What is a Burn-In Test?
It is a sad but hoary truth that all things wither down and decay, that they are imperfect, fallible, and therefore liable to wear down, break, or malfunction in any of a myriad of ways. Knowing this, many companies subject their products to rigorous testing regimens long before those products ever hit the shelves. By anticipating the kinds of stresses that their merchandise is likely to encounter in the ordinary course of use — or even by subjecting it to greater-than-normal stresses — companies can extend the lives of their products, decrease average maintenance or replacement costs and generally improve quality. It’s all part and parcel of building a better mousetrap.
But what if what you seek to build is a tad more intricate than a mousetrap? What if it’s an advanced CPU studded with millions, or even billions, of microscopic semiconductors, or an LCD computer monitor or TV screen meant to treat its users to only the clearest and most vivid of displays?
Electronic devices have their own special set of testing standards. These standards emerge from the product quality goals that electronics manufacturers set for themselves as well as from whatever regulatory requirements exist in the country or area in which the product will be released. Some of the most important such standards are connected with what is called a burn-in test.
A burn-in test is a special type of procedure designed to detect early failures or malfunctions in electronic devices and allow manufacturers to correct them. This test can be performed either on the device as a whole or modularly — that is, on each electronic component of that device separately.
This process is called a burn-in because it typically entails subjecting the device to temperatures that are significantly higher than normal and observing how that stress affects performance. Below, we will give a detailed description of burn-in tests and of the various methodologies and bits of technical jargon that surround them. With this, we hope to clarify the meaning and purpose of this critical procedure, explain the pros and cons of its many variations, and acquaint you with a part of the electronics industry that often sits below the surface of attention.
The Burn-In Test: An Exact Definition
A burn-in test is a special type of stress test conducted on electronic devices prior to their public release which uses some combination of high temperature and/or high electrical voltage to determine whether the device being tested, or any part of it, is likely to fail or malfunction in the early stages of its product life.
Depending on the type of product being tested, they can be conducted according to a variety of different methods and protocols, but they generally last anywhere from 48 to 168 hours. Thus, the purpose of a burn-in for a device is to decrease as much as possible the so-called “infant mortality” portion of its bathtub curve.
How do Burn-In Tests Work?
The Bathtub Curve
Before we begin explaining the mechanics of the burn-in test in any detail, we must pause to explain what a bathtub curve is.
A bathtub curve is an an important tool in engineering — specifically, in deterioration modeling. It is a mathematical function which relates the failure rate for a given product — in this case, a semiconductor or other piece of electronic equipment — to the amount of time that that product remains in use. It is called a bathtub curve because it usually has a semi-ovular shape that somewhat resembles the outline of a bathtub.
This bathtub shape is created by combining together three component functions, each related to a general time interval in the product’s lifespan. The first function, sometimes called the “infant mortality” function, relates to the earliest stages of the product’s life. When time is 0, the infant mortality function generally starts off at a fairly high value, but then sharply decreases as the product moves out of its early stages of life. When time has moved to the middle and late stages of the product’s life, the infant mortality function levels off to a low value. This describes the fairly high number of early failures — due usually to various manufacturing defects — often seen in electronic components and devices.
The second component function relates to the middle stage of a product’s life and is meant to track random failures. Because such failures are random, they are impossible to predict on a case-by-case basis. Therefore, they are simply represented as some constant value across the product’s entire lifespan. This is depicted graphically as a horizontal line.
The third component function maps a product’s “wear-out failures” — that is, the failures that a product experiences as an inevitable result of aging components and of prolonged use. When time is 0, this function’s value also begins at 0, and it only increases slowly and very gradually as time travels through the early and into the middle stages of the product’s life span. Then, as we enter the late stage, the wear-out function’s value suddenly shoots up.
The bathtub curve’s shape — its downward-sloping early section, relatively flat middle section and upward-sloping later section — comes from each of the peculiar properties of these three component functions, as manifested in the respective time intervals in which they are most significant.
One of the main purposes of a burn-in test, therefore, is to acquire the information necessary to construct the infant mortality functions of various pieces of electronic equipment — and hence, the early sections of their bathtub curves. That information can then be used to decrease both the peak and the overall value of the infant mortality section as much as possible.
For those interested, a detailed discussion of the mathematics behind the bathtub curve can be found here.
The Mechanics of a Burn-In Test
Now, it is time to actually explain the facts surrounding the setup and functioning of a burn-in test. To conduct a burn-in test, four things are typically required: the device or component to be tested, a printed circuit board (PCB), a socket and a special burn-in oven.
The PCB is used to connect all of the relevant electronic components together and allow them to send signals to one another. Virtually every major electronic device made today has a PCB inside of it which performs this function. It is not possible to observe how the parts of a device interact during a burn-in if those parts cannot send signals to one another, so a PCB is obviously a critical part of the test. Even in cases where the burn-in is restricted to only one isolated component, the test will still require a PCB because testers need there to be some kind of electronic interface through which they may gather performance data as the test proceeds.
There are two ways to connect electronic components to a PCB and hence, to each other. One way is through soldering, which is permanent. The other way is to use a socket, which is a special type of connector that is meant to efficiently and compactly connect circuity in a small space.
There are various types of sockets, each with its own respective pros and cons. Which type happens to be used in any particular burn-in will depend on the kind of device being tested. The key thing to understand about them all, however, is that they are removable. This is obviously useful for testing purposes, as it makes it easy to remove, replace or tweak any components which fail the test without destroying the general device setup.
There are also certain sockets which are built specifically to be used in burn-in tests. They have peculiar features which it is best to explain in the course of discussing burn-in ovens.
