P28 Stages 1, 2 and 3 explained

A flowchart, not a menu

The three stages of P28 are often described as though they were increasing levels of rigour you could opt into. They are not. They are a decision tree, and where you land is determined by your equipment and your connection point, not by how much effort you want to spend.

The practical consequence is worth stating plainly, because it is where most wasted money goes: people commission Stage 2 studies for equipment that qualified under Stage 1 all along, and people submit Stage 1 declarations for connections that were never eligible. Both are avoidable by reading the gate conditions properly.

Two rules that remove the choice

Before any threshold is checked, two conditions can take Stage 1 off the table entirely. Both are easy to miss because they are not about the equipment's rating.

1. If the PCC is at HV, Stage 1 does not apply

This holds even when the equipment itself is LV. The relevant note sits with the assessment flowchart (Figure 9) and Clause 6.3.3.1 confirms it: all HV connections are assessed under Stage 2. A 400 V motor fed from your own transformer, where your point of common coupling with other customers is on the 11 kV side, is a Stage 2 problem no matter how small the motor is.

This one catches out industrial sites constantly, because the equipment looks like textbook Stage 1 equipment. The question is never "what voltage is my kit?" — it is "where is the point I share with other customers?".

2. Equipment that works as a system is assessed as a system

Clause 6.3.1 requires disturbing equipment that operates together to be treated as a single item. The two cases that come up in practice:

• Micro-inverters forming one PV installation. Twenty inverters that each individually sit under the threshold do not give you twenty Stage 1 assessments.
• Heat pumps with separate indoor and outdoor units. Clause 8.5 is explicit that these must be tested as one integrated system as well as individually.

Stage 1: the simplified route

Stage 1 exists because it would be absurd to require a flicker study for a washing machine. Its logic is that if a manufacturer has already demonstrated conformance to the relevant EMC product standard, and the supply is at least as strong as the one that standard assumes, then the equipment's flicker behaviour is already known to be acceptable.

The two product standards that do the work:

• BS EN 61000-3-3 — equipment up to 16 A per phase, tested against a reference impedance, connecting unconditionally.
• BS EN 61000-3-11 — equipment up to 75 A per phase, connecting conditionally: the manufacturer declares the maximum supply impedance (Z_max) the equipment needs, and it is your job to show the actual supply meets it.

The four Stage 1 routes

Clause 6.3.2 sets out how equipment qualifies. There are effectively four outcomes, and it is worth identifying which one you are arguing before you gather evidence.

Route A — up to 16 A per phase, conforming to BS EN 61000-3-3

The clean case (Clause 6.3.2.1). Not subject to conditional connection. Valid provided the supply impedance at the customer's terminals does not exceed:

In practice this is rarely the binding constraint. The standard notes that PME (TN-C-S) LV systems typically present a supply impedance of 0.35 Ω or less, comfortably inside the single-phase figure. Services installed before about 1950 are the ones worth measuring.

Route B — up to 75 A per phase, conforming to BS EN 61000-3-3

Clause 6.3.2.2 allows connection without reference to the network operator provided the service current capacity is confirmed adequate — a check BS 7671 Regulation 132-16 requires of you anyway.

Route C — 16 to 75 A, conforming to BS EN 61000-3-11 only

This is conditional connection, and the route with the most detail. The equipment qualifies under Stage 1 only if either:

• (a) the supply impedance at the customer's terminals — confirmed by measurement, or by calculated values the network operator provides — is no greater than the Z_max the manufacturer declares in the instruction manual; or
• (b) both of the following:
  • the service current capacity is confirmed as at least 100 A per phase, as required by the manufacturer, and the equipment is clearly marked to that effect; and
  • the supply impedance is confirmed by measurement as no greater than the values below.

Route D — conforming to neither product standard

Stage 1 is unavailable. The equipment may be assessed under Stage 2, though the standard itself notes that equipment failing both product standards is unlikely to meet the Stage 2 limit either. Treat this as a design problem rather than an assessment problem.

What passing Stage 1 does and does not excuse

This is the subtlest part of P28 and the part most often got wrong, including — until we corrected it — by our own calculator.

