Quality control within manufacturing boils down to how much money you are willing to spend to make sure there are as few problems as possible. One of the keys to success is being able to track a few data points that will allow you to anticipate problems before the product is shipped.
Most companies aim for a 0.5% failure rate up to a maximum of 3% failure rate - anything above is an “epidemic ” and triggers recalls for manufacturers.
The obvious answer is to spend as much money as possible to make sure there are no problems right? Big brands such as Apple, Microsoft and Sony would certainly agree to this tactic however would you, the consumer, be willing to spend “premium” for every product purchased?
Probably not, I certainly wouldn’t want to spend £50 for every cable I purchased, I would be willing to spend £10 - £20, cables are just cables at the end of the day.
How about cheaper products - how can they get away with being so cheap?
The cheapest product will have little to no quality control in place and will just be produced, hoping that the batch is “fairly good” and consumers would just purchase another cable because it’s so cheap.
So if you spend too little money, then the batch will have more failures and if you spend too much money, nobody will be willing to purchase the product because it’s too expensive.
What’s the balance?
This is the problem that every Quality Control manager and Quality Control Management System hopes to answer. “What is the perfect balance between too much testing and too little testing?”
To answer this question you always need to start with getting some core data. These Key Performance Indicators (KPIs) will act as a great start to get insights into your production.
This blog explores a few Key Performance Indicators (KPIs) that you can expect to track within factories that do or do not utilize good traceability systems. If the batch can not be traced within their system for any reason, these KPIs can be requested before the start of production.
In principle, all factories should have good traceability however there are times where this can not be accomplished for example:
The batch is too small
The traceability system is not advanced enough
Modifications to the system would take too long
The additional time add a significant cost to the final product cost
Question:
What if you’re about to start production and the factory doesn’t have a good traceability system?
Answer:
You can sudo create one by setting a few important KPIs and asking for the data upfront.
Question 2:
What if the factory doesn’t want to do it or you’re afraid that they might make up the data?
Answer 2:
Go to China and collect the data yourself; this might sound like a crazy idea but many problems I’ve solved were solved by being present at the factory during the day of production. I had a translator ready and collected the data myself over the course of a couple of weeks. These weeks turned to months and these months turned to years but that is a story for another time.
So what do I track?
The First Important KPIs to ask for:
“What happens when you first receive these sub-components or sub-assemblies, and what are the inspection tests?”
I.e Incoming Quality Control pass/reject rates of incoming components
Every product has a factory where the “Final-Assembly” takes place. At these factories, I ask “What happens when you first receive these sub-components or sub-assemblies, and what are the inspection tests?”
This is a great indicator because it tells you if the factory is inspecting goods as it arrives at the factory. Many avoidable quality control problems can be detected before the batch is ever produced by sampling components, running basic measuring tests, and having a basic pass / reject mechanism. If a 1mm change can heavily affect your design, recommend to the factory to measure those points of the design.
The same applies to critical electronic components. If you need transistors to be at a certain parameter for an adequate safety margin - ask the factory to randomly select some from the batch and measure the component values. If any of them are not up to scratch, either carry a full inspection of the products or reject the batch to the original supplier. Mistakes can happen anywhere along the supply chain - it’s your job to detect all of those mistakes.
If you are not sure what to measure, first decide what are the critical parts of your design and why they are critical to your design. Will it heavily affect the features of your product? Is it serving a part of your design that protects the user? If yes, then these are critical parts of your design
Mechanically, you should take into account natural deviations in the sub-assembly dimensions. If parts of your design were to +- 1mm, does it cause interferences or will it not affect your design? A lot of the time, tolerances are pre-agreed and you would build this into your mechanical design. If there are interferences then you should request for these parts to be measured by hand at the factory before assembly. If it deviates too far then the batch should be rejected.
If the batch is failing your first set of tests then they will fail when you put the whole product together. A 50V capacitor measuring at 40V will still measure at 40V when you assemble your product together. If the batch is good, it will be built to your specification and pass the above tests.
This forms a simple, yet effective, test for your factory to cover. This can then be added as a vital activity for the factory to carry out. My favourite part of incoming quality control is that you don’t need to rely on the factory to have these answers, the engineers who have designed the product will have these answers. Any good product designs will have identified critical components to measure, created test criteria, and will have designed test jigs to be used at the factory.
All of this information comes to light by asking “What happens when you first receive these sub-components or sub-assemblies, and what are the inspection tests?”
The second KPI stems from the question
“What is your First Pass yield rate on PCBs?”
This is an important question because you might be finding that your products are of good quality however it takes a surprisingly long time for the product to be produced or more costly than expected. This could be due to on-the-fly re-works occurring as soon as Printed Circuit Boards (PCBs) are placed through their Surface Mount Technology (SMT) oven.
This process is when you’ve placed the components onto your circuit board, usually by a pick and place machine which is then moved into an SMT machine which is effectively a large oven. The high temperature of the oven heats up the solder on the PCBs which allows components to mass soldered onto the boards - creating your PCB sub-assembly, ready for board testing; this electronic test tells you if the components are properly mounted onto the board and have electrical connectivity. The number of PCBs that pass the first it is tested gives you your “First Pass yield rate” on PCBs, given as a percentage.
A lot of different factors can affect this KPI:
Storage conditions of the Solder used
Type, quality and consistency of the solder paste
Maintenance of the PCB machine and how often maintenance is carried out
Quality of the batch of components ordered
Design problems associated with poorly placed components
Dimension issues with individual components
I once came across a 97% failure rate - yes 97 out of 100 PCBs failed the first time test because a button that we ordered varied by 1mm because the supplier thought this would have been acceptable. The solution was easy, resolder one of the legs of the pin by hand. The final-assembly factory didn’t see it as a major problem and continued to run this for two full weeks before reporting this problem. However, once we added the extra time this process took (roughly 30 seconds), it mounted to a $3 increase in the Bill of Material (BOM) cost/cost per unit; roughly 5% of the total cost of the product.
