DFx reviews: tools, checklists or graybeards?

DFx–Design for Manufacturing / Assembly / Test / Cost / Reliability /Supply Chain / Environment  etc (hence the x)–should be part of the DNA of every design team, an integral part of designing to meet requirements just like getting the oscillator frequency or the temperature cutoff right.  But despite best efforts and best intents, it’s still prudent to do a DFx review at key milestones.  So then the question is, how best to do the DFx review?

1. Utilize CAD based DFM / DFA tools that check dimensions , spacings, and just about everything imaginable against a set of design rules, flagging the violations

2. Utilize DFx checklists to manually check the design against a predetermined list of subjective criteria

3. Utilize a highly experienced engineer, with gray in his or her proverbial beard, to go through the design, poking it for issues

The modern proxy for conventional wisdom overwhelming supports approach 1. It’s repeatable, automated and can be run by inexperienced personnel and / or in low cost regions.    There are two main issues however with approach 1:

a) GIGO–garbage in garbage out–the check is only as good as the database of design rules.  It’s challenging to do this in a classic, self contained design-and-manufacturing-all-under-one-roof type company; it’s next to impossible to do this in a global enterprise.

b) You find lots of trees but no forest.

The next most politically correct answer is checklists. The underpinnings of many an ISO process, checklists are also repeatable and have the ability to be checked by lower skilled or remotely outsourced personnel.  The challenge here is that the answers are subjective and also more dependent on the skill level of the person doing the check.  It also of course is not nearly as accurate for minute tolerance issues etc that rely on accurate numbers.  All this said, checklists are an invaluable augment to other DFx options.  Zebulon Solutions’s DFx checklist is available for free on our website.

The bottom on most manager’s list is to let the gray-bearded types with years of experience and manufacturing in their blood have a go at the design.  This is typically at the  forest level with zoom downs into the trees and often into the bark and sap and…  Downside is that this only works with highly talented engineers with the right experience; the upside is this path yields the best results over and over.  The winner by a landslide.  And as bonuses, a) because such engineers can do their view in a fraction of the time of a greenhorn with a checklist, it’s often the most cost effective; and b) in the course of doing a DFx review these same engineers can often spot other issues in functionality.

Of course, combo approaches are often even better, and arming a graybeard with a  checklist and an automated rule checker can only help.  And getting ahead of the curve–learning how to design for manufacturability etc up front and utilizing predictive tools like DFMEAs–is even better.

Here’s to the graybeards!

Chuck

6 Comments

  • Pingback: Zebulon's Thoughts · Wisdom and Graybeards

  • Al Onderick

    About 7 years ago we went from Grey Beards to Valor Trilogy (now Mentor Graphics Vsure) and haven’t looked back. We initially went to Valor to implement the Valor Parts Library to validate footprints but we also complete a myriad of DFX checks. The problem we have is bandwidth. We have 17 Designers and two people performing DFX checks and with the current state of the economy adding staff isn’t an option. We do some pretty complex products and a Valor check takes 3-5 days, but it’s an investment in quality and reliability in addition to improving time to marked. I cannot provide you list but I can tell you we perform 362 different checks and keep statistics on overall violations.

  • Chuck

    Valor / Vsure / other DRC are excellent tools in an enterprise environment for catching objective and parametric design issues. Great to hear it’s working well for you. I always recommend to our customers that they do DRC checking as well. The challenge for many small to mid sized businesses is the ROI on setting up and maintaining sucb a DRC system, and for companies that rely on outsourced design and / or manufacturing it gets even more complicated (Been there, done that. I spent 13 years running design services at a major contract manufacturer).

  • nigel mills

    I’d support a more organic approach. It’s easier ensure graybeard exposure before you arrive at a CAD model. Once you’re looking at a model you will have already progressed a long way down a particular road. Reworking that isn’t very lean.
    After even the smallest of ‘assist’ from the greybeard, you have established a small team and network that the CAD guys can call on if/when they need a second opinion. In my experience, that second opinion speeds the decisions made at the screen and makes them more informed. Better, you lose any antagonism between functions as people work together at the right stage in the project.

  • Emi

    Thanks for posting this Marina. I also have stglegurd with this over the years. I grew from 0 to 400 subscribers over 2.5 years in my nature photography business. In my new business, I’ve grown from 0 t0 100 in about 1.25 years. That’s slow growth for sure. My passion has always kept me going. But I’m in the mode of learning about new ways to accelerate readership growth. I especially like Ricardo’s idea above. I need to get out and network again. Todd Smith recently posted..

Leave a Comment