The temptation is to treat the step voltage check, the RVC category check, the flicker check and the Stage 1 check as four independent tests that all have to pass. They are not. Where equipment qualifies under the EMC product-standard routes, that qualification is the demonstration of acceptable voltage-fluctuation performance. Running a separate step voltage calculation and failing the equipment on it contradicts Clause 8.6.2, which says in terms that such equipment may be connected.

The Annex A motor route is narrower, and the difference matters:

Annex A tells you a motor of that size may be started without reference to the network operator; it does not repeal the 3% step voltage limit of Clauses 5.4 and 6.2. A motor inside the Annex A table on a supply weak enough to produce more than a 3% step is still a problem, and the network operator's discretion under Clause 6.2 is the route to resolving it — not a claim that Annex A made the limit go away.

Motors: the Annex A shortcut

Annex A gives ratings for LV motors that may be connected without reference to the network operator, which for a large share of real projects is the fastest answer available. Which table applies depends on how often the motor starts and where the PCC is.

Started very frequently — intervals under one minute (Table A.1.1)

PCC not at the LV busbar of a substation (Table A.1.2)

PCC at the LV busbar of a substation (Table A.1.3)

The most generous case, and the one that decides most industrial motor questions. It applies to three-phase motors started at intervals of ten minutes or more, and scales with the distribution transformer:

Star-delta starting multiplies these ratings by 1.5 (Clause A.2). A 500 kVA substation supports a 45 kW motor direct-on-line, or a 67.5 kW motor star-delta. That factor is often the cheapest way to turn a failing assessment into a passing one.

Stage 2: calculating P_st

Stage 2 is where you stop citing product standards and start calculating. The requirement (Clause 6.3.3.2) is to establish the worst-case short-term flicker severity your equipment produces at the PCC, under worst-case supply impedance, and show that P_st ≤ 0.5.

The 0.5 figure is deliberately half the LV planning level. The headroom is what allows the network operator to accept your connection without measuring what is already on the network — if every connection stays under half, the summation of several of them still fits.

For repetitive events, the assessment usually runs through the P_st = 0.5 characteristic of Figure B.1.2, which maps how often a voltage change happens to how large it is permitted to be. The relationship is steep: a change every few seconds is held to a small fraction of a percent, while one every eight minutes or more is allowed the full 3%.

Stage 3: measurement and summation

If Stage 2 does not pass, the remaining route within the standard is Stage 3, undertaken by agreement with the network operator. It differs in kind, not just in degree: instead of assuming a worst case, it measures the actual flicker background already present at the connection point, combines it with your predicted contribution using the summation rules, and compares the total against the planning level.

Stage 3 can succeed where Stage 2 fails, because Stage 2's assumption that the network is already half-loaded with flicker is frequently pessimistic. On a quiet rural network there may be far more headroom than Stage 2 credits you with.

It is also slow and expensive. It needs a flickermeter installed at the connection point for a representative period, network operator involvement throughout, and a specialist to do the summation properly. Budget weeks, not days.

Two situations push you towards Stage 3 regardless of what a simplified calculation says: several disturbing installations sharing one PCC, and RVC events frequent enough that summation between them matters. Both are outside what any single-equipment calculator, including ours, can settle.

What to do if you do not pass

A failed simplified assessment is not a refusal. It means the cheap route is closed, and there are usually four ways forward — roughly in order of what they cost.

1. Change the starting method. Star-delta buys a factor of 1.5 under Annex A. A soft starter or a variable-speed drive effectively removes the step change altogether. Watch the consequence: a drive is a harmonic source, so you may trade a P28 problem for a G5 one.
2. Change the connection point. Step voltage change is inversely proportional to short-circuit level, so connecting nearer the substation — or at the LV busbar rather than down a service — can move you into a different Annex A table entirely.
3. Change the operating pattern. The permitted change grows sharply as events get less frequent. Staggering starts, or interlocking equipment so two items cannot start together, can be a control change rather than a plant change.
4. Go to Stage 3. When the design is fixed and the numbers are marginal, measured background levels may show the headroom is there.