Future Trends in Flying Probe Testing
Peter Brandt, director of sales for Europe and Japan at atg, sits down with Pete Starkey and Barry Matties, gives his views on market requirements and testing technologies, and explains how flying probe testing is becoming the industry standard at all levels of production—and in many cases, the only practicable solution.
Pete Starkey: Peter, it’s the end of what’s been a very busy and successful productronica show for you. What are the main requests you’ve had from customers this week?
Peter Brandt: At atg Luther and Maelzer, we are focused on high throughput flying probe test systems for the bare board market. Not only in Europe, but also in the Far East, we see more and more customers asking for higher throughput for the flying prob. In Europe, customers want the higher throughput to take out the existing grid system and fixture test system from their market, and in Asia, they push us for higher throughput and quantities for flying probe to more flexible and cost-efficient than in the past with the fixture test systems.
Starkey: To build fixtures for some modern designs is extremely expensive and sometimes impossible anyway.
Starkey: Probe testing was always considered a very elegant way of doing small quantities, but you had to accept that it was a slow way of doing it. We now see it as a proper production test method.
Brandt: Yes. In certain cases, we see it as an essential—especially for big server boards, for example, where it’s nearly impossible because of alignment problems to build fixtures anymore, or in the direction of four-wire testing. If you have a BGA pad of 120 microns, it’s nearly impossible to place two pins on one pad to make a four-wire measurement happen. And it’s not only the economic decision to go with flying probe; it’s also a technical decision to go in the direction of automatic flying probe machines.
Starkey: What people are asking for four-wire testing, and for what sort of applications?
Brandt: Four-wire testing is a way of checking mainly the pre-processes for the plating of the through-holes or for the connection of the microvias to the next layers.
Starkey: You’re looking for very small changes in resistance?
Brandt: Correct. They are looking for a very small difference in the resistance, but the secret is to do it before solder mask because, for the contact, you have to get close to the microvia; otherwise, track resistance causes noise in the four-wire testing. For four-wire testing by flying probe, you can guarantee a measurement accuracy of less than one milliohm. And you need this accuracy to qualify the stability of the microvia.
Starkey: Just to change the subject slightly, do your customers understand the power of latent testing methodology?
Brandt: Honestly, no, because the latent test is an old R&D test development by IBM. It’s the only way to measure electrically for bottleneck or mouse-bite effects in long tracks. The latent test is a high-current indirect measurement, which means we measure the quality indirectly by observing the temperature change in the net. Also, for this measurement, you need a four probing technology, meaning a double probe. The latent defect measurements recognize many more defects than a four-wire measurement. But the technology comes at a higher price because the latent measurement is much slower than the four-wire measurement, and it needs more support from the technical department of the PCB manufacturer.
Starkey: Except for some special cases, it’s not necessary.
Brandt: Right. Latent measurement is for very expensive boards. For instance, for the aerospace industry, you have only certain boards to test, where the stability and the long-term life-cycle of the boards are critical. It’s still a niche market from an economic point of view for mass production or higher quantities.
Starkey: How is the Far East bridging the gap between flying probe, grid tests, and dedicated fixtures?
Brandt: When you see all the PCBs produced worldwide, the dedicated tester just with a press and a dedicated fixture with wired cables is the most common technology because the quantities are high, and the machines are inexpensive. The next steps are the universal testers, but universal testers have their limitations with density and technology. With our flying probe machines, and with automatic machines, we try to get market share from the universal test market. When the technology comes to a higher level—and when the customer asks for higher threshold levels, four-wire testing, and bigger boards with alignment problems—we try to enter the medium quantities in Asia with our flying probe. In the past, the breakeven point was in the range of 100–300 PCBs per lot; now, we reach 1,000–2,000 sometimes. Step by step, we try to enter the medium quantities, as well as in the Far East market.
Starkey: And you’re doing this successfully?
Brandt: Yes. At the moment, 40% of the machines we deliver worldwide are equipped with automated loading and unloading systems.
