Success Stories

The following case studies illustrate complex engineering problems that were solved and/or business/leadership successes achieved by Luis J. Navarro, Founder and Engineering Solutions Architect of LuNa Consultants.

Problem-Solving Success Stories:

Business/Management Success Stories:


The Case of the Disappearing Defects (Weak Inductors in Ultrasound Assembly)

The problem: The inductors used in an ultrasound assembly were 100% tested before shipment. However, upon receiving the products, customers would test them and find some inductors to be "weak" or defective. When the products were returned and re-tested at the manufacturer's facility, all would pass the factory tests. The customer suspected poor PCB cleaning procedures in manufacturing.

The solution: Systematic measurements proved the inductors could be damaged by temperatures very close to the temperature necessary to melt the solder. The inductor insulation was damaged by temperature and the "healed" / "weak state" could be changed by the application of the proper voltage and temperature. Two sources of overheating were identified and removed and the supplier of the inductor was changed.

The result: After the changes were made no more problems were encountered.

>> Return to Top


Management by Superstition (Contaminated Coaxes)

The problem: For several months, a manufacturing division in México was plagued by wrapped tape cables randomly exhibiting abnormally poor signal transmission. Sometimes, the problem could be cured by baking. Humidity during transportation or in the México assembly plant was initially blamed. Special packaging and desiccants were used but problems continued.

The solution: Observation of the coax manufacturing process showed that after the tape was wrapped, it was fed through a trough of full of water before adding the shield. Measurements made after the coax was completed showed no dissipation therefore, no water absorption. A cleaning operation after the product was made used alcohol in excessive amount. The tape would readily absorb alcohol which normally contains some water. When the alcohol evaporated, it left the water trapped inside. The cleaning process was changed.

The result: The defects disappeared 3 days after the first measurements were made and did not reoccur.

>> Return to Top


A Part Delaying a Multi-Million Dollar Announcement (High Speed RF Cable Variation)

The problem: The customer required a very phase-stable RF cable regardless of how far the cable was bent or twisted. The cable was tested using the industry standard tests for cables of this type and passed at the specified frequency. However, the cable's pulse edge changed more than what the customer could tolerate and the issue was delaying the schedule of their product's imminent launch. The customer stated "this is exactly what your competitor provided, we expect better from you" so the customer relationship was in jeopardy.

The solution: The customer was using the cable as a high speed strobe pulse, making the industry standard tests meaningless for their specific application. Recommendations were made to test with a pulse signal to fit their specific application. Once the test was done, the design was changed and evaluated for performance.

The result: A new sample was made in one day based on our recommendations and the customer had an acceptable product two days after. The customer was able to introduce their new product on-time and the customer relationship was strengthened.

>> Return to Top


Standard Tests Do Not Anticipate Every Flaw (Exploding Resistors)

The problem: In a patient-monitoring device commonly used by Emergency Medical Technicians, a 1 kOhm 1 Watt resistor is used to isolate the monitor (normally reading a 0.001 volt heartbeat) from the defibrillation potential applied to the patient that has cardiac arrest (normally 5,000 volts). If the resistor fails, either the technician would not detect the presence of the heartbeat and continue defibrillation or the patient would get an insufficient dosage. The 1 Watt resistor must withstand 18,000 times its maximum rated power for 5 ms. This means that only certain resistors with large cross-sectional areas would survive. The standard test is to shock the resistor with a simulated defibrillator three times and ensure its resistance had not changed more than 10% after the three shocks. However, when measuring a specific resistor brand, its resistance was changing permanently after each shock. With three shocks, the resistor had changed 9.5%, technically passing. During the next shock the resistor exploded. Other samples of this resistor exploded in 10 or less shocks. Resistors from other suppliers never exploded.

The solution: The appropriate measurement eliminated resistors from unsuitable construction that would have risked the life of patients.

The result: No loss of life due to an unsuitable resistor.

>> Return to Top


A Compromise May Be Good Enough (Camera Assembly Affects Image Quality)

The problem: Two "identical" assemblies used with the same CCD camera showed a consistent color shade difference. Measurements of the products indicated no difference between them.