Remember that any re-works, “small/quick/easy fix” always add more time spent producing the product. Time spent on product range from 4 minutes to 12 minutes, with most factories spending 5 - 6 minutes on average. So even a 30-second increase can have a significant increase in the cost to your product.
A $3 saving times 10,000 units is $30,000; small savings, identified early will save money over the course of a product’s life span which can be anywhere between 100,000 to 500,000 units = $300K to $1.5M which is enough to fund a major research & development team!
Therefore first pass yield will be a great indicator so that you can start investigating on your side where these improvements can be made. Typically factories accept first pass yield issues and correct them without notifying the customer. By being open and shifting the focus to using this metric you can gain a lot of insights, especially early on in your production cycle.
Using this data you can improve your design which will save on your product cost. Simple component changes can make very big differences to your end cost by reducing the amount of re-work on products; all of this is gained by asking “What is your first pass yield rate on the PCBs” and tracking it.
The Third KPI stems from...
All great factories have one thing in common - they focus on data and the training of their factory workers. Ask “How often is training carried out by the factory? “
Most factories carry out training once per year - usually in February, straight after Chinese New Year. During this yearly festive holiday for China, most people go back home to their hometowns to visit families and friends. The most recent generational change is that 60% of Millenials and Gen-Z workers switch factories or their jobs during this holiday which has caused a new Quality challenge to factories. Their labour workforce is relatively new and requires re-training so that the quality of production does not fall in March.
Great factories have identified this and actively manage training - with dedicated training time before every new production run, spending anywhere between 3 days to a full week dedicated to building products, analyzing problems and optimizing workspaces.
The best factories publicly display a “Badge system” where workers have undergone specific training, earning them badges in that specific area of a product; they can then be confidently placed onto those assembly areas of the product.
By asking this question you open yourself up to some key insights. Factory quality and lead time variance based on time of year and how seriously the factory takes training into account; this level of dedication is usually carried forward into all the elements of the factory’s operation. If they take training seriously then they certainly care for the Quality of your product.
Fourth and Final KPI
A final and fourth KPI you would get by asking “What are your Final Quality Control tests before shipping the product and what are your typical FQC pass/fail rates”
Here you are asking the factory what their defect rate is when the products are assembled and to carry out an additional test before shipping.
It is very easy for a factory to finish production and immediately ship your product without sanity checking the batch again. Why would they? It takes more time and if there is a problem, it will take a lot of time and effort to fix an entire batch of products.
The way to think of this is, this is your final defence before the customer receives your product.
Leave the product fully packed into its pallets at the factory for a couple of days. Then get an inspector, or yourself, to randomly select units from the batch and carry out a full product feature test.
Depending on the batch size, a set of products are randomly selected from the shipment and inspected. The Acceptable Quality Limits (AQL) are 0% Critical, 1% Major and 2.0% minor issues. If the shipment is found to contain more than this number of issues, then the whole batch is subject to a full inspection; this increase to cost is agreed in the Manufacture Supplier Agreement to all Ardencraft Technology suppliers. Industry-standard for consumer electronics are 0% Critical, 2.5% Major and 5% Minor.
A 'Critical Defect' corresponds to a defect that is likely to result in unsafe conditions for end-users. Accept if there are ZERO critical issues and reject if there is a CRITICAL issue then stop the shipment and investigate the issue. Critical issues are the main reason for product recalls because it poses unsafe conditions for the end customer.
The 'Major Defect' corresponds to a defect that reduces the function of the product, or reduces the usability of the unit of product for its intended purpose or specifically requested by customers.
A 'Minor Defect' corresponds to a defect that is not likely to reduce the usability of the products for its intended purpose or is a departure from established standards having little bearing on the effective use or operation of the unit.
The core idea of this KPI is to identify if the factory carries out this final check before shipping or whether they ship the product straight to you or the end customer. We’ll write another blog post that dives deeper into this section.
Summary
With that in mind, hopefully, you are now armed with four key questions to ask your supplier that provides you with a set of easy KPI to track, knowing that the KPIs provide valuable insights into your product’s health. For high-quality products, there is almost an infinite number of steps and precautions that you can take but it all boils down to how much customers are willing to pay for the product and how much risk is associated with producing your product.
The four questions are:
What happens when you first receive these sub-components or sub-assemblies, and what are the inspection tests?
What is your First Pass yield rate on PCBs?
How often is training carried out by the factory?
What are your Final Quality Control (FQC) tests before shipping the product and what are your typical FQC pass/fail rates”
This will give you the KPIs of:
Incoming Quality control pass / fail results for critical components as a number per batch, per product and as a percentage
The first pass yield rate of PCBs per batch, per PCB, and as a percentage
Number of hours of training carried out by the factory for your product
Internal Quality defect rate per batch before shipping as a number per batch and percentage
The end goal is to get insights into the common faults experienced by the factory during production so that you can improve your designs for future production runs by identifying wasteful processes.
Quality Control is at the heart of what Ardencraft Technology does and why so many customers trust us with the production of their products. In this blog, we talked about some of the high-level Quality Management System Planning. We strive to be experts in this field and can provide valuable insights into your product during products. We provide design services that ensure products are made to high quality and at a reasonable cost.
Feel free to leave a comment or get in touch with us if you are looking to develop a product, need to improve the design of a product, or simply have a question on anything written in the blog.
- Your Friends at Ardencraft
Comments