Project Case Studies
Audio Electronics: Test Systems
Medulla was recently commissioned by Audinate to develop the system control software for the testing of a new Dante product.
Audinate required a solution that could test multiple variants of the product family while supporting multiple test configurations. Medulla worked within a cross functional team, led by Audinate engineering and including Danalog who was commissioned to design and build the test fixturing and hardware.
This was not a trivial system to design, for either the hardware or software as it incorporated a rich set of switching operations to support all the different test configurations.
The Medulla ViPER test executive met the critical system requirements: A user interface that configures dynamically based on user access rights (Operator vs. Engineer) and a debug mode with soft front panel access to all instrumentation, allowing test engineers to view or interact with the test instruments during runtime.
The system was delivered on time and to budget. It has since been deployed to a manufacturing site in China where production volumes are increasing at a significant rate. Additional systems are now being assembled for deployment to manage the high volumes.
“I worked with Kurt from Medulla on a test jig design and implementation for a new product line at Audinate. This was a complex project that involved multiple parties besides Medulla and required various adjustments as the requirements and the design firmed up along the way. Medulla was responsible for implementing the test cases. Kurt showed a high level of competence in NI LabView, great communication skills and most importantly flexibility. Kurt has also provided ongoing support. The test jig is in production and plays a key role in ensuring Audinate can deliver a quality product to its customers every time.”
Director of Product Delivery
Telecom: Test & Calibration time reduced by 40%
Medulla was contracted by a global manufacturer of optical telecommunication products to optimize an existing system, designed in-house by the client, that tests and calibrates a new product.
The process to test and calibrate one device takes 3 hours. They developed a system with the capacity to test up to 8 devices in parallel. However, the system couldn't cope with the load. Performance degraded with each additional device, limiting the client to a maximum of 4 units at a time, and taking over 5 hours to complete a test cycle.
This limited capacity and longer than expected test time was causing significant bottlenecks at their manufacturing facility. With production volume set to double, this bottleneck was a serious concern.
Medulla worked closely with our client’s optical engineers and software developers to understand the system's intended design & function. It is a highly complex piece of software with over 6,000 separate subroutines and a parallel architecture.
Medulla was able to resolve the issue well within our initial project estimate. Test time of a system loaded with 4 devices was reduced by 40%, from 5 hours back to the expected 3 hours. In doing so we also uncovered and resolved other issues related to memory leakage within the system. The fix should now allow the client to build up the system to full capacity and test up to 8 devices in just 3 hours, with a potential x4 increase from current capacity.
This result was well received, particularly by management overseeing the manufacturing operations. They were preparing a Plan B, to build multiples of test systems (>$100k each) to support increased capacity. They are now able to maintain the planned number systems and re-purpose the hardware purchased on Plan B to other projects.
The ROI of this project was immediate.
Medical device manufacturing: Process automation
Medulla worked with a large global medical device manufacturer to automate and streamline a process step in the assembly and calibration of a high-end implantable device.
Our client had a problem. Due to the variability of performance inherent in electroics, each device they made required a unique set of components, ID resistors and tuning capacitors to ensure every device performed consistently to a very tight tolerance. The process at the time involved loading carriers with blank circuit boards (24 at a time) onto a conveyer belt that fed into a conventional Pick & Place machine and reflow system in order to position and solder all of the miniature electronic components to the circuit board, but omitting those components that required individual selection. Each board would then be assessed by a calibration system to determine the value required for those select-on-test components. An operator would take the print-out from the calibration system, locate the components and manually solder them in place under a microscope. One device at a time.
This was tedious, slow, expensive and had a relatively high level of failure. This inefficiency was costing our client about $700,000 per year.