Starkey: Are there any standards and accepted traceability support for test results?
Brandt: There are, especially for automation, because it’s a preferred solution to guarantee traceability in the test process. In all of our automatic machines, we offer ID recognition for all the boards—barcode or DMC code recognition—where we check the board number in the beginning, make the test, mirror the test result to the related ID number, and write it back to the company server. At the moment, everybody is working on this traceability and loop system as a single solution for electrical test.
In the future, we will have complete process capability. That means we would be able to include the results from the optical alignment system in our flying probe system, and the defect images from AOI, in this ID process and write them back to a common server structure. There is also the need for complete processes in the PCB industry. We now have a customized island solution, but in the future, it will cover the whole production process at our customers.
Barry Matties: Traceability is an interesting area because we see more and more the ability to trace down to the smallest level of detail these days, which creates a defect liability traceability. Do you ever hear about that in your conversations with your customers?
Brandt: They want to have a closed-loop from the printed circuit process to the loaded board process because the ideas in the PCBs when they have labels on the board are different than those in the loaded board business. The idea is to have a barcode ID on the PCB, which can also be used in soldering, pick-and-place, and the final electrical test process. Then, you can come back with a closed-loop, especially in the end the final production; if a part number shows any failure at the final customer, they can see when the boards were soldered, where the boards were produced, what were the test parameters in production, and what operator was at the end responsible for the handling of the boards.
Matties: And responsible for the failure; that’s where the financial liability would lie.
Brandt: It also proves where the problem comes from and which parameters were used for testing or measuring this kind of PCB.
Matties: It puts a whole new pressure on building it right.
Brandt: Yes, because in the past, if it was a manual machine, it was easy. When you had a problem with a PCB, you could have a manual machine. It was operated on the night shift to test the board, and, at the end, it was put on the wrong stack. But with automation and traceability on the board, it’s not as easy anymore.
Matties: And it also comes back to what we’re seeing in digital factories, such as being able to integrate with the digital factory like at the GreenSource Fabrication facility. We saw your equipment there—the atg A8a automated machine. How’s that process going?
Brandt: The easiest integration is a data handling interface because you can define software formats to read in and read out. Everything can be adapted with software. The challenge for a company like atg supplying a solution with automation will be integration with automatic guided vehicles. This means you have routed boards coming in and out with an automatic vehicle on a special track to your machine, which will be interfaced automatically to your machine. Then, you need an intelligent loading system. Bring the loading inside the machine and make a separation again to the AGV system with separated good-board and bad-board AGVs.
This is a final solution, but the challenge will be that there is no standard. If you have a customer asking if this for the future AGVs, there is no standard AGV machine. This means you have to develop a completely new mechanical solution for customer A than for customer B with a different security system, robot system, and covers. Data things are easier to handle than mechanical changes.
Matties: That leads to the new developments that are expected because this is one of those new expected developments. What other developments are you working on?
Brandt: As a facility, and as a system integration process, integration is huge. We are also working toward more automation and full automation. But for us, the key to market improvement for our machines is always throughput because it gives us the possibility to offer our machines at more facilities, as customers come to rely more on flying probes and less on-grid systems. Twenty years ago, you may have seen some non-contact testers behind curtains at the exhibitions. Here, we do not see any kind of theoretical solution, so we have to take motors and mechanical solutions to move the probes faster with higher accuracy to the point. We do not see any kind of indirect electronic beam or laser beam solutions over the next 10 years; we only see electrical test systems and electrical measurements to guarantee this test.
Matties: But you have to have the lightest possible touch.
Brandt: It’s a light touch, but it’s a touch, and it has to be done faster and faster. It’s a sequential measurement, and we have to improve this. This is not a big jump, but a constant process optimization.
Starkey: Peter, thank you very much for this conversation.
Brandt: Thank you, and see you at IPC APEX EXPO.
Starkey: We look forward to it.