Global parts procurement presents challenges to any product designer. If a component is sourced from multiple suppliers, how do you keep track of these suppliers from initial design to production, especially in a global manufacturing environment where products are made in multiple locations? Do you really have the right manufacturing and process controls to manage such an environment? The unit submitted to a certified test facility had a specific supplier but is there assurance that other suppliers will perform the same during testing? If not watched carefully, these variables could wreak havoc with product certification and regulators worldwide.
Coupled with the drive to continually decrease cost and the pressure to reduce the time to market, product performance can be at risk due to the scenarios noted. That, coupled with the fact that circuits generally contain more than a single component, creates a complex situation to analyze and multiple risks could be present.
The following example takes the reader thru a real-life example of the complexities and inter-relationships between: 1) the component, 2) its deployment into a typical consumer product, and 3) how the manufacturing environment can exacerbate an already difficult situation. The investigation speaks to a consumer-based product that contains radio frequency transmitters and receivers as part of its architecture. Buckle your seat belts!
What a Starting Point – The Customer!
A consumer product employing a transmitter and receiver is produced in high volumes while being manufactured in multiple locations around the globe. Additionally, multiple suppliers are used due to the volume demands placed on the manufacturing facility. After six months of production and customer shipments the service team (Px) began reporting warranty returns from customers. The primary customer complaints were two-fold: poor product performance along with excessive heat being felt on the exterior packaging, especially after long product usage times.
As the frontline, customer-facing organization, the ‘Px’ team was the recipient of those happy customers returning their units. Their diagnosis noted:
- low RF power output or no power from the transmitter
- 3x increase in RF PA field failure rate as compared to the prior 3 months
- conducted spurious emissions that were over the specification limit
- multiple sources of the RF PA were present in the customer returned products
- several manufacturing locations for the product were used
- excessive or abnormally high surface temperatures on the products surface
The Basics of Risk
In the real world, most problems that a manufacturer, supplier, or design engineer faces have multiple reasons, or variables, that contribute to the problem and its occurrence. This is represented in Figure 1 where a problem could be caused by anyone, or multiples of, four (4) separate root causes or variables.
Figure 1: Four separate root causes or variables
Some of these variables can be major contributors to the problem occurring while others are less likely to create the problem unless combined with other less minor ones. Variables can take many different forms, such as process or manufacturing defects, actual material defects in the component, design defects or a mix of all of them. Trying to sort thru these in a logical manner can be time consuming as well as frustrating.
Part of the challenge in fixing problems with multiple variables is to properly identify all the possible root causes or variables that could potentially contribute to the problem and not rule out the “obvious”. This process can be problematic as the number of possible variables increase. In situations where there are many variables, more discipline and rigor are needed in order to result in an effective solution to the problem. Industry has adapted many tools to help identify and resolve problems with multiple variables, probably the most popular being the Six Sigma approach. Figure 2 is a general representation of how complex it can be to solve a problem as the number of variables increases.
Figure 2: Relationship to complexity relationship
Problems do not always occur at the same interval or frequency, and when they occur it could be either a major or minor issue. There is a logic that states the cost to solve a problem, i.e. minimize the risk of it occurring again in the future, may be more than the total cost had the problem actually occurred. For example, if it costs $5000 to prevent a problem, but the actual impact of the problem occurring is $2000, why worry about it? Why spend $3000 more on a bet that the problem may occur again? The answer here lies on what type of impact the problem has. Is it regulatory? Is it safety related? Is it a customer issue? Is it a business or market ethics challenge? Did it create a situation where multiple clients could be lost? The bottom line is that caution is needed when examining and identifying risks.
Problem Analysis and Results
The point of this article is not to debate the effectiveness of a Six Sigma or other approach in problem solving. Rather, the results are summarized in a context intended to point out 1) how a
defective process impacts product performance and 2) the importance of component variables in that environment.
Assign the Right Owner – Who Is the Best Owner?
When faced with many variables and unknown ties, the first step is to assign an owner. This is paramount to anything else. Thinking this is a simple decision can be a dangerous assumption. The right owner needs to be multi-task orientated, possess the ability to step out of the box and think strategically in a broad sense and, most importantly, know the tools that are at his/her disposal (i.e. Six Sigma) to resolve the issue. This person should not be quick to rule out the obvious or to dismiss the items or data that appear unrelated.