The solution: Upon further investigation, it was discovered that one of the assemblies used a customer-supplied adaptor built with a flex circuit while the other did not. When the customer provided the flex circuit adaptor, it was noted that 1 ground connection was used for 6 signal conductors, as compared with the other assembly that had 6 grounds for the 6 signal conductors. We recommended the customer to modify the adaptor to fix the grounding issue.

The result: The customer instead decided to use a shorter cable to minimize (but not remove) the shading effect.

>> Return to Top


Change The Design If You Can (LVDS Cable Yield)

The problem: For several months, a cable using unshielded differential pairs exhibited a poor yield of the rise time to support the required data rate. The root cause was assumed to be in the construction of the pairs. The manufacturing group performed several design of experiments and none were able to identify the defect in the pairs.

The solution: A cable was methodically dissected and in 1 week, the root cause was identified to be in the process of cabling the pairs, not in the process of making them. Design changes were recommended.

The result: Initially, management deemed the design changes to be more expensive than the yield loss. Manufacturing later implemented some unrelated improvements to the cabling process and the yield fell to 0% (all products failing). Once the recommended design changes were implemented, the yield rose to 100% (all products passing).

>> Return to Top


Blame The Supplier First? (Increasing Data Rate On A Ribbon Cable)

The problem: A customer started using existing ribbon cables at faster speeds than they had previously used. As production of the customer's new product started, a few of the cables we supplied would work at the faster speeds. Further confusing matters, the customer thought that a recent (unrelated) change to the manufacturing process was the cause and that pre-change products worked at the faster speed.

The solution: Measuring samples before and after the change did not indicate any significant difference and the customer eventually discovered that pre-change products were failing at the same rate as post-change products. We identified the root cause and recommended a construction change to solve it.

The result: New samples were made within a day that outperformed the original product by a significant margin. The customer was delighted to receive better product at the same cost.

>> Return to Top


An Interaction Causes Problems (Noisy Doppler Ultrasound)

The problem: A cable used in a Doppler ultrasound cable would become significantly noisy as the cable aged. Every parameter measured on the aged cable was within specifications. However, the capacitance changed over a larger range as compared to a brand new cable.

The solution: The customer was given six brand new samples that had been aged a different amount (from no aging to beyond expected usage) and asked to rank their performance. All six samples were free of "noise". The defect observed was an interaction between the cable and the electronic amplifier it supplied. The customer was advised to compensate for the negative input impedance of the FET follower amplifier in the system.

The result: The customer followed the recommendations and the noise disappeared.

>> Return to Top


A Picture Is Worth A Thousand Numbers (Identifying The Root Cause Using A DOE)

The problem: A Six Sigma Black Belt organized a 16 trial experiment to understand the sources of variation of a coax parameter. The analysis indicated too much variation but not the root cause.

The solution: The data was analyzed using two different graphs not used in the DOE analysis. This graph identified the families of variation and clearly illustrated the best process setup to minimize the variation.

The result: The engineer was able to find the source of the largest family of variation and reduce the variation by a factor of three.

>> Return to Top


Measure The Problem In Its Environment (Motion Induced Triggers)

The problem: A Logic Analyzer cable used to monitor digital faults would falsely trigger when it was slightly moved. This occurred in product that was made during 1 month in the year, several years after the product had been in field use with no false triggers. Many complex root causes were proposed.

The solution: Measurements made on the defective product produced motion induced noise much smaller than would trigger the Logic Analyzer. Measurements made simulating the usage in the Logic Analyzer indicated a different noise generation mechanism that would trigger the Logic Analyzer. An unexpected root cause emerged as a result of analyzing the mechanism.

The result: Recommended changes were implemented and no failures have been observed in the subsequent three years.

>> Return to Top


Mistake-Proof the Measurement of Defects (Measuring DC Leakage)

The problem: When measuring dielectric breakdown, a large voltage is applied across the electrodes. If there is a breakdown, a large current will flow. If there is no breakdown, the current is small. If the product is not connected properly, the current will also be small. Therefore an improperly connected failure will not be detected.

The solution: Measure the current during the ramp-up time. The current must be commensurate with the calculated inrush current. This approach requires no additional connections or very accurate inrush current measurements while maintaining a low leakage current sensitivity.

The result: The Company has proof that the products were tested appropriately and no false acceptance of defective products is possible.

