies; verify the released single-line diagram for the proposed configuration.

This ordering distinction should be drawn on the supplier’s single-line diagram and carried into the acceptance plan. Packaging does not change the electrical boundary: a transformer hidden inside a cabinet still participates in the waveform seen by the DUT.

The back-end (output-side) architecture is the historical route to medium voltage: take a proven low-voltage programmable source, add a step-up transformer, and the bench reaches kilovolts. Its logic is reuse and capital efficiency — the LV source may already exist, and transformers are well-understood machines.

In the front-end arrangement defined here, the isolation transformer is upstream of conversion stages that synthesise medium voltage directly. Cascaded power-cell topologies are one implementation, but the exact converter and isolation design must be confirmed rather than inferred from a category name. The controlled boundary is now the converter’s output network rather than an output transformer winding. That removes one magnetic stage from the DUT path; it does not make the commanded waveform identical to the DUT-terminal waveform without measurement. The broader system boundary is covered in How to Test a Medium-Voltage SST at Full Power.

What does an output transformer do to your test?

Each mechanism maps to a different acceptance check. The size of the effect is design- and load-dependent, so avoid treating every output transformer as either ideal or unusable.

Frequency response and phase: IEC 60076-18 and IEEE C57.149 define transformer frequency-response measurement practices. A commanded fast transition or harmonic component can therefore arrive at the DUT with different magnitude and phase after the transformer, cabling and load. Measure the complete source chain at the required operating point instead of assigning a universal bandwidth penalty.
Flux offset and saturation: asymmetric volt-seconds, DC content, point-on-wave timing and sag recovery can offset core flux. The cited IPST study documents saturation and magnetising-current distortion after voltage-sag recovery. Some profiles may consequently need different recovery timing or transformer design, but a mandatory delay cannot be inferred without testing the proposed chain.
Harmonic transfer: each harmonic and inter-harmonic encounters the combined magnitude and phase response of the transformer and test circuit. Commanded and delivered spectra should be compared at the DUT terminals under representative load.
Vector group, earthing and common mode: winding connection, neutral availability, capacitance and the selected earthing point become part of the test boundary. Alternative secondary arrangements may be possible, while output filters, cable capacitance, insulation monitoring and protection can be just as important as transformer placement.
Fault and zero-voltage response: leakage impedance, protection settings and core transients affect the voltage and current seen during faults and recovery from deep dips. ZVRT capability should be demonstrated at the DUT terminals with the intended earthing and load, not accepted from a front-panel command alone.

None of this makes the back-end route unusable. It means the LV source, transformer, filter, cabling, protection and instrumentation must be treated as one test source. Characterise ratio and phase versus frequency, transient response, zero-sequence path, saturation margin and delivered event tolerances; then state which limitations are corrected, bounded or accepted.

What does front-end isolation enable?

Moving transformer isolation upstream of a direct-MV converter removes the output transformer from the DUT path. That can reduce one set of magnetic effects, but the benefit remains conditional on the converter, filter, cabling, protection, earthing and measurement design.

Sequential ride-through testing: there is no output transformer core whose flux must recover between events. Repetition rate can still be limited by converter current, DC-link energy, thermal duty, protection logic or the DUT, so ask for demonstrated sequence limits.
Fast transitions: the terminal edge is governed by the direct-MV converter and output network rather than an added step-up stage. Verify rise/fall time under the specified voltage, load, cable and measurement bandwidth.
Earthing flexibility: placing isolation upstream may let the output earthing arrangement be engineered around the test. A line-to-PE or leakage-current mode is available only when the released converter, filter, insulation-monitoring and protection design supports it.
Spectral synthesis: direct output avoids transferring a programmed spectrum through an output transformer. Achievable harmonic order, amplitude, phase, load dependence and verification bandwidth still need stated limits.

The published XGY MVGS Series uses a cascaded power-cell topology for direct 5 kV to 35 kV output without an external step-up transformer in the output path. Its public series specifications include less than 1 ms rise/fall time, harmonic generation to the 50th order, independent three-phase control and four-quadrant regeneration. Those are series-level capabilities, not proof of a particular project’s input isolation, earthing mode, fault envelope or event repetition rate; the quoted configuration and FAT method must state those details.

For bidirectional work, a four-quadrant architecture can return absorbed energy through its grid-side power stage to the facility mains. Facility power factor, current distortion, protection coordination and import/export limits remain separate acceptance items.

When does the back-end route still make sense?

Three scenarios keep the output-transformer architecture on the table.

Modest dynamic-fidelity requirements: steady-state withstand, efficiency mapping and slow voltage changes may not exercise the transformer’s high-frequency or transient limits.
Existing assets: a lab with a suitable LV source may extend it to medium voltage with lower initial equipment cost. Include the transformer, switchgear, protection, floor area, losses, commissioning and terminal verification before calling it the lower-cost system.
Interim or bounded capability: a characterised transformer chain can keep a defined programme moving while a direct-MV platform is evaluated or reserved for higher-fidelity campaigns.