Program Manager – Identified the Right Team Members
In our real-life scenario, the business tapped a program manager who had broad skills in interfacing across the organizations, was a strong networker, and possessed the leadership skills to manage multidisciplinary teams. Supported by a Six Sigma Black Belt, he identified key team members needed to resolve the issue. Talent in the design engineering, supply chain manufacturing, product service and component supplier support teams were chosen. The program manager assembled the team and the ‘Px’ service organization was contacted to get first-hand, non-filtered information.
Validate the Currently Known Data – It Pays to Double Check!
The process of making sure the existing facts are accurate ensures a foundation exists upon which future effort can be based. In many cases, wasted time is prevalent when an issue is being addressed without having accurate data. Inaccurate data can quickly take investigations down an entirely wrong or parallel path that is inconsequential to the real cause.
In our example, the spurious report provided by the service organization was reviewed along with the test fixture/system used by the ‘Px’ service team that took the original data. That investigation discovered the ‘Px’ team’s test fixture for measuring spurious used an out-of-calibration RF spectrum analyzer and had cables not traceable to a calibration date. Engineering then re-measured the same defective RF PA devices and found all spurious to be in specification.
Figure 3: Service versus engineering data
As part of validating the reported heat rise, it was found to originate from an older prototype unit that was pre-production. There was no indication on any service repair log for any incoming repair on this pre-production unit. Yet why then did the service team include it in the failure rate calculations and reports? The heat topic became real, as opposed to being a non-issue, because the ‘Px’ service organization thought that the original prototype issue may be related to the current problem and appended to the service report with some comments in an attempt to provide added information that might be of help.
Get the Global Data
Globally, all service locations were queried for the problem but the resulting data indicated that only the Asia regional center was reporting the customer issue. Sample units of all failed product were collected and catalogued from the Asia branch. Each unit and suspect RF PA was analyzed for the RF PA manufacturer ID number, lot number, and manufacturer. Service records were analyzed to get the RF PA manufacturing location, RF PA date code, and other information pertinent to the manufacturing history. The manufacturing history was summarized.
From these records, two RF PA suppliers (VG, XJ) were found in the suspect units, each with two versions of the RF PA device – d2 and d3. Three RF PA manufacturing date codes (V2, V3, V4) were logged from two manufacturing locations for the RF PA (EU4, EX4). All available RF PA devices were catalogued for ID marking details. The failed RF PA’s were validated as defective with limited or no power output in the standard test fixture. Heat rise was also measured on the RF PA surface on the operating RF PA’s. Shown below are some examples of how the RF PA was marked and the specific meanings.
Using the catalogued information from all defective RF PA’s, the results were summarized as shown in Table 1.
Figure 4: Examples of RF PA markings and specific meanings
|RF PA Suppliers Noted||RF PA Versions Found||RF PA Manufacturing Locations||RF PA Manufacturing Date Code|
|XJ||d2, d3||EU4||V2, V3, V4<|
|VG||d2, d3||EX4||V2, V3, V4|
Table 1: Summary of defective RF PA data
RF PA Supply Chain Investigation
Since the data indicated more than one supplier, an entire history of the RF PA device was reviewed. Each supplier, XJ and VG, was contacted and requested to send all of their release records pertaining to revisions of the RF PA. In house purchasing sent all purchase records for review and comparison to the supplier provided data.
Examination of these records indicated that the two versions of the RF PA from each supplier (d2, d3) were approved for purchase along with the original d1 from each; the versions were qualified by the component qualification team, and delivered quality audits passed. These RF PA versions (d1, d2, d3) were shipped globally to all locations, had the same base part number, and had version indicator stamped next to the base part number.
Globally, all RF PA product manufacturing histories were examined from initial production to current inclusive of changes. A total of five (5) RF PA manufacturing date codes (v1, v2, v3, v4, v5) were noted as approved for production in all RF PA manufacturing facilities, including EU4 and EX4, globally for the RF PA’s by the Vendor Quality Organization.
Engineering writes and releases all changes to the factory, including component changes or revisions with suppliers. It then relies on the manufacturing team to complete all other tasks. Examination of the engineering records indicated they were aware of all date code versions v1-v5 of the RF PA. However, in their records the RF PA manufacturers (XJ and VG) advised them that some internal grounding changes inside the RF PA device would be present in date codes v2-v5, for all versions of the RF PA itself (d2-d3), which resulted in a different power output versus frequency performance of the RF PA. To compensate for that, engineering had to change the phasing requirements for the actual consumer product so its RF performance would continue to meet specifications when used with date codes v2-v5 and RF PA versions d2-d3.
