Optimizing the Prototyping and Design of Custom Magnetics
This article highlights Bourns Inc. idea on power advantages gained in using Finite Element Analysis to identify the optimum winding order before prototypes.
“To meet application-specific feature specifications, many high-frequency magnetics designs must now be customized. What has proven successful for customization is where experienced engineering teams can supply both the software (magnetic design and FEA) and hardware (prototyping tools) portions of the design from one power electronics laboratory location. Being able to supply this combination of engineering talent provides time-to-market and configuration benefits, better supporting a customer’s initial converter prototype design.”
This article highlights the power advantages gained in using Finite Element Analysis to identify the optimum winding order before proceeding with physical prototypes. Furthermore, it determines how actual prototype measurements to simulations can vary. Therefore, having the ability to create prototypes in the same location as the simulation software is a critical requirement.
Initial Specification Case Study
The first steps in creating a magnetic component involve a preliminary specification of the power supply itself, including additional information – such as the topology – as well the manufacturer and identifying the power management chip series. A basic electrical block diagram of the system will indicate the number of windings involved.
Figure 1: Block Diagram of a Flyback Power Supply.
Figure 1 shows an isolated AC/DC 70W Flyback power supply with reinforced isolation. The control loop consists of secondary voltage feedback and primary current sensing. The coordination of the MOSFET and synchronous rectifier is done using the auxiliary (AUX) winding and the controller IC for measuring the demagnetization time.
In this case, the leakage inductance measurements will determine the time that the leakage inductance spike lasts and determine the cycle time of each oscillation when the secondary is completely demagnetized. Designing the transformer to ensure optimum leakage inductance with multiple windings will be imperative for desired performance.
Figure 2: Auxiliary winding voltage and synchronous rectifier control voltage (blue) in Flyback power supply.
If there is an application note from the power controller supplier, it can help determine the electrical parameters of the transformer, including the primary inductance and peak saturation current (in the case of a Flyback). For this case study example, the primary inductance will be calculated using the energy equation for a Flyback transformer:
Figure 3: Example Calculation for Creepage and Clearance.
fsw = Switching Frequency
Ip = Peak Current in the Primary
VF = Synchronous Rectifier Voltage Drop
The dimensions of the transformer will be determined first and foremost by the target power to be dissipated in the transformer and the specified operating temperatures. Also, essential is the customer’s board and enclosure that are strictly dictated by the safety requirements as stipulated by the customer.
Bourns in Europe uses Solid Works for mechanical design in which Figure 3 shows one example of a Solid Works transformer design. The blue line highlighted is measuring the shortest distance between a secondary SELV pin and winding. Solid Works supports its partners in helping meet safety standards such as IEC62368-1 Edition 2.0 2014-02.
Ideal Prototype Support
If a customer requires engineering samples urgently, then having all materials in stock is a clear advantage to help avoid delays. Typically, Bourns stocks more than 179 different shapes and sizes of MnZn Ferrite cores for new designs.
These cores are “un-gapped”, although custom laboratory machines can produce a flat uniform gap in less than 30 minutes. In addition, a Form-labs 3D printer is typically able to create bobbins of various types of plastic within four hours.
Optimized Simulation Software
The time-consuming trial and error in assembling and testing different variations can be simplified by first relying on tools and software, such as ANSYS, for identifying optimum structures. The Flyback transformer example in Figure 1 is designed for 12V /6A isolated output (SELV) with a non regulated, non-isolated 12V/0.18A output (NSELV).
Figure 4: Diagram of Winding Order for Three Different Scenarios.
Some controller manufacturers will have a maximum time allowed for the leakage inductance spike on the AUX winding. Figure 2 (in yellow) shows the AUX winding voltage, sampled by the controller. The peak-to-peak variation (ringing) will also have a minimum value and is dependent on the leakage inductance.
The coupling between the NON SELV in Figure 1 and AUX may also need to be controlled. This can be necessary for standby power situations with the NON SELV output being switched ON or OFF. The control loop stability could be affected in these situations without optimum coupling between the AUX and NSELV windings. Therefore, placing the NON SELV close to the AUX is necessary for this situation to maximize their coupling.
Figure 4 shows three different winding structures that have been analyzed by ANSYS with the leakage inductance shown plotted in Figure 5 and Figure 6.
Figure 5: ANSYS Finite Element Analysis.
Figure 6: ANSYS Finite Element Analysis.
The software will optimize the layout of the windings, but also allows for manual placement of the windings. Likewise, it allows for insulators, such as tape and margin tape, that can have an effect on leakage inductance. The isolation requirement between two non-isolated windings (500 Vac) is not possible if the windings are placed side by side, which would be more efficient. They have to be separated by at least one layer of tape. The spacing between turns can also be adjusted.
The primary to SELV leakage in a high-power Flyback with auxiliary winding using secondary regulation is halved by splitting the primary winding. This doubles the magnetic field path length and halves the magnetic field intensity.
Winding Order B provided the optimum balance between the Mains to SELV and AUX to NSELV leakage. Winding Order A has the lowest Leakage inductance Mains to SELV but had a higher AUX to NSELV leakage. Spreading out the windings actually increased the leakage inductance, despite the fact that the path length increases through this approach, hence lowering the field intensity (Ampere Turns Per Meter).
Increasing the distance between the turns allows uncoupled flux to pass into space between windings. However, there is a trade-off in the space between turns and the overall length of the winding.
Therefore, using margin tape to keep the turns close together was used successfully when making initial samples. The measured results confirmed that Winding Order B was the better option. These measurements demonstrated that spreading the winding across the bobbin had the opposite effect on the leakage inductance.
Figure 7: Physical Measurements of Prototypes made with Three Different Scenarios (80KHz HP 4285)
Made at 80kHz using an HP 4285 LCR
The differences in real and simulated measurements of leakage inductance can be partly due to the following factors:
A) Short circuit bar resistance
B) Distribution of coil along the surface of the bobbin
C) Tolerance in insulation material thickness
It is important to note that while simulation helps to compare different scenarios and select the appropriate winding structure, it is imperative to build a sample and test it thoroughly in software.
Power Laboratory Capabilities
Occasionally, customers might require support with testing transformers on application boards under certain conditions. For instance, Bourns has a license for Altium Designer for circuit and PCB design. The laboratory has a range of power sources and electronic loads together with a temperature chamber and infrared camera for testing boards. Bourns also assists customers with EMI board testing.
Ideally, custom magnetics suppliers should also have production facilities that are certified to IATF16949 with automated manufacturing both for high volume and low power transformers, along with more complicated magnetics assemblies. This includes high-power converters (toroidal or split cores) such as power factor corrected, soft switched half-bridge converters. Experienced application engineers are needed to ensure prototypes successfully transfer from initial engineering to production using industry standard Advanced Quality Control Procedures (AQCP) so the design can maintain the highest quality levels.
The most efficient power electronics laboratories are set up to support customers with design, simulation and engineering samples of high-frequency power magnetics. They also need expertise with advanced software design tools that will allow them to select the optimum magnetics design for the customer before making engineering samples.
Having mechanical and electrical engineering in the same location, as well as the available stock of ferrite cores and 3D printer allows for a quick, sometimes as fast as 24-hour, turn-around on engineering samples. While simulation tools save time by identifying the optimum design, there is still no substitute for testing actual samples and verifying results.
About the Author
Cathal Sheehan is a Senior Technical Market Manager at Bourns, Inc. He studied Electrical and Electronics Engineering at University College Cork and pursued his MBA at The Open University. He has worked with various companies that related to electronics.