PCB Testing Quotes: Fixture or Flying Probe
Electrical test is either flying probe, which needs no fixture and suits low volume, or a dedicated fixture, which costs more up front and far less per board at volume. The crossover is usually in the low thousands of boards. Bare-board test proves continuity and isolation; it says nothing about whether the assembly will work.
Lead times are industry ranges, not commitments. Any factory's actual date depends on its queue that week, and on what else is already scheduled.
ChinaPCB.org does not quote boards and takes no commission. This page exists so that the enquiry you send to a factory is complete the first time. Send it yourself — see how to send an RFQ.
What this quote is priced on
Five to seven variables move the number on a testing quote. They are not equally important, and only the first two are usually discussed.
- Fixture or probe. A fixture is a one-off cost that pays back per board. Flying probe has no fixture and charges machine time per board, so the cheaper option flips with volume.
- Net count and density. Test time and probe count scale with nets and test points. A dense board with 3,000 nets takes longer to program and longer to probe.
- Coverage level. Continuity and isolation, or adds impedance, microsection and solderability. Each addition is a separate test with its own equipment.
- Fixture storage and maintenance. Fixtures take space and need probes replaced. Suppliers charge for storage or scrap fixtures after a period, and that is a real line.
- Reporting. A pass or fail count is cheap. Net-level fault data and serialised records cost more and are what makes a failure traceable.
The parameters a factory needs
This is the set that has to be in the enquiry before a testing price means anything. The third column is why each one changes the answer, which is also the reason a quote that skips it is not really a quote.
| Parameter | What to state | Why it changes the answer |
|---|---|---|
| Netlist and test points | The netlist, plus which points are accessible and which are covered by mask or components | Coverage is defined by accessible points. A netlist without test points leaves the coverage level undefined. |
| Coverage level required | Continuity and isolation, or adds impedance, microsection, solderability | Each added test is different equipment. "Full test" means different things to different suppliers. |
| Expected volume | Units now and per year, and whether the design will be reordered | This decides probe versus fixture. Quoting a fixture for 50 boards is a large cost for no benefit. |
| Fixture ownership | Who owns the fixture, where it is stored, and what happens at end of life | An unowned fixture cannot move with the product. Storage charges appear a year after the order. |
| Acceptance criteria | The resistance and isolation thresholds, and any IPC-9252 class | Test thresholds decide what passes. Defaults vary, and a board that passes one supplier can fail another. |
| Handling of failures | Whether failed boards are returned, scrapped, or analysed | Failure analysis is additional work and the only route to finding a process problem. Scrapping without data hides it. |
| Panel or single board | Whether testing happens on the panel or after separation | Testing a panel needs a larger fixture and different probes. It also prevents separated boards from being tested at all. |
| Reporting format | Pass or fail count, or net-level data with serial numbers | Serialised net-level data is what makes a field failure traceable to a manufacturing step. |
Send with the enquiry
- The netlist and the test points, and which points are physically accessible.
- The coverage you need, in named tests rather than "full test".
- Volume now and per year, and whether the product will be reordered.
- Acceptance thresholds, or the IPC-9252 class you require.
- What should happen to failing boards: return, scrap, or analyse.
Where it goes wrong
- Bare-board test assumed to cover assembly. It proves copper continuity and isolation. It says nothing about whether components are placed, oriented or soldered correctly.
- A fixture bought for a short run. The fixture costs more than the boards it tests. Flying probe would have been cheaper and just as conclusive.
- Failure data not retained. Failures are scrapped without analysis, and the process problem continues into the next batch at full cost.
- Coverage claimed without accessible points. Nets under components or mask cannot be probed. The advertised coverage was arithmetic, not physical.
Comparing two testing quotes
Two prices are only comparable when the parameters above are held constant. Most of the spread buyers see comes from a difference in one of these, not from a difference in factory efficiency.
- Compare the coverage, test by test. "Electrical test" is not a specification.
- Compare fixture ownership and storage terms. They are small now and significant at end of life.
- Compare the acceptance thresholds. Two suppliers can disagree about the same board.
- Compare what happens to failures. Analysis is the only path to a process fix.
What this page does not tell you
This page gives no crossover volume figure, because it depends on net count, fixture complexity and local machine rates. The structure of the trade-off is stable; the number is not.
Related
- PCB testing: flying probe, fixture and coverage
- Assembly quotes: where functional test belongs
- 6-layer quotes: why inner layers need test coverage
- Coating quotes: coat after test, or lose the option
Fee structure and editorial independence are described on the about page. What we can and cannot confirm about a factory is on how we verify.