App Notes
Magnetics Design for a Dual Active Bridge (DAB)
Optimizing transformer design for Dual Active Bridge converters with a focus on leakage inductance, core choice, and Frenetic’s simulation tool.
A design that can’t pass the isolation standard should be seen as a failed design. Design for safety standards is not trivial and can increase both R&D costs and complexity. But before starting to implement safety standards, we need to understand the basic definitions that are used in this context.
Let’s get the definitions straight, before moving forward:
Another way I see clearance and creepage is by thinking of high voltage as electrons that go from point A to point B, either by plane (clearance) or by car (creepage). In my opinion, the silliest analogies are usually the easiest ones to remember! Have you ever heard of the memory palace?
In Figure 2, we can see that the clearance between the primary (blue wires) and the secondary (orange wires) is the direct distance between them. However, since there is insulating tape, the two windings never face each other directly. The dielectric strength of a single layer of insulating tape normally used is about 4.5kV (e.g. Tecroll 11B), way above the voltage difference between the primary and secondary winding.
If the windings were touching the margin tape, the creepage and clearance distance would be the same and the double of the width of the margin tape. Just look at the tracking path, in black color. So, if for example the requirement is 6mm creepage distance, then we would use 3mm of margin tape in the primary layers and 3mm margin on the secondary layer, as shown in the picture.
Following a safety standard protocol is going to dictate how much creepage and clearance is needed for a particular transformer. A good rule of thumb for mains powered transformers like flybacks in USB PD chargers is 3 & 6mm of creepage required for basic and reinforced isolation, respectively. However, depending on the standard and some other operating conditions, those numbers can be different.
As mentioned in the description of Figure 2, the margin tape is inserted just to keep the windings a specific distance away from the side of the bobbin. You can think of it as an insulating spacer. The drawback of the margin tape is that it shortens the bobbin width, thus reducing the space available for the windings. For smaller cores, the margin tape could potentially take up most bobbin space. There is a way to avoid margin tapes by using fully & triple insulated wires, and that’s the approach we’ve used in today’s library design.
You can have a look at the full project here.
How did I choose the RM8 core (see Table 1) for this design? I checked the power density achieved in the previous flyback designs! Let’s see some numbers:
The first design was an EQ25 60W compatible for USB PD 3.0 charger applications, and that design managed to achieve the highest power density. But problems with the bobbin availability made that design hard to manufacture. Use of flying leads and core insulation requirements led to the next iteration.
The second design (ETD29 - iteration of EQ25 design) achieved a lower power density. Essentially, power density was sacrificed to make sure that a reinforced isolation, standard core, no flying leads, no core grounding considerations were needed, this way addressing all the drawbacks of the first design.
The third design is a 25W similarly speced flyback transformer that employs an RM8/I core. The transformer passes basic insulation, avoiding the need for flying leads, but needs core grounding. It’s in a sense somewhere in between the first two designs, as far as drawbacks go. Power density-wise could probably be better, but the difference is actually minimal.
Using a simplistic approach, and realizing that the creepage requirements are similar in that design, I’ve made the decision to go for a power density somewhere between 3.5-5kW/L. The idea is that the higher the power density, the more compact the design is, leaving little wiggle room for isolation needs.
Remember that choosing a power density goal is entirely up to you. You may disagree with my approach here and suggest a different way to choose the power density goal, but please remember to include isolation needs in your approach. Not just textbook equations…
I would be more than happy to receive some feedback on that!
There is no margin tape in this design, as I mentioned earlier. Notice that only the secondary winding is chosen to be a triple insulated wire (TIW). That way, creepage and clearance specs are not applicable between windings. However, they are applicable between the bobbin pins and core! Remember the core is made of ferrite, which is considered a conductive material in compliance protocols. The distances of the pins to the core allow for a Class I transformer with basic insulation.
The winding arrangement is always a crucial step for any designer. The solution that Frenetic provides is simple enough, though. I’ve copied my original project two times and changed the winding arrangement with just a few clicks. Using the comparator option of the platform we can find the best solution:
Although power losses and hotspot temperature are similar out of the three iterations, V0 is the best achieving 3.91uH of leakage inductance*. I wanted the leakage to be below 5uH to minimize power losses in the RCD snubber filter of the flyback, increasing the overall flyback efficiency.
*Leakage inductance here is the primary reflected leakage inductance of the secondary winding.
Let’s get now into detail about how to select the appropriate distances in a transformer design. The goal is to decide exactly how much creepage and clearance is needed for the flyback design that I want to build for our library designs.
Long story short, I needed a flyback auxiliary psu a while back for a big project of mine. The input voltage was about 40-50V and the output 12/8.4V. I had no idea back then about safety concerns so, I built the transformer, and everything has worked great until today. The reason why it’s still working is the low input-output voltage requirements. Years later I found another application where my little flyback card would work just fine, if I changed the input voltage to about 450V. Owning the rights for this design I decided to freely share aspects of the actual design with all of you, starting with the spec sheet shown in Table 3.
