Flying squadIf you need a fully populated board tested and in the customers lap in a couple of days then call in the flying probe – second generation systems have improved immensely. Richard Ball reports Improvements in printed circuit board density and reduction in pitch on surface mount components are making test access to pins, pads and vias extremely difficult. This has dealt a blow to the effectiveness of automated test fixtures.
A solution gaining more recognition and use is the flying probe, also known as the fixtureless tester.
Flying probe test systems started life as a slow and crude method of checking connection and whereabouts of components on a prototype PCB. Four independent probes check component connection, values and look for short and open circuits.
The latest systems are increasingly being used for production runs and have functions that carry out in-circuit test, boundary scan and include vision systems.
The software shipped with the systems takes standard CAD design data and removes the need for purpose built test fixtures. This means a fully populated board can be tested and be at a customer or in the lab within a couple of days.
"Fast prototype turnaround is very important. You need probing accuracy, measurement stability and speed," said Keith Fairchild, north European operations manager at GenRad. A system that combines these key attributes is required to get boards up and running quickly.
"Early systems were pretty crude," Fairchild said. Tolerances of resistors, for example could only be measured plus or minus 50 per cent. "The probe looked for the component, but not much more," he said.
Second generation systems, now appearing on the market, have considerably improved the important metrics of a flying probe.
Leading systems are now using air bearings to improve the speed and positional accuracy of the probes. This is assisted by the use of ACbrushless or linear motors which have closed loop feedback so the system always knows where the probes are.
Positional accuracy has been improved to the point where most manufacturers are quoting an accuracy of around ?5?m. This sort of precision means the flying probe can get to points inaccessible to a more traditional bed of nails tester.
"There's no other way of testing the boards that companies are now producing," said John Chubb, technical director at SPEA. "A bed of nails tester's limitation the ability to make good contact with every net on the board."
If the tester can't do this, there is an option to do cluster testing, where a fault in a cluster of nets is determined in software after testing, but this is expensive. "It's cheaper to buy a flying probe," Chubb said. This also improves turnaround time in the testing process as a bed of nails tester can take up to five weeks to manufacture.
One of the accepted downsides of the flying probe is the rate at which they carry out tests. They have been improved, and can now carry out upwards of ten tests a second. With the improvements in speed, people are willing to take ten minutes to test each board with batches of ten, 50 or even 200 boards, said Fairchild. This moves the system from purely prototype to production boards.
If this is not good enough and a traditional tester is required for high speed once in production, the flying probe can be used much like an FPGA is used before switching to an Asic.
The flying probe is used during prototyping and for the first few weeks of production. "You start with the flying probe and are quick to market," said Chubb. "Then you build the bed of nails."
By the time the design is in production, it should be fixed, making the fixed bed of nails tester safe to produce.
Flying probes can also be used in conjunction with a bed of nails test fixture when checking high density boards. After the bed of nails completes its part coverage, the flying probe finishes off the high density areas or those with fine pitch surface mount components. Although it is versatile, Chubb sees cost as prohibitive to this form of testing.
The actual tests that the flying probes can perform are improving. "The technology is more analogous with in-circuit testing," said Fairchild. GenRad's gr Pilot, for example, measures most resistors to one per cent or less and all capacitors and inductors better than ten per cent.
Software is also playing a part. "The software is much better - reducing the number of travels," Fairchild said. In a typical system, the probe may actually spend just seven per cent of its time testing - the rest is spent flying around.
Software can calculate the best order to do the tests and reduce the wasted time, making significant reductions in testing time.
Companies producing these more up to date testers include Bath Scientific, GenRad, SPEA and Teradyne.
Bath Scientific, a subsidiary of Everett Charles Technologies, one of the biggest test companies in the US, has produced fixtureless testers for many years.
Its latest testers, LinearProbe I and II, are also examples of second generation systems.
Although primarily aimed at prototype or one-off PCBs, the improved speed of the test system means that short production runs can also be considered, the company says up to 160 units per shift is possible.
GenRad's flying probe, mentioned before, is called gr Pilot. "What GenRad has done is adopted a relationship with SEICA, an Italian manufacturer of flying probes," said Fairchild. "We're buying in technology to get a fast start into the industry."
SPEA claims its latest 404X series to be capable of 40 to 50 tests per second. "We're finding that people buying flying probe equipment are making incredibly complex boards," said Chubb. "Bed of nails testers just don't have the capacity."