DMT (Design For Manufacture And Test):
What Does This Really Mean For Controltek Customers
Jeff Skinner, Vice President of Engineering

Acronyms such as DMT, DFM (Design for Manufacture) and DFT (Design for Test) are often heard by companies looking for engineering design services. These terms are usually received very positively from prospective vendors without any real assurance of what that will translate into with regard to their particular product. Certain assumptions are often made that any level of DFM or DFT implemented will result in lower manufacturing costs and higher reliability. Let the buyer beware! What you actually get from a design that supposedly includes DMT, DFM or DFT can vary widely from one design house to the next. By its very definition, design for manufacture must include significant detailed information about the actual manufacturing processes that will be used to fabricate and assemble the product on the manufacturing floor. The less specific information a design firm has about these processes, the less effective the DFM aspect of their design will be. Controltek, as both a custom engineering design firm and contract manufacturer has a distinct advantage over other design firms with regard to DFM and DFT. Products designed at Controltek are also manufactured and tested at Controltek. As a result, Controltek engineers have access to every detail of the manufacturing and test process. In general, companies seeking engineering design and manufacturing services do not select a contract manufacturer until after the design is complete. Contract manufacturers can not quote on the manufacture of a product until they have specific information about what they must build. This information is not available until after the design is complete. As a result, the design firm can not obtain detailed information about the manufacturing processes that will be used or how the device will be tested. Design for manufacture in such a situation becomes very limited. The story with regard to DFT is much the same. Without detailed information on production test techniques used by the manufacturer, effective design for test can not be accomplished.

The Controltek Difference DFM:
Having access to detailed information on the manufacturing process is only half the story in DFM. How that information is applied is the other major piece that determines how effective DFM is in any design. The goals of DFM are lower manufacturing costs, improved product delivery and higher reliability. This starts in the design phase with component selection. Issues to be addressed with each component include:
i.) Part cost
ii.) Part availability (lead time)
iii.) Minimum purchase quantity
iv.) Second sources
v.) Part maturity (obsolescence)
vi.) Package (compatibility with manufacturing processes and capabilities)
vii.) Internal part number ( Is this part already stocked)

Generation of a bill of materials should result in a data set that allows the client to submit the data to any EMS (Electronics Manufacturing Service) provider and obtain a manufacturing quote. It is surprising how many design services produce a bill of materials as output from a design that provides insufficient information to actually quote or buy parts. Distributors must at a minimum have a valid manufacturers name and part number to access information on a part. With that information any design service can provide data on items (i) through (iii) above. With some additional work and support from distributors second source and maturity information can be obtained. Items (vi) and (vii) can only be addressed with information on manufacturing process limits from the EMS vendor that will be used to manufacture the product.

The importance of component selection can not be overstated. Improper component selection with regard to DFM will cost the OEM client money on every production unit throughout the life of the product. Part cost is somewhat obvious, why design in a $3.00 part when a $1.50 part will meet the applications specifications? Sometimes design engineers will buy a precision part when they don’t need to in order to save time and NRE (non-recurring engineering) charges in a design. To determine the minimum grade part you can use requires analysis of the total error contribution in the circuit. This takes time and costs money. Sometimes over specification of a part is a good trade off. If the production volumes are low the client may not be able to recoup enough savings through lower unit costs to cover the NRE charges associated with determining the least expensive part that could have been used. Other times production volumes justify the time and expense of getting down to the lowest cost possible. Over time, the cost savings on your bill of materials per unit can add up to a very significant amount of money. Part lead times are another issue to be addressed in component selection. Lead times change due to the fluctuating nature of the electronics industry, however, some parts are consistently a problem and should be avoided if possible. Additionally, some fabless IC manufacturers are notorious for the poor availability of their parts in production. Such manufacturers should be avoided as well. Long lead time parts make product availability difficult and force the EMS provided to purchase and stock sufficient parts to cover production volumes over the lead time. These stocking fees are passed on the customer and increase the cost of their product. Minimum purchase quantities are another source of stocking fees. If your annual production usage of a part for an assembly is 1000 pieces but you have a 3000 piece minimum buy you are forced to stock this part for 3 years carrying the cost of that inventory the entire time. Again these fees are passed on the customer. The number of parts that are stocked should be in line with the annual usage of the part to avoid unnecessary fees. Perhaps the worst thing that can happen to a production design is to find that you simply can’t get a part required in the design. This will effectively shut down production of the product. This can happen due to a manufacturers part going obsolete with no second source available. Selection of sole source mature parts near their end of life (many semiconductor manufacturers will list these parts as “not recommended for new designs”) should be avoided. You are not doing your client any favors by putting them in the kind of bind this issue often creates.

Good DFM includes consideration for the EMS providers process limits. Designing in a part with a 0.5mm pitch BGA package when the EMS providers equipment and process limits can not place a part with a smaller pitch then 0.8mm is going to create a problem. Unfortunately for custom engineering design firms with no tie to an EMS provider, information on process limits is simply not available. Other process limitation issues to be considered are listed below. It is important to note that form factor and component selection are determined based on industry standards and capabilities that are common within the EMS world.

• Maximum or minimum PCB size for stencil printing, SMT pick and place, reflow oven or wave solder machines.
• Smallest device lead pitch for stencil printing or SMT pick and place.
• Component height limits for SMT pick and place, reflow or wave solder.
• Maximum and minimum PCB thickness for stencil printing, SMT pick and place, reflow or wave solder.
• Smallest component size for stencil printing and SMT pick and place.

