Survey results on the causes of power supply failures due to magnetics.
Introduction
In this third article about power supply failures, the magnetics are examined for their contribution to the failure rate. These are usually the least understood of all components, and poor magnetics design can lead to many different failure mechanisms in power supplies.
Power Supply Failure Survey
The LinkedIn site “POWER SUPPLY DESIGN CENTER” [1] is a valuable source of design information with over 3500 members contibuting to discussions. We surveyed the group members on “Why do power supplies fail?”, and the results of the survey are repeated in Figure 1 below.
Figure 1: Survey Results for the Cause of Power Supply Failures
From Figure 1, the semiconductors were seen by the survey group as the main cause of failures, and this was discussed in the first article of this series. Second on the list is capacitors, discussed in the second article. Magnetics are key to succesful operation of a power supply, and these are next on the list for causing failures.
Magnetics Failures
Magnetics are the key component in a power supply. Without proper design and manufacturing of these parts, a power converter can be unreliable, inefficient, or it may simply not work. It is usually the least understood of all components in a power supply and simulation models are woefully inadequate. Very few engineers receive any formal training in design and testing of magnetics leading to many misunderstandings and mistakes being made.
Figure 2: Survey Results for Causes of Magnetics Failures
Thermal Stress: 33%. As with capacitors, thermal stress is at the top of the list for causing magnetics failures. This can be due to excessive dissipation, lack of accomodation for high-frequency losses, or simply poor heat removal. Reliable magnetics design is very much a function of experience of the engineer. In this article, we thought it best share the opinions of our LinkedIn members to illustrate the numerous problems that are encountered with magnetics.
“It's interesting to note that most semiconductors are typically placed on a heatsink to remove the 1W or more of dissipation and keep the junction below 125C. How many magnetics do you know that are placed on heatsinks in order to remove the heat from the core and the windings? Some of these magnetics have a lot more than 1W of dissipation and they are just on FR4 boards. The thermal coefficient is over 30C/W. What do think the life of such magnetic is at 80C or 100C ambient? Thermal, in my opinion and experience, is the number one factor for magnetic failures in power application.”
Saturation: 24%. Second in the list is core saturation of transformers and inductors. Many designers these days are not familiar with proper test procedures of power supplies, and this leads to the omission of crucial current measurements. When saturation occurs, a power supply may continue operating, but current stresses are greatly increased. The best way to detect the saturation current is to observe it directly with a current sensor under a wide array of test conditions.
“I see saturation occurring in perhaps 25% of offline converters that I review. It usually happens during transients due to inadequate design margins.
For bridge converters, there are many reasons why saturation can happen – inadequate number of turns, poor layout, noisy controls, and so on. Once you see saturation, failure is usually just a matter of time. But if you don’t look at the waveforms, you won’t know the problem is there. ”
Figure 3 shows saturating current in an inductor. The characteristic sudden change in slope of the current is due to saturation of the core, and the resulting reduction in permeability. These waveforms should be avoided at all costs if you want to ensure a reliable power supply.
Figure 3: Inductor current waveform with saturation.
Mechanical Failure: 15%. Mechanical failure can mean many things for a transformer or inductor. Ferrite cores are made of extremely hard materials, but they are prone to cracking with mechanical stress. Proper handling and shock and vibration stress release can be key.
The following statement appears on a ferrite manufacturers website, and should be taken to heart in your manufacturing process.
“CAUTION: The edge of the surface of ferrite core is sharp. Minute burrs may be present. Ferrite cores are weak and prone to shock damage. Shock may cause cracking and chipping in cores. Inspect ferrite cores for cracks prior to use.
If ferrite cores are used without inspecting for cracks, deterioration of characteristics and heating may result.”
Since the performance of magnetics is very much dependent on mechanical arrangements of copper, tape, and cores, it is very common for poor design and construction to lead to mechanical failures.
Voltage Breakdown: 13%. Since transformers are the isolating elements in high-voltage applications, voltage breakdown is a primary concern. Voltage breakdown can cause catastrophic failures and safety hazards. The mechanisms of voltage breakdown can range from the very simple, involving inadequate spacing, to very complex, involving corona effects and material physics.
process.
“Some of the voltage breakdown cases I saw might have been triggered by core losses, where the enamel of the wire becomes so hot that the insulation it provides eventually fails. But hard to say because by that time the converter was long dead and everything was cooled off.”
Other: 8%. A multitude of reasons make up the last category. Primary under here is the quality of manufacturing. The modern manufacturing world will send magnetics design and construction work to the lowest bidder, and this can easily result in the obvious results for a component that requires much skill and experience to get right.
“The problem is getting the transformer manufactured correctly. I've seen many transformer houses swear they know how to do vacuum potting and HV insulation and so on only to have the part fail in seconds.”
“In my world of aerospace consulting, a lot of problems are caused by unrealistic size requirements from the customer coupled with the company insisting on using the "cheapest" outsource vendor.”
“I agree 100%. The biggest problem transformer houses have is unskilled labor, poor QC and inadequate testing to weed out failures. My previous company did mostly high voltage supplies and transformers were designed and wound in-house. Once they decided to outsource for cost savings, the failure rate increased dramatically.
Figure 4 shows a classic manufacturing error for a power supply transformer. The core has been gapped on only one side. This transformer will meet all of the electrical test specifications, but once in the circuit, a changing inductance will be observed even with light load.
The one-sided gapping creates an edge contact between the two core halves on the ungapped side, and this will progressively saturate as current is applied. While this may not be a problem electrically, and the control system may absorb the changing value, it is a poor way to construct a transformer. There is a lot of mechanical stress on the edge contact, and this can easily lead to cracking of the ferrite with time and vibration.
Apart from this, the transformer was well-built, no doubt a copy from some other vendor. The manufacturer simply did not have the experience to know that this technique is to be avoided.
Figure 4: Flyback transformer gapped on just one side.
In virtually every design of a custom transformer for power supplies, there are iterations with the manufacturer to eliminate all of the problems that may arise when large quantities are built. Close attention is needed when ramping up quantities, changing vendors, and tracking quality during ongoing production.
Summary
The survey results in this article highlight the major causes of magnetics failures in switching power supplies. There are many reasons why this often hand-built component can cause a power system to fail, and they must be well understood if you want to have reliable power supply products. Experience is key, both from the design engineer’s point of view, and from the manufacturer.
Very few engineers receive formal training of any kind in magnetics design, and learn the art piecemeal from hard-won experience. If you want to improve your skills in this area, Ridley Engineering offers a hands-on workshop where you build and test your own magnetics for power circuits [3].
References
- Join our LinkedIn group titled “Power Supply Design Center”. Noncommercial site with over 7000 helpful members with lots of theoretical and practical experience.
- For power supply hands-on training, please sign up for our workshops.