Before IEC 60034-27-3, dissipation factor limits for stator insulation were mostly private — each manufacturer and utility worked to its own internal numbers. The standard put a public set on the table. That was useful, and it was immediately controversial, because a single set of numbers has to cover insulation systems that behave quite differently.
So the limits are worth knowing exactly — the values, the narrow conditions they hold under, and where they stop applying. Used carefully they accept or reject a new coil. Used carelessly they reject good insulation or pass bad. This is the acceptance-criteria detail behind the dissipation factor pillar.
Table of Contents
The three limits
For new single bars and coils, tested guarded at room temperature, up to a rated voltage of 21 kV:
| Characteristic value | Maximum |
|---|---|
| Initial tan δ at 0.2 UN (tan δ0.2) | 20 × 10⁻³ |
| Delta tan delta per 0.2 UN step (Δtan δ / 0.2 UN) | 5 × 10⁻³ |
| Tip-up, tan δ0.6 − tan δ0.2 | 5 × 10⁻³ |
The 20 × 10⁻³ baseline is the same as 2% in absolute terms. The two 5 × 10⁻³ figures cap how fast the insulation is allowed to rise with voltage — one as the worst single step, one as the 0.6-to-0.2 UN tip-up. Together they say: start low, and don’t climb fast.
These also hold for global-VPI test objects impregnated in slot simulations. For what each value means diagnostically, see the tip-up page — this page is about applying them.
The conditions are not optional
Every word in “new single bars and coils, guarded, room temperature, up to 21 kV” is load-bearing. Change any of it and the limits may not apply.
- New. These are manufacturing-acceptance limits. An aged coil pulled from service is read by trend and by curve shape, not against this table.
- Single bars and coils. Not complete windings — see below.
- Guarded. The values assume guard ring electrodes are in place to strip out end-of-slot-coating surface currents. An unguarded reading is not comparable.
- Room temperature. The tan δ-versus-voltage curve shifts with temperature, so the limits are tied to ambient measurement.
- Up to 21 kV. Above 21 kV rated voltage, limits are set by special agreement between manufacturer and client, not from this table.
- Standard voltage steps. The tip-up and Δtan δ are defined at specific points — 0.6 and 0.2 UN, in 0.2 UN steps. Use different steps and the published limits no longer apply.
The accuracy trap
This is the part most people miss, and it can flip a result.
The limits are not compared against your raw reading. They are compared against the reading plus its measurement uncertainty. The standard says to compute the accuracy of each characteristic value from the instrument specifications, which gives a spread around each computed value — and the maximum of that spread must fall below the limit, not the nominal number.
So a coil reading a tip-up of 4.8 × 10⁻³ on paper does not automatically pass. If your instrument’s uncertainty puts the upper end of the spread above 5 × 10⁻³, it fails the criterion. On values this small — single-digit thousandths — instrument accuracy is not a rounding detail. It decides borderline coils. Know your measuring system’s uncertainty before you quote a pass.
Where the limits don’t reach
Complete windings have no limits at all. The standard provides none, because a complete-winding reading is strongly affected by the end-winding stress control coating, which no guard ring can remove. A new complete winding still gets measured — to control manufacturing quality and to set a baseline — but there is no table to pass or fail it against. Aged complete windings are the same: ambient and surface conditions of the stress-control coating rule out absolute limits, so condition is judged by trending the value over time. A consistent upward trend is the trigger for visual inspection and further diagnostics, not a fixed threshold. For aged windings, a reduced test voltage down to 0.6 rated voltage can be agreed between the parties.
Above 21 kV, by agreement. As noted, higher-rated machines fall outside the table and need negotiated criteria.
Why the limits are debated
These are among the most contested numbers in machine-insulation practice, and it is worth knowing why before you treat them as a hard gate.
The core issue is that one set of limits has to span many insulation systems that legitimately behave differently. Field studies across large populations of coils and bars — and long-term aging work correlating dissipation factor and tip-up against service life using Weibull statistics — have repeatedly found that insulation systems exceeding the Table 1 values can still be entirely reliable in service. Work within CIGRE’s rotating-machines committee has pushed to reassess whether the published limits are appropriate. The limits are viewed by some in the field as too restrictive and by others as too lenient — which is roughly what you would expect from a first attempt to make private criteria public.
None of this means ignore them. It means treat a Table 1 exceedance as a flag to investigate, not an automatic condemnation, especially on an insulation system with a proven service history.
Building a real acceptance criterion
In practice, the strongest acceptance decision layers three things:
- The Table 1 limits as a baseline screen, applied with the accuracy spread and only under the conditions they’re valid for.
- Manufacturer-specific criteria for the particular insulation system, which often carry more weight than the generic table because they are tuned to how that system actually behaves.
- Trend data on your own coils, bars, and windings — the single most reliable signal, because a coil’s change from its own baseline is more meaningful than its position against a population limit.
For a new production run, that means testing a statistically meaningful sample — enough coils to show the process is under control — rather than one or two. For an in-service machine, it means comparing against that machine’s own history, not against a table built for new single bars.
FAQ
What are the IEC 60034-27-3 limits?
For new single bars and coils, tested guarded at room temperature up to 21 kV: a baseline tan δ at 0.2 UN of 20 × 10⁻³, a delta tan delta per 0.2 UN step of 5 × 10⁻³, and a tip-up of 5 × 10⁻³.
Do the limits apply to complete windings?
No. The standard provides no limits for complete windings, because the end-winding stress-control coating strongly affects the reading and cannot be guarded out. Complete windings are judged by trending over time, not against a fixed threshold.
Do the limits apply above 21 kV?
No. For rated voltages above 21 kV, the acceptance values are set by special agreement between manufacturer and client, not from the standard’s table.
Why do I have to include instrument accuracy?
Because the standard compares the limit against the computed value plus its measurement spread, not the nominal reading. The maximum of that spread must fall below the limit. On values this small, instrument uncertainty decides borderline coils.
Are the IEC 60034-27-3 limits reliable?
They are debated. Field and long-term aging studies have found insulation systems that exceed the limits yet remain reliable in service, and industry groups have pushed to reassess them. Treat an exceedance as a reason to investigate, and lean on manufacturer criteria and trend data alongside the table.
Can the limits be used with different voltage steps?
No. The tip-up and delta tan delta are defined at specific voltages — 0.6 and 0.2 UN, in 0.2 UN steps. If you measure with different steps, the published limits no longer apply.
