Contact Us
  • Home
  • BLOG
  • Which Test Plan Pays Back at 500 Boards?

Which Test Plan Pays Back at 500 Boards?

At 500 boards a year, flying probe is almost always the cheaper plan; at 500 boards a month, ICT usually wins. The crossover has nothing to do with a magic quantity of boards. It is the fixture cost divided by the per board saving, and both of those numbers come from the board itself: how many nodes need access, how long each method takes to touch them, and whether a fixture can survive the engineering changes the product will still see.

What each method actually measures

In circuit test puts a bed of nails on every node at once and measures them in parallel: opens, shorts, resistor and capacitor values against limits, diode junctions, and on powered fixtures a set of functional checks. Because all probes contact simultaneously, the cycle time is mostly the machine settling and the software running, typically 20 to 40 seconds for a mid size board.

Flying probe touches the same nodes with four to eight moving probes, one after another. It needs no fixture, so nothing is spent before the first board, and it can reach a 0201 pad or a 0.4 mm pitch BGA that a bed of nails cannot. The price is time: a board that ICT finishes in 25 seconds can take two to four minutes on a flying prober, and a board with 1,500 nodes can take longer than that.

Where the money actually goes

Cost lineICTFlying probe
Upfront toolingFixture, probes, wiring, program debugProgram only
Per board time20 to 40 seconds2 to 4 minutes, longer with high node count
Node count effectFixture gets denser, not slowerCycle time grows roughly with node count
Engineering changeFixtework or a new fixtureProgram edit
Access neededOne side, probe pads on a gridEither side, small pads, fine pitch
Best fitStable product, volume, high node countLow volume, prototypes, dense boards

The break even is arithmetic, not opinion. Take a fixture at 4,000 USD, an ICT cycle of 25 seconds, a flying probe cycle of 180 seconds and a loaded line rate of 60 USD per hour. The saving per board is 155 seconds of line time, which is 2.58 USD, so the fixture pays for itself after about 1,550 boards. Change the fixture to 12,000 USD, which a double sided 1,500 node board can reach, and the same arithmetic needs 4,650 boards. Run the numbers with your own fixture quote and your own cycle times before choosing, because the answer flips quickly.

The board features that decide it first

Test point access is the first gate, and it is usually closed before anyone thinks about test. Probes need a pad to land on, and on a dense board those pads get removed to make room for routing. A needle needs a pad of at least 8 mil on an IC or BGA site, with clearance around it, and a via used as a test point has to be open and on the probed side. If the board has no room, the choice is already made.

Coating is the second gate. Probing through conformal coating is unreliable, so either the test points are masked before coating or the board is tested before the coating step, which changes the order of the line and the rework path. Double sided probing is a third: a fixture that probes both sides costs more and needs a handler that flips the panel, while a flying prober does it by turning the board over.

Three mistakes that hide the real cost

Choosing the method before the node list exists. A test plan written from the BOM alone misses the nets that have no accessible pad. Build the node list from the Gerber data and the assembly drawing first, then ask which method can reach the nodes that matter, and accept that coverage will be a percentage rather than a promise.

Counting only the fixture. A fixture with marginal contact delivers false failures, and at 1 percent false fail on 500 boards five units go to rework that had nothing wrong with them. Add the debug time, the rework time and the second test pass to the fixture cost before comparing.

Skipping flying probe on the first articles. The most useful role for a flying prober is to validate the fixture and the test program: run the first ten boards on flying probe, compare the results against the bed of nails, and fix the fixture before volume. That step costs a day and saves a week of false failures later.

How SUPERB PCBA runs the test sequence

SUPERB PCBA builds the test plan from the customer files, not from a template. The sequence before test is 3D SPI at print, 3D AOI after placement, X-ray and CT for BGA and thermal pads, and DIP AOI after selective or wave soldering. Then ICT on a bed of nails, flying probe where the fixture cannot reach or the quantity does not justify one, and functional test on a fixture built for the product, with the pass limits agreed at first article. Probe pads of 8 mil on IC and BGA sites are within the process window, 01005 parts and 0.30 mm pitch are placed on the same lines, and every board carries a serial that ties it to the test records.

Frequently asked questions

Can flying probe replace ICT completely?

For low volume and prototypes, yes, and it is the better choice because the program changes with the board. For a stable product at monthly volume, the cycle time becomes the constraint: the same line produces fewer boards per shift, and that cost keeps repeating while a fixture is paid once.

Do test points have to be on both sides?

Not necessarily. A single sided fixture is cheaper and faster, so the usual answer is to bring as many nodes as possible to one side and leave the rest to flying probe. A board with power planes and connectors on the bottom often cannot avoid a second side, and that is a fixture cost to price in early.

Does conformal coating change the test plan?

It changes the order. Either the test points are masked before coating, which costs a masking step, or electrical test happens before coating, which means a coated board that fails cannot be reworked without stripping the coating locally. Both are workable; deciding late is what hurts.

Next step

Send the Gerber data, the BOM, the assembly drawing and the expected annual quantity. The reply states which nodes are reachable, the coverage each method gives on that board, the fixture cost if one is needed, and the test sequence with the pass limits that will be applied at first article.

Need a quote or a test plan review on your files? pcba@superb-tech.com · WhatsApp +86 133 9608 1443