Knowing the RF PA history thoroughly, the next step was to contact the manufacturing locations that produced the consumer product. In examining the engineering change notice sent to the manufacturing site for v2 – v5 approvals, it was verified that version v2 date codes and greater did require the factory making the consumer product to change the phasing criteria for the product transmitter to match with the d2 and d3 versions of the RF PA that had v2-v5 date coded RF PA’s. Knowing this, the ‘Px’ team was contacted and asked to measure each defective customer unit to determine the phasing table values. The phasing data embedded in each of the defective consumer products was then checked and found to be the older version of the phasing data, which is not compatible with RF PA devices with date codes between v2-v5.
Figure 5: Summary of phasing values and RF PA data codes
Product Factory Test Organization
Since there was a needed change in the phasing of the product, the test organization supporting the factory was contacted. They confirmed they were contacted by engineering and did change the parameters but had no effective date of implementation into production, so the older phasing version stayed as is pending them hearing back on when date codes v2-v5 would be incorporated into the consumer product. No confirmation was ever received so the master IT system that controls all phasing data used in the consumer product was not activated.
Once this was discovered, inventory control was contacted to check their records of material release to the factory. Their files indicated all versions of the RF PA, d1 – d3, were released to production along with all versions V1-V5 of the RF PA.
The Story Closes – Summary Results
After calm heads prevailed, the following solutions and improvements were implemented:
- The change control document was updated requiring signatures from all parties involved as opposed to verbal communications. Crossover dates were added that specifically tied to all organizations involved.
- A new supplier change control order was implemented that required all suppliers to report in writing any change relative to the specification of components, including the RF PA.
- Going forward, unique RF PA part numbers were assigned to minimize intergroup coordination during complex crossovers.
- The quality program overseeing the ‘Px’ team updated their procedures to ensure only customer reported data was entered in the ‘Px’ defect reports and that traceable calibration activities were implemented.
The Bottom Line
The story above reflects a real-life scenario, with names changed to protect the innocent. It points to how important a robust process can be and delineates the linkage to the component, how it is used and manufactured in a product, the effects it can have on a customer and/or business, and the potential to create major disruptions in delivery.
Exposure to so many variables can result in a comedy of errors and assumptions, with the victims being cost and performance, as well as the customer. From a regulatory view, it creates significant risk which could potentially lead to recalling products from the market place and/or fines and business impacts such as a dropping stock price or reduced market share. It is critical that variables be understood and embedded into day-to-day operations before they can create problems. It is an incorrect to assume that including them always increases complexity. Separately, product certification needs careful examination to determine if and when a product needs to be tested and/or resubmitted to a country regulator for approval. Not doing so, or assuming that no variables exist that could impact regulatory approval, is wishful thinking.
Change is always healthy and having an independent view on just how robust a business process is should be considered.
|Peter S. Merguerian is President and CEO of Go Global Compliance Inc. (www.goglobalcompliance.com) and provides regulatory engineering consulting and global certifications for companies worldwide. He has 30 years global regulatory compliance experience with an emphasis on safety, EMC, wireless and telecoms where he had corporate-wide responsibility in various global test laboratories for Market and Conformity Surveillance, Regulatory and Testing Services, Global Engineering, Accreditation and Global Certifications. Mr. Merguerian holds a Bachelor of Science Degree in Electrical Engineering from the Illinois Institute of Technology, Chicago. He speaks five languages and owns and moderates two popular global regulatory groups, one on Linked In: “Global Regulatory Compliance” and the other his blog at www.goglobalcompliance.com/blog. Mr. Merguerian can be reached at email@example.com.|
|Dennis W. Bartelt is President of Bartelt Consulting Corp. (BCC). He has 30+ years of past experience in telecommunications and consumer products with Motorola Mobility Inc.(formerly the cellular division of Motorola Inc.). His credentials include product design, software quality, service and repair, product safety, regulatory and environmental compliance, customer satisfaction and process enhancement. Under his leadership the teams he managed provided approvals for more than 100,000 unique products. Dennis has a Master’s Degree in Electrical Engineering Technology from Northern Illinois University along with 30 hours pre-doctorate work. He is currently consulting with industry firms and specializes in process enhancement in the product regulatory and safety compliance areas. Dennis can be reached at firstname.lastname@example.org or dennis@gogloba `lcompliance.com.|