The Board Authority : How would you rate the importance of a parametric Electrical Test (e. g. actually measuring a resistance) for various applications (mainstream PCB, HDI PCBs) ? Will this issue be addressed in next 356B revision ? Do you think non parametric test (capacitance or similar approaches (field measurement, etc.), and the first generation on non-contact testers) will meet a strong reluctance of the market because actually not measuring resistance ? What would make their acceptance rather wide ?

Duane Delfosse: This is a tough issue because it relates directly to the quality and reliability of the circuit. The role of 356B should be to convey from design and CAM the specific board information needed to accomplish common test methods. If a particular test method has unique data requirements, the standard may not support it. I hope that by giving notice in periodicals, such as The Board Authority, that the industry will come forward and ask for inclusion of the kinds of data they think are needed.

As it is, 356A supports embedded passives, including the capacitance of a net relative to a reference. The question is, are there upstream CAD or CAM systems that calculate these values and include them in the 356 output? Generally, the answer is no. There are sources quite recently for embedded resistors, but none that I am aware of for capacitance or inductance.

As for the basic need for parametric measurements in different types of substrates, I think the need will remain quite strong for these types of measurements. In fact, with some of the new board constructions that are emerging, new measurements are being requested to assure high speed performance and reliability.

There is no question that moving from a contact test methodology to a non-contact methodology will produce more test escapes. The question is how many, and what kinds, of defects will be missed. There should be a direct correlation between the defects that are to be detected, their frequency of occurrence and the test method employed.

In today's environment, board manufacturers own responsibility for delivering "good" boards. OEMs might audit and require particular test methods, but in so doing they begin to accept responsibility for defining what is "good". We have not witnessed many cases where the board manufacturer and the OEM discuss and agree to a compromise in test coverage to make use of a faster or less expensive test method, but these kinds of discussions are starting to occur. Such discussions are unlikely early in a product or process life cycle, and much more likely after some data is available.

Today it seems contract manufacturers are a middle man creating further confusion on the issue of test coverage and liability. Perhaps most difficult is the lack of good data on what sorts of escapes occur and what are their root causes and detection method(s). There has been little structured experimentation on this subject, although I am aware of a recent ITRI project targeted at just this.

TBA : Do you think the PCB industry will widely accept combinational electrical test (Prober + bed of nails, or AOI + bed of nails, etc.) as a new test standard ? Why ? How will 356B cope with this in order to guaranty a 100% coverage ?

DD : These techniques have been pretty widely used over the years and I think they have a role to play in the future. A board manufacturer can't be expected to buy a new test methodology for every new design they produce. Thoughtful use of combinational test makes sense to fill this gap. These techniques do come at a cost, however.

There is increased handling, two sets of results that must be correlated and the accompanying increase in labor, floor space and equipment capacity. Modern shop floor control systems and bar coding help, but these systems will never be as nice as a one pass test.

We will be considering how 356B might be made more robust for these kinds of applications. As it is, you could express a 2 pass test, grid and prober for example, by use of 099 records. We need to provide more examples and check the logic. I am not aware of many instances where 356 has been used that way.

TBA : Just a few years ago, Electrical Test simply meant "bed of nails"? Probers have recently changed this situation. Now we are talking about controlled impedance, crosstalk, etc. How will 356B cope with these new requirements ? How do you envision the future of Electrical Test : a single equipment performing the whole range of test, or various specialized equipment ?

DD : 356A supports the identification of impedance test points and some of the other emerging test requirements, but up stream data sources are not usually able to output such information.

If working from a CAM (graphical) data source, it usually takes a person to identify the particular points of interest. Once you get into it doing this, you will see that the vast majority of signals routed on boards are incompatible with current high speed performance measurements. The traces are too short, are routed through too many vias or have other anomalies that cause signal distortion.

Quite a few people in the industry are of the belief that for the next few years more stringent correlation of coupons to boards, and better modeling in CAD systems can limit the test requirement to coupons and a small sampling of boards. On the other hand 6 different board manufacturers showed automated TDR at the JPCA show. The market is not really clear.

As to your question about a single piece of equipment performing a whole range of tests, it is certainly possible. This approach is most easily accomplished in track type load/unload systems where additional modules can be inserted to accomplish additional tests. Balancing the test time of each component becomes a challenge so that the overall throughput of the system is not seriously diminished.

TBA : What is, according to you, the proportion of PCB shops using IPC-D-356 for test purposes, today in North America, Europe, Far-East and Japan, respectively ? How will these figures evolve in the next coming years ?

DD : It's very difficult to answer because I don't have any real data. My guess is it is commonly used at 50 to 70% of board shops in North America and Europe, substantially less in Asia. In the future I see more penetration.

There has been a tremendous consolidation of board manufacturers and their test and CAM suppliers. 356A has served as a link between the various parts of these new organizations and as a convenient mechanism for competing suppliers to communicate. Also, with the experience of 356A under our belts, I would also expect 356B to be a significant improvement.

If the ambiguities and deficiencies that exist in A are corrected, we will have a very robust and straightforward data format which will help with it's acceptance.