>> Return to Top


Make Good Measurements (Protecting Air Force One From EMP Damage)

The problem: In the 1970's, Air Force One was going to be replaced with a new 747. Kelly Air Force base was in charge of the project. Underneath a huge wooden trestle sat a 747. The trestle contained an antenna. A huge bank of capacitors were charged in parallel and then discharged in series through the antenna to simulate the electro-magnetic pulse. Everywhere in the site there were hundreds of sensors each connected to an oscilloscope with a camera. The oscilloscopes were triggered with the pulse and the cameras recorded the event. Not all the cameras worked all the time due to human setup error. The film was developed and hand digitized to feed a computer that would do the analysis. Typically, it would take at a long time before another experiment could be set up. Tektronix was approached to use a new transient digitizer that would capture the event, digitize it electronically and send it to the computer in seconds, not weeks or months. The specification called for the second harmonic distortion of a test circuit to be 20dB below the fundamental. All attempts to meet this specification failed.

The solution: Because of the methodology used to capture the event, the waveform was obviously distorted by beam spot blooming. The assumption was that the blooming was causing the distortion so software was developed to correct the trace for it. However, it was pointed out that blooming could not generate second harmonic distortion. Simulations proved that another type of distortion would be enough to generate a second harmonic at -7dB. The digitizer software was then designed to correct for both distortions.

The result: When Kelly Air Force Base was informed of our discovery, they had to request additional Congressional appropriations to correct several years of data distortion that was produced by the previous manual oscilloscope-camera-film digitizing methodology. They bought many transient digitizers units. Cycle time was reduced to less than ¼ and 100% capture since the instrument was computer controlled.

>> Return to Top


Failing a Spec May Not Be the Root Cause (Arcing Traces)

The problem: Two instruments are returned from the field with a burned circuit board due to arcing between two traces. The traces normally carried had 400V AC across them. One engineer noted that the traces are 0.005" closer together at one spot than good design practices recommend. He decided that the traces needed to be cut and wires soldered to bypass the "problem" spot.

The solution: We noted that 0.005" variation was not enough to cause the defect by itself. Raw boards were tested by applying a voltage to the traces until they arced. They arced at around 1,500V. The same test was performed using an assembled board and found that it arced at 1,100V. When humidity was introduced to both boards, the bare board arced at 1,200V while the assembled one arced at 300V. Application of litmus paper to the assembled board found acidic contamination. Washing the boards in soap and water and keeping track of contaminants in the bath water eliminated the contamination.

The result: An expensive rework that would have not addressed the root cause was eliminated. The wash process was monitored for contaminants. No arcing failures were found again.

>> Return to Top


Is the Problem in the Process or the Parts? (CCD Dark Current)

The problem: A CCD design was the crucial transducer to a high speed digitizer. The data would be sampled using the fast clock until the CCD memory was full. The memory would then be clocked out at a slow speed commensurate with an inexpensive A/D converter and store it on more permanent memory. However, the dark current measured on the CCD was too large to retain the charge long enough for the slow clock to move the data out without significant distortion. The engineers could not devise any way to improve their process to reduce the current by about a factor of 1000.

The solution: At a design review, process engineers were interviewed. Since the process was similar to other industrial processes that had acceptable dark current, the problem would not be resolved by improving the process. Instead the problem had to be defects in the equipment used for the process. Within a week, they found an old section of pipe contaminated with HCl and corrected it.

The result: CCD's produced after the correction now had acceptable dark current.

>> Return to Top


Understand Customer Usage (Probe Reliability)

The problem: A new line of wide band, small size oscilloscope probes was introduced with a new flagship product line. A short time after introduction, the probes were experiencing a high failure rate. The tip, designed to contact traces on a printed circuit board would break. The comment from the field was that the probes were inferior to the older, larger size.

The solution: It was found that the probes were not only used to contact the circuit board's trace, but to scratch away the epoxy conformal coating. The designer complained "I did not design them for that use!" The old probe was measured to find the breaking point. It was 11 lbs. The new probe was measured and it broke at 2 lbs. The engineer changed the material and the shape of the tip. Without affecting the size, the new design broke at 15 lbs.

The result: The new design was introduced to production in record time and samples given to the sales force. The customers were delighted with the improved performance and the fast response.

