If your design reviews are not covering the needs of the test, product engineering, failure analysis, packaging and systems community you may have a design that is rock solid and not meet volume production requirements. The design may be too expensive to test, have a failure mode that is difficult to debug or may not meet the intended application. What a disappointment to find out you have a design that’s worthy of first time success status only to find it must be spun for issues not directly related to specific design functionality.

To justly make it to first time success through production manufacturing it is essential that a review process is in place to cover the requirements of all of the product support functions in addition to the design details. It is suggested that the reviews be broken down into two levels. One tier is a high level review and covers communication to the non design disciplines (manufacturability) and the other tier is a detailed peer (design to designer) review. Find out what product, test, FA, packaging and systems would be concerned about in a design and make sure to prepare and present that information in the high level review. Extra work? You bet. If you feel the additional effort is not worth it, consider the risk of at least an 8 week schedule slip for a spin. It comes down to a conscious tradeoff of production schedule risk against the additional time required to properly prepare for a thorough review.

One thought to help minimize the “extra work” perception would be the integration of the necessary product, test, FA and marketing review deliverables as part of the normal design process flow. Having a design process template or guide in place that covers these additional review requirements will highlight these deliverables, allowing them to be created real time when the specific design activity is fresh. The perception of extra work fades as it becomes 2nd nature through integration as part of your normal process.

The peer design review is a nuts and bolts review of a design with a small team of designers. The review is a discussion between designers and must not be in a formal presentation format. If formality creeps into the process you will have minimized the desired discussion atmosphere and the required depth of review will be lost. The intent is to go through the design schematics and layout in detail, identifying areas of concern related to the implementation and/or validation of the design.

The ideal makeup of the peer review team is 2 experienced and 2 fresh (0-3 years) designers. Ideally, they should have limited knowledge of the design to be reviewed. The experienced designers bring along previous battle wounds from their history of design work. The fresh designers bring a lot of good, probing questions that come about by their zest for learning design techniques. It may make sense to share design resources with other organizations for the purpose of peer reviews to ensure previous knowledge of the design does not cloud the discussion.

The input to the peer review team should be schematics, simulation stimulus/results, layout and the block level requirements. Appoint one of the reviewers as recorder of the discussion and capture any issues that surface during the review. The discussion summary should be forwarded to the design manager for final resolution of any issues identified. The output of this process must be specific actions to resolve any design/validation concerns and must be tracked to closure.

When your in the midst of debug on your first silicon you may find that some of the issues that are found may be able to be fixed, or least improved with a metal and/or poly layer spin. You will want to spin this at your earliest convenience and get a new version of silicon back for customer samples. This is a reasonable course of action.

What happens when you find an issue that requires an all layer spin? This should demand a more complex decision matrix. When looking at an all layer spin another component that must be considered is the manufacturability of the design. If manufacturing validation fixes are not part of your second all layer spin then you have added a reasonable risk that you will need a 2nd all layer spin to address manufacturability issues later.

Manufacturability can best be summed up as parametric margin. Can the device wafer yield to the level that was identified for the business case? You must minimally look at full temp/voltage, ESD and latchup to gain the necessary insight into the manufacturability for the design. If manufacturing issues are found the required fixes will generally require an all layer spin. You will reduce the chances of a 2nd all layer spin by ensuring that manufacturability is worked in concert with design debug.

If you have any specific questions that you would like to see answered here please send me an email and I will address them here anonymously, unless you indicate differently.

email: jeff@jorvigconsulting.com

phone: 480-895-0478

web: http://jorvigconsulting.com