The main insulation categories are functional, basic, and reinforced (or double) insulation. Functional insulation only accounts for a bare minimum insulation that can only guaranty functionality of the transformer. Basic insulation provides protection against fault conditions, such as overvoltage conditions at the primary transformer side. Reinforced isolation is the strictest option, providing maximum protection against fault conditions, making sure that isolation doesn’t break down from the primary to the secondary side.
For this design reinforced isolation was chosen.
That is the maximum voltage that the insulation of the transformer can experience under normal operating conditions. In our case that’s about 450V between the primary winding and the secondaries.
The standard specifies different pollution degrees depending on the actual machine/product the transformer is used in. For example, an epoxy potted transformer will always be clean, away from humidity and dust. But the most common thing is unpotted transformers that are subjected to dust humidity and other “pollutants”.
IEC 61558-1 defines pollution degree as below:
Pollution degree 1: “Pollution degree in which no pollution or only dry, non-conductive pollution occurs.”
Pollution degree 2: “pollution degree in which only non-conductive pollution occurs, except that occasionally a temporary conductivity caused by condensation is to be expected.”
Pollution degree 3: “pollution degree in which conductive pollution occurs, or dry non-conductive pollution occurs which becomes conductive due to the condensation which is to be expected.”
Most times pollution degree 2 is chosen, as is the case with this design as well.
That is a probabilistic categorization that hasn’t got to do with the transient voltages that a transformer will experience necessarily, but it’s rather a categorization depending on the application the transformer is used in.
We have 4 overvoltage categories (OVCs) named I to IV. Take a look at Table 4 to understand the applications and description of each overvoltage category.
Overvoltage category II was appropriate here.
CTI index is the voltage which causes tracking after 50 drops of 0.1% ammonium chloride solution have fallen on the material. The results of testing at 3 mm thickness are considered representative of the material's performance in any thickness. Tracking is the breakdown of insulation on top of an insulator created from a voltage potential which gradually creates a carbonized conductive path along the surface, creating a leakage-conductive path.
When we talk about insulating materials we consider the bobbin, insulating tape between layers, margin tape and wire ending sheeves like heat shrink tubes.
The cost effective/good compromise is to select material group II (400V ≤ CTI < 600V) for all of the aforementioned components.
We have previously selected:
From IEC61558-1 in Table 4 we see that the necessary clearance is between 3-5.5mm. Although most standards allow linear interpolation between values in tables, for this table the standard forbids this. So, the minimum clearance is 5.5mm.
We have previously selected:
From IEC61558-1 in Table 5 the necessary creepage is between 4.3-8.6mm. In this case linear interpolation between table values is permitted, thus 6.5mm of creepage distance is calculated for 450V.
This design was built with margin tape. Using 3.5mm of margin tape on each layer the final creepage is set at 3.5x2=7mm, as shown in Figure 7.
Now you’re in the position to understand in more detail the way we comply with a safety standard. To be honest there are a lot more things that need to be taken care of to make sure the transformer complies, but this is a major step in this direction.
[1] Sudberg, 2016, “Clearance, creepage and other safety aspects in "MySensors" PCBs.” Available at: forum.mysensors.org/topic/4175/clearance-creepage-and-other-safety-aspects-in-mysensors-pcbs
[2] (2018) Transformer Design Considerations. tech. Wurth Electronik. Available at: www.st.com/content/ccc/resource/sales_and_marketing/promotional_material/newsletter/group0/6b/10/c1/aa/ba/76/41/cf/Wurth_Design_Considerations_Munichhttps://freneticweb-2020-prod.s3-eu-west-1.amazonaws.com/files/Wurth_Design_Considerations_Munich.pdf/jcr:content/translations/en.Wurth_Design_Considerations_Munich.pdf (Accessed: February 13, 2023).
[3] IEC 61558-1, Safety of transformers, reactors, power supply units and combinations thereof. part 1, General Requirements and tests (IEC 61558-1:2017) (2019).
Optimizing transformer design for Dual Active Bridge converters with a focus on leakage inductance, core choice, and Frenetic’s simulation tool.
Optimizing transformer design for Dual Active Bridge converters with a focus on leakage inductance, core choice, and Frenetic’s simulation tool.
In this webinar, Bob White (President and Chief Engineer at Embedded Power Labs) presents an insightful lecture on the Two-Transistor Forward Converter.
In this webinar, Bob White (President and Chief Engineer at Embedded Power Labs) presents an insightful lecture on the Two-Transistor Forward Converter.
Richard Schlesinger (Research Scientist at Frenetic) presents an insightful lecture on the Leakage Inductance Model of Frenetic Simulator.
Richard Schlesinger (Research Scientist at Frenetic) presents an insightful lecture on the Leakage Inductance Model of Frenetic Simulator.
Reach out now and see what Frenetic can do for you.