>> Return to Top


Project Leadership (Responding to the Competition)

Role: Project Leader

The problem: A new generation of products was introduced after several years of development. Within a few months, our main competitor introduced a superior product at a lower cost, therefore reducing the expected sales.

The solution: A team was formed to respond to the challenge. A decision was made to improve the performance in one area 1.5 times better. At the same time we would reduce the cost by 30%. A fast introduction was important.

The result: By reusing other designs and changing only what was necessary, the performance and cost objectives were achieved and the product was introduced within a year. The product was removed from the catalog after 17 profitable years. In contrast, our main competitor abandoned this market segment within 3 years after the introduction of my product.

>> Return to Top


Time to Market (Teamwork Develops New Product in Record Time)

Role: Group Leader

The problem: The Company was making highly profitable analog storage oscilloscopes. However, the measurement paradigm had changed. Instead of wishing to view the event after the trigger, the customers wanted to see the event that caused a fault. This required changing the technology from analog to digital. Two additional problems existed: (1) the design specifications would typically change frequently, often causing project delays; (2) the typical development cycle was 5 years (if the product did not get cancelled before introduction).

The solution: The team proactively attacked the "Creeping Feature Creature" by developing a thorough Functional Definition. We committed to have the product available in 1.5 years in accordance to the Functional Definition. If a change was proposed, the team would issue a statement of the additional time required. Since time was paramount, no changes were proposed.

The result: The first major milestone was celebrated by trimming the beard of an engineer that believed management would cancel the product. He loved it! The product was introduced within 1 week of the promised date even though the team had to solve an unexpected problem with a major redesign. We celebrated the success by attending the introduction show. The product immediately won the approval of the customers and orders exceeded the forecast by 200%

>> Return to Top


Changing the Culture (Manufacturing Cost Reduction)

Role: Group Leader

The problem: The Manufacturing operation was moved 25 miles away from the main plant. All of the products were under cost pressure by foreign manufacturers. The company culture preferred Design engineers over Manufacturing engineers.

The solution: We created a new group called "Profit Engineering" with the intent of making major cost reductions by improving the design tolerances. We hired top notch engineers for the group and purchased the same CAD tools that Design Engineers used.

The result: Actual cost savings during the first year were 10 times larger than the department budget.

>> Return to Top


Anticipate What Your Customer Wants (Product Strategy Development)

Role: Division Engineering Leader

The problem: The prevailing market segmentation strategy was the critical performance specification and price. This did not forecast how the products were used by the customer or how many were needed.

The solution: We developed a two dimensional segmentation of performance and size of the device being tested. This defined how the customers were using the product and what the mechanical constraints were.

The result: Our market share increased. Competitors continuously introduced products to imitate our introductions.

>> Return to Top


Reduce Measurement Variation (Automate Measurement System)

Role: Test Engineering Leader

The problem: The company's product required sophisticated electrical measurements to assure the products met the customer's requirements. Some of the measurements had the potential of creating unsafe situations for the operators. Other measurements would accept defective product if it was not connected properly. The unskilled operators were very creative in how they measured the product thus generating excessive measurement variance.

The solution: The different measurement systems were automated using a common platform. This standardized the test methodology and setup. Unsafe operation was eliminated with a new fixture design. Confidence tests were utilized every time a new product was tested assured appropriate connection and performance of the test equipment.

The result: Measurements covered the range from DC to 40GHz. Signals from 10kV to 100nV, resistance from 10mOhm to 1TOhm, characteristic impedance repeatability to 0.05 Ohms, rise time and time delay skew to 0.1ps. No safety accidents or false product acceptance occurred.

>> Return to Top


Predict What You Will Build (Models of Parts and Processes)

Role: Senior Engineer

The problem: New products were designed using approximate formulas and the design centered by making products by trial and error until the characteristics met customer requirements.

The solution: Simulated electronic products using PSpice to understand tolerances required to produce what the customer wanted. Simulated processes using Excel and equations to determine material and process parameter tolerances required to build the desired product.

The result: Tolerance of components and process variables required to meet unusual customer requirements were understood and predicted. Predicted performance was within 10% of actual. The models were used to identify potential tolerance issues in respect to customer requirements.

>> Return to Top



Contact us today to see how LuNa Consultants can help and/or train your organization.