The burn-in oven is the most important part of the whole testing setup. It is what actually administers the heat stress to the electronic device being tested. Testers place the device being tested inside the oven and set it to whichever temperature the parameters of the test require. The device is then kept inside the oven at the set temperature for however long the test requires. The device’s perfomance, as well as anything else that happens to it, are observed and recorded in the meantime.
Most burn-in tests use a temperature of anywhere from 125°C to 150°C, though temperatures as high as 180°C are not unheard of. The automotive industry, which is typically required to fashion the electronic components in its cars to higher standards, can run burn-ins at anywhere from 200°C to 220°C. Tested devices are continuously kept inside the burn-in oven at the set temperature for as little as 48 hours or as much as 168 hours. For some highly specialized radio frequency equipment built to especially exacting standards, a burn-in can last for up to 1,000 hours. However, this is highly unusual. The 48-168-hour range covers nearly all cases.
This setup is a fairly straightforward one, but as electronic devices began growing more complex, test engineers encountered snags that required them to minutely tweak the procedure in all sorts of ways. For one thing, semiconductors, even when used normally, emit a certain amount of heat. This heat will combine with the heat in the burn-in oven and may thereby subject the device to greater heat stress than was originally intended. This may not seem like much of a problem, but as the semiconductor itself is the source of the additional heat, it can have an outsized impact. Moreover, the data gathered during a burn-in is very precisely calibrated to many decimal places. Even small disturbances can throw the balance out of whack and render the results unusable.
Still other threats exist to this delicate balance. As they examined burn-in ovens more carefully, engineers began finding that heat had not always been equally distributed across every corner in the oven. This produced small variations in test results that depended on where in the oven some particular decide had been placed.
To counteract these difficulties, engineers decided to place special types of sockets onto the PCBs that had been stacked up in the oven. These sockets included features like heat sinks, fans and liquid cooling systems that cooled components down if they became hotter than the test required; heaters that warmed them up when they were not quite as hot as the test required; and temperature sensors that determined which of the above issues was happening when and at which location. This was all controlled through a feedback system powered by specially designed software.
The result was the idealized and even heat distribution required by the test. This then made progressively more and more accurate results possible.
Types of Burn-In Tests
The final thing to discuss in this connection are the types of burn-in tests. Broadly speaking, there are two types. One, called a thermal test, straightforwardly puts the tested device under some pre-established temperature in the burn-in oven for some pre-established length of time. The other, called an environmental stress burn-in, runs a semiconductor at some given speed at some given initial temperature. Then, it gradually increases the speed at which the semiconductor is run and the surrounding temperature. The test goes on in this way for some predetermined length of time or until some pre-determined temperature is reached.
Equivalent sorts of testing conditions exist for burn-ins which subject devices to electrical charges. Indeed, they are often tested under both high temperature and high charges. It’s also important to note that burn-ins are just one of the many kinds of quality and lifespan tests that electronic devices are put through. Some others include:
- In-circuit tests
- X-ray tests
- Flying prober tests
- Auto Inspection (AOI) tests
How Does One Arrange a Burn-In Test?
A explained above, companies usually conduct burn-in tests by taking the electronic devices that they wish to test and placing them inside of specialized burn-in ovens. These ovens have shelves of PCBs on them. The devices are placed onto the PCBs, attached thereunto by specially designed sockets and subjected to whatever kind of heat or electrical stress is deemed appropriate for whatever length of time is deemed appropriate.
Ordinary people at home generally do not have access to such specialized equipment. However, in some cases, even you can conduct a burn-in test on some devices in your own home. It is easiest to do this with computers. There are special kinds of PC stress testing and reliability software that you can download which will perform something roughly analogous to a burn-in test for your laptop or desktop computer.
The Origins of the Burn-In Test
Burn-in testing did not have any peculiar history or pioneers all its own but instead developed of a piece with the evolution of electronic devices themselves. As technology advanced and electronic equipment became more and more sophisticated, not only did governments around the world begin to enact testing and safety standards that all manufacturers had to comply with, but electronic engineers and hobbyists themselves began experimenting with their own hardware.
Thanks to these developments, people converged on a set of effective testing protocols. Engineers, intimately familiar with the physics of their hardware and its functioning, knew the kinds of stresses that electronics would have to withstand on a day-to-day basis, and therefore the kinds of bars they would have to clear in order to be consumer-ready. Specifically, they knew that semiconductors inevitably generated both heat and electric charges as they did their work, and also that said heat could affect the physical integrity and reliability of the silicon and metal components of which they were made.
Electronics companies, eager to stay competitive, needed ways to ensure that their products would be the best that they could make them. It is not difficult to detect the emergence of burn-in testing in the confluence of all of these facts.
As technology continued to develop, the resulting procedures would be refined as needed through trial and error — a process that continues to this day.
What Are Some Applications of the Burn-In Test?
Burn-in tests are used in product testing. Their purpose is to maximize the quality of electronic devices by decreasing their probability of early failure as much as possible. This is especially important for electronic equipment because it tends to suffer from far higher rates of early failure and malfunction than other types of products do. By weeding out the products that fail this test from the pool of products eventually sold to consumers, manufacturers can maximize product quality and hence, customer satisfaction.
Examples of Burn-In Tests in the Real World
Burn-in tests are everywhere in the electronics industry. Major global semiconductor manufacturers like Samsung, Intel and TSMC use them all the time, as do GPU manufacturers like Nvidia and AMD. Computer manufacturers like Apple, Dell, HP, Toshiba and IBM all use them, as do smart TV manufacturers like LG. Indeed, as burn-in tests feature prominently in the work of essentially every firm that creates any kind of electronic hardware, the full list of companies that do burn-in tests would be gargantuan.
On a smaller scale, electronics hobbyist might occasionally do burn-ins on their computers or LCD TVs.