On a metal core PCB, the circuit copper and the metal base are separated by one thin thermally conductive dielectric. In power supplies, LED drivers and motor controls, that layer often carries safety isolation, so drawings often say "hipot ××× V". A single voltage value is not something a fab can execute. AC or DC, how long, what leakage counts as failure, between which points, 100% or sampled: none of it is defined.
This article answers two questions. Which inputs complete a hipot requirement so it becomes a test condition, and which manufacturing steps decide whether the board passes.
One point first. Routine 100% electrical test (flying probe or fixture) checks opens and shorts. It is a different inspection from dielectric withstand. SKPCB electrically tests 100% of orders; hipot conditions have to be specified separately in the order.
What a hipot test measures
A hipot test applies a voltage above working voltage between circuit copper and the metal base, holds it, and checks for breakdown or leakage above a limit. It answers whether the insulation keeps isolation at the specified voltage. It does not measure thermal performance or continuity.
Current can take two paths:
| Path | Where | Main factors |
|---|---|---|
| Through the dielectric | Directly under copper | Dielectric thickness, voids, contamination, copper burrs |
| Surface tracking or air flashover | Board edges, hole edges, V-score | Copper-to-edge distance, routing burrs, exposed metal |
Many failures are not a weak dielectric. They come from metal at an edge or hole sitting too close to a trace.
Test conditions the drawing must state
| Input | What to state | If missing |
|---|---|---|
| Voltage type | AC or DC | Criteria differ; results not comparable |
| Test voltage | Value and source standard | Fab falls back on habit |
| Ramp and dwell | Ramp method, dwell seconds | Different time, different pass rate |
| Criterion | Leakage limit, or "no breakdown, no flashover" | No pass/fail decision |
| Test points | Which nets go to HV, base to return | Wrong points, no real test |
| Sampling | 100% or per-lot sample | Cost and lead time cannot be quoted |
| Preconditioning | Humidity or thermal shock before test | Passes at shipment, fails in field |
| Records | Test record shipped with goods | No traceability |
Test voltage and leakage limit should come from the safety standard of the end product and be supplied by the designer. The fab executes and records the condition; it does not choose the safety margin.
Test points: which nets are tied together
A common setup ties all circuit nets to the HV terminal and the metal base to return, testing copper-to-base insulation for the whole board in one shot. Any net meant to be at base potential (for example a ground bonded to the aluminum through a plated hole) must be excluded, or the board reads as a short the moment voltage is applied.
So the drawing needs a net grouping note: which nets take part in hipot and which connect to the base. A missing note is the most common reason a hipot clause comes back as an engineering query (EQ).
Manufacturing factors that decide pass or fail
Dielectric thickness and uniformity. A thinner dielectric conducts heat better and leaves less voltage margin. That is the central metal core trade-off, covered in aluminum PCB insulation layer thermal trade-off. Local thin spots, voids and particles after lamination all become breakdown starting points.
Copper burrs and etch residue. Sharp trace corners and copper slivers concentrate the field and break down first at the same voltage.
Exposed metal at holes. A hole drilled straight through the metal base exposes aluminum or copper on the hole wall, with no isolation between the plating and the base. Holes that must be isolated are usually drilled oversize in the base, filled with resin, then redrilled in the resin. The drawing must mark which holes need this.
Edge burrs and V-score. Punching, routing and V-scoring leave metal burrs at the edge or smear base metal onto the dielectric surface, shortening the creepage path. Aluminum machining limits are in aluminum PCB fabrication and machining limits.
Copper-to-edge distance. Traces close to the edge flash over along the edge more easily. The higher the hipot requirement, the more pull-back must be reserved at design stage.
Base material. Aluminum and copper bases machine differently, so edge and hole treatment differ too. See aluminum vs copper base manufacturing.
Checklist before release
- Voltage type, test voltage and source standard
- Ramp method, dwell time, leakage limit
- Nets under test, nets bonded to the base
- 100% or sampled, and sample size
- Preconditioning such as humidity exposure
- Holes that need isolation
- Minimum copper-to-edge and copper-to-hole distance
- Whether test records ship with the goods
SKPCB aluminum PCBs go up to 1500 mm long. Longer boards mean more edge and V-score length, so edge burr control should be stated on the drawing. See metal core PCB capabilities, quality assurance, and request a quote once the package is complete.
FAQ
Q: Can 100% electrical test replace a hipot test?
A: No. Electrical test checks opens and shorts. Hipot checks insulation between circuit copper and the metal base. Specify them separately.
Q: What test voltage should the drawing state?
A: The value set by the safety standard of the end product, supplied by the designer. Name the source standard too, so the fab can check the condition.
Q: AC or DC?
A: Whichever the safety standard requires. Criteria and leakage readings differ between them, so the drawing must state one.
Q: Why do boards break down at the edge?
A: Usually edge burrs, V-score smear or copper too close to the edge cause surface flashover. The dielectric itself is often intact.
Q: How are base-bonded ground nets handled?
A: Exclude them from the hipot test and mark them in the net grouping note, or the test reads as a short.
Q: Is 100% testing required?
A: It depends on the product safety class and customer requirements. 100% and sampled testing affect cost and lead time differently, so state it with the order.