At Controltek, our manufacturing engineers are a part of our design review process. This process includes a bill of materials review where package types are examined and a PCB layout review. In the PCB layout review several things are considered. Below is a partial list of the items reviewed by the manufacturing engineers in the PCB layout review.

• PCB form factor is within the limits of common equipment.
• Parts have sufficient clearance to the edge of the board to accommodate a wave solder machines attachment to the PCB.
• Board traces have sufficient clearance to panelization tabs such that depanelization will not break traces.
• Global and local fiducials are provided in the layout for accurate part placement.
• Vias are located a sufficient distance from part pads so as not to pull solder away from the part lead.
• Part footprints on the board have sufficient pad size to allow a good solder fillet.
• Connections to copper planes have thermal relief’s so that solder will flow and wick to part leads without excessive heating.
• Copper planes are distributed symmetrically in the horizontal plane around the board mid plane to prevent potato chipping (curling of the board due to uneven thermal expansion of the copper planes during heating) of the board in reflow or wave solder.

Careful attention to DFM issues in the design process yields a production product that has far fewer issues over its production life and can be produced at a lower cost. These are all desirable results but can only be fully achieved when the design engineers have access to manufacturing expertise. DFM without such knowledge is very limited and will not produce the expected result.

The Controltek Difference DFT:
In most custom engineering design firms, production testing is not addressed at all. This is seen as the responsibility of the EMS provider. Designs are released to the client by the design firm who then selects an EMS provider based on quotes to produce the product in the estimated annual volumes requested. Once an EMS provider is selected they have to determine how they can test the product prior to shipment to the customer. Test coverage in such cases can be limited and fixtures expensive. There is another way to address production testing. It can be a part of the initial design requirements if the design firm has knowledge of the test capabilities of the EMS provider. If an EMS provider has not been selected, it is not possible to fully address production testing. Because Controltek is both a custom engineering design firm and an EMS provider, it is possible for Controltek to fully address this issue. In fact, at Controltek we have a dedicated test engineer who designs production test fixtures for our designs. Our test engineer is involved early in the design phase. When the architecture for a design is established, our test engineer is a part of the process, making sure a production testing methodology has been established. At the project schematic review our test engineer is involved checking that the methodology established at the architecture level has been properly implemented. Test fixtures are designed to maximize test coverage in a cost-effective manner. There are two primary methods Controltek uses for testing at the board level. For products that have significant volume, in circuit testing (ICT) provides the highest percentage of coverage and the volume justifies the expense of the test fixture which typically run between $8,000.00 and $12,000.00 each. For lower volume products where the volume can’t justify the expense of an ICT fixture Controltek uses a custom ATE system designed by our test engineer. This system consists of GPIB programmable test equipment under the control of a PC running National Instruments Lab-View software. A standard set of connectors electrically connect the test system to the fixture which is unique to each product. The device under test (DUT) connects to the fixture. Both of these systems can support the JTAG (Joint Test Action Group) interface and boundary scan. Boundary scan testing through a JTAG interface is fully supported by the HP3070 ICT system and can be supported through custom software in the ATE system. For product testing at the system level the custom ATE system must be used.

Testing methodologies are numerous and generally based on product cost and volume. High volume product is generally tested at the board level using the ICT system. This does not require any special hardware in the design just to support board level testing, although if it is a processor based design, a processor with a JTAG port supporting boundary scan makes testing easier. Products that have a high cost can support special hardware to facilitate testing, as the added cost is not significant in the total cost of the board or system. In such cases one can consider a JTG boundary scan chain where every circuit can be exercised through the JTAG connection. Where coverage does not exist in the design a JTAG supported part can be added to interface with that circuitry to allow testing control and response. At the system level, the JTAG chain can allow for nearly 100% coverage of the design. What must be kept in mind when designing a production level test fixture is testing time. Time is money and the customer will be paying the manufacturing rate or higher for test time. Test fixtures should be designed to be fast. This also helps support throughput of product. Another thing to keep in mind is that production test fixtures should not be designed to validate the design. That work was done before the product was released to production. What you are looking for with a production test fixture are manufacturing defects only. Often times a customer will want to provide a test fixture for production. More often then not, this is the engineering test fixture used to validate the design and requires 20 or 30 minutes to test a unit. This type of fixture is completely unacceptable in production testing. Time required in production test should be measured in seconds or just a few minutes at most. A 20 or 30 minute test per unit will be cost prohibitive and kill your throughput. In densely-populated designs, sometimes there is very little access to test points and it is difficult to get good test coverage with standard methods. In this case, built-in self-test (BIST) is often the best solution. At design time, this can be an easy function to add to a circuit. If the need for BIST is not discovered until the product is going to production, it can result in expensive delays or product going to customers poorly tested.

By addressing a production test methodology at the initial architecture level in a new design, a fast, efficient, cost effective and high coverage test approach can be realized. This requires knowledge of the testing capabilities of the EMS provider. Armed with such knowledge, the product design team together with test engineering can design the selected test approach into the product from its inception. Through such an integrated development the customer will have better test coverage, fewer field issues and lower testing costs per unit. As both an engineering design firm and EMS provider, Controltek is in a unique position to offer the fullest realization of DFT possible and that is the Controltek difference.