The same principle applies across the five common methods for testing MW-scale power electronics: select the least complex method that meets the evidence requirement at the DUT boundary. A back-end bench with documented limits is a legitimate instrument. The test plan should state which events it delivers within tolerance, which it approximates and which are outside its validated envelope.

What should you ask a vendor?

Use these six questions alongside the broader grid-simulator selection checklist:

Where does each isolation transformer sit, and what components are between the controlled converter and DUT terminals?
What rise/fall time and waveform tolerance are measured at the DUT boundary under the proposed voltage, load and cable conditions?
Can the required ride-through events run consecutively, and what electrical, thermal, protection or recovery limit sets the repetition rate?
Which output earthing arrangements are released, and which phase-to-earth or leakage-current scenarios are supported by the complete protection design?
How are harmonic and inter-harmonic magnitude and phase generated and verified at the DUT terminals?
In regenerative operation, through what path does absorbed energy return to the facility, and what grid-side limits apply?

A credible supplier of either architecture should answer all six with a single-line diagram, configuration-specific data and an acceptance method. The decision is not which label sounds better; it is which complete system reproduces the required terminal conditions with evidence.

Product fit

Where the MVGS fits

The published XGY MVGS Series uses a cascaded power-cell topology to provide configurable 5 kV to 35 kV direct medium-voltage output without an external step-up transformer in the output path. Published series capabilities include 1 MVA to 10 MVA standard power, expansion to 20 MVA and above, less than 1 ms voltage rise/fall time, harmonic generation to the 50th order, independent three-phase control, and four-quadrant regeneration. Input isolation, earthing, fault modes, event repetition and acceptance limits remain configuration-specific and are confirmed against the project test plan at quote stage.

Power Supplies

MVGS Series Medium-Voltage Grid Simulator

XGY MVGS is a four-quadrant medium-voltage grid simulator and programmable AC source/load platform for megawatt-scale power electronics testing. Its modular Power Cell Cascaded topology delivers direct medium-voltage output without external step-up transformers.

View product

FAQ

Frequently asked questions

Does front-end isolation mean the system has no transformer?

No. In the architecture defined here, the transformer sits on the facility side of the converter rather than between the converter and DUT. Suppliers do not always use the term consistently, so confirm the released single-line diagram, winding arrangement, earthing points and protection boundary.

Can a step-up transformer bench run LVRT profiles?

Yes, when the combined source, transformer, cabling and load reproduce the required profile at the DUT terminals within the stated tolerance. Transformer frequency response and flux transients can alter some profiles or constrain repetition, but the size of that effect and any recovery interval must be measured for the actual system rather than assumed.

Why does transformer placement affect ground-fault testing?

An output transformer's vector group, neutral availability, winding capacitance and earthing connection become part of the DUT boundary. Moving isolation upstream may provide more design freedom, but transformer placement alone does not guarantee a ground-fault mode: converter topology, output filters, insulation monitoring and protection coordination must also support it.

Which architecture is cheaper?

A lab that already owns a suitable LV source may find the back-end route has a lower initial equipment cost. Compare total installed cost, however, including the transformer, switchgear, floor area, protection, losses, commissioning and terminal-waveform verification. Direct MV can be the better-value route when removing the output transformer materially improves required tests.

Primary sources and standards

Official standards, regulators, metrology bodies, and first-party technical references used to verify specific claims. The notes identify the evidence boundary; a citation does not imply endorsement of XGY Tek.

IEC 60076-1:2011 — Power transformers — Part 1: General — IEC
Provides the general transformer framework, including rated quantities, harmonic content and DC-current considerations relevant to qualifying an output transformer.
IEEE Std 1547-2018 — Interconnection and interoperability of distributed energy resources — IEEE Standards Association
Defines DER response-to-abnormal-condition, power-quality and interconnection requirements that motivate repeatable voltage and frequency event testing.
IEEE Std 1547.1-2020 — Conformance test procedures for DER interconnection equipment — IEEE Standards Association
Provides the associated conformance-test context for verifying equipment functions against IEEE Std 1547.
Medium Voltage Power Electronics Laboratory — Oak Ridge National Laboratory
Reference practice for MV converter test facilities.
XGY MVGS Series medium-voltage grid simulator — published specifications — XGY Tek
Supports the public direct-MV topology, voltage and power ranges, rise/fall time, harmonic generation, independent three-phase control and four-quadrant regeneration statements used here.
IEC 60076-8:1997 — Power transformers application guide — IEC
Supports the application-level discussion of transformer connections, voltage drop, converter loading and measurement when qualifying a back-end step-up chain.
IEC 60076-18:2012 — Power transformers — Part 18: Measurement of frequency response — IEC
Establishes that transformer frequency response is a measurable equipment characteristic and provides the relevant measurement framework.
IEEE Std C57.149-2024 — Frequency-response analysis for oil-immersed transformers — IEEE Standards Association
Provides instrumentation, test and interpretation guidance for transformer frequency-response analysis, supporting system-specific rather than idealised bandwidth treatment.
Transformer Saturation after a Voltage Sag Recovery — International Conference on Power Systems Transients
Analyses transformer flux offset, saturation and magnetising-current distortion following voltage-sag recovery.