There is a better way. Medulla worked closely with the manufacturing team to revise and automate the process. The goal was to re-use the pick and place machine for attaching the missing components in a second-pass run after calibration. The challenge was to identify each of the boards and dynamically inform the pick and place system which components were required for each device (i.e. a dynamic placement file)
The client first modified the calibration system to store the calibration data to a cloud server, rather than print the result for each device. Medulla developed a machine vision system that identified each of the 24 implants in a carrier as it rolled past on the conveyer belt. Once identified, Medulla's system would pull down the calibration results for each implant and compile a specific placement file representing those 24 units, sending it to the pick and place machine.
This project has resulted in a highly significant cost saving to our client and a very good ROI
Medical Device: Modern test architechture
Medulla* was contracted by a large global medical device manufacturer to upgrade an existing system that tested circuit boards prior to being assembled into one of their implantable products. The test hardware and control software in the current system was ageing, most of the equipment had become obsolete and hard to maintain.
Our client had a variety of test systems in use company wide for performance testing and quality control across numerous product lines. Each system had been developed by different consultants with little consistency across systems. They had a critical mass of systems that were becoming an incredible burden to maintain and keep up to date. With a significant number of new products on the company roadmap they needed to adopt a more systematic design architecture that could unify maintenence procedures and, where system testing between products was common, utilise the same pre-verfied code base.
Our ViPER test architecture was designed for exactly that purpose. We rebuilt the legacy test system using the precursor to our ViPER test system architecture.
It became the gold standard for the company.
They have used this design in the development of all subsequent test systems supporting a wide range of new products. Leveraging and re-using pre-verfied test modules across systems shortened the development cycle and greatly reduced the V&V burden. Using the Medulla framework, the company now develops all of their test systems in-house, no longer dependent on external consultants.
Rail #1: Renewal of obsolete design
Medulla was contracted by a global leader in rail maintenance to re-design and supply a critical sub component of a train braking system.
The problem our client was experiencing is that the existing supplier of this sub component had discontinued that product and our client was quickly running out of stock. With 15 brake systems to service every month it was a serious problem that would impact their business and disrupt their client.
The sub component is a PCB that manages the regulation of pressure in the braking lines, a very critical piece of hardware.
Our team was able to examine the existing PCB (without schematics), understand its function and develop a new design using modern rail spec components and a much more rugged physical footprint.
We subjected our PCB to harsh environmental and electrical testing, qualifying it to the applicable rolling stock standards:
- BS EN 61373:2010 (Shock and vibration)
- BS EN 50121-3-2:2016 (Electromagnetic compatibility)
- BS EN 50155:2017 (Electronic equipment)
The boards were delivered on-time and within cost. They are in the process of being deployed across a fleet of trains serviced by our client.
Rail #2: Real time in-situ testing
Medulla* was contracted by Downer EDI to develop a high-speed stress and strain logging system to assist in the commissioning of the Waratah train carriages for Sydney Rail.
The carriages needed to undergo a series of live trials to ensure the stresses, strains and deformations experienced were all within design limits.
To accomplish this Downer EDI engineers placed 60 strain gauges throughout the carriage. Our task was to develop the logging system that simultaneously sampled all transducers and streamed it to file. We also developed a tool that allowed the engineers to perform fast waterfall analysis of large data sets.
The system was delivered on time and on budget. Data collected and processed assisted Downer EDI in achieving their test and commissioning goals.
“As part of the Test and Commissioning process for Sydney’s new Waratah trains, Downer Rail engaged Kurt to develop LabVIEW software to efficiently acquire, store and post-process large volumes of dynamic strain gauge data from 60 channels using the compact RIO hardware. Kurt's extensive experience in software engineering, object-oriented programming and problem solving skills resulted in the delivery of a sound and well designed software application. His professionalism, initiative, ability to meet project requirements, offer improvements and provide valuable training to staff has left us with great confidence in his abilities and in the end product.”
Senior Design Engineer – Rolling Stock Structures
*project conducted by Medulla co-founder Kurt Friday while trading as Sciware Pty Ltd. All IP owned by Medulla Pty Ltd.