🤓 Good news for you guys, something a bit different is coming today!
In the last weeks I've focused on the design and analysis of a 130W and a 500W LLC Transformers. While for the 130W Transformer I had the evaluation board at the Lab and I was able to test the sample, as you saw in last week’s Newsletter, for the 500W LLC Transformer there was no evaluation board at hand to make some tests.
💡 Instead of buying an eval board, which is not dirt cheap, I decided to design and build a new one, and to share the whole process with you. As you all know, without properly designed Magnetics and Power Electronics there is zerochance of a successful product!
🙌 Ready? Let’s go!
Our goal is to build a half bridge LLC Converter that can support 500-600W of power in an open loop configuration, just to be able to characterize our magnetic samples. I wanted to design this PCB with a minimum design effort in order to reduce the design time, so we can have a real advantage compared to buying a similar eval board.
Figure 1. LLC half bridge power circuit
🔎 In Figure 1 you can see the Power Circuit. First, we have CN4 as the input connector. Input voltage range freely, usually from 300-430V that is supplied from a PSU/PFC stage. After that, a capacitor at 10uF is enough to make sure that sufficient decoupling is provided. This capacitor should be film type, because of the high RMS ripple current rating, as well as its low ESR value.
If any ringing is observed to the DC bus rail, that means that the inductance of the input power cord to the board is oscillating with the decoupling cap. A 2.3uH in parallel with a 10R, 0.5W, smd resistor can be inserted in series before C4 to damp the oscillation, as shown in Figure 2.
Figure 2. Series damped LC circuit
Then we have R35, R41 which serve as bleeding resistors, whose purpose is to discharge capacitors in a matter of seconds/minutes so that electrical shock hazard is avoided.
☝️ Q2, Q3 form the half bridge. An RC snubbing network is added in parallel (R1, R13, C3, C12). During normal LLC operation, the mosfets experience ZVS, therefore these filters are unnecessary. In fact, they add 440pF of extra capacitance to the switching leg in addition to the Coss of the mosfets, making it harder to achieve ZVS. But the reason why I’m putting these filters is to account for burst mode initial cycles or any hard switching conditions that might occur during testing. The pcb layout in that case will dictate the spike on the Vds that can cause the mosfets to go into avalanche mode and if the energy is enough to fail.
💰 A cheap way to measure AC current with an oscilloscope.
No need for current probes or Rogowski coils here!
In series with the Inductor, Transformer (TR7A, T1), there is a current Transformer (TR2 - PA1005.100NLT). This miniature Transformer has 1 turn on the primary and 270nH of inductance, so it doesn’t affect our circuitry, as you can see in Figure 3. That’s an easy way to accurately measure the resonant current.
📈 From personal measurements and simulation, with 1m of coax 50Ω impedance cable we can measure signals up to 20A or 2V on the scope (1A ->0.1V) with a rise time of ~30ns to be expected, practically about 10MHz of BW for our custom current probe. A 50R series resistor is enough to damp any ringing and provide series termination. Figure 3 shows an LTspice simulation with measured parasitics included. This Transformer saturates above 52Vus. For example, a 20A signal will be displayed as 2V, so a Ton=52/2>25.5us will drive the current transformer into saturation, distorting the waveform.
Figure 3. LTspice simulated current transformer-coaxial cable
⚡ C7, C9 capacitors are providing a current return path for the common mode currents that are transferred from the primary side to the secondary side of the Transformer during rise/fall times via the parasitic capacitance of the primary-secondary windings. About 1nF of capacitance in total is a good rule of thumb for these capacitors. Here I connected x2 X1/Y1 capacitors (although X/Y type caps are not mandatory here) in series to double the voltage rating of the effective capacitor.
Across D3, D4 RC snubbers are connected to damp the ringing caused by any reverse recovery currents. That’s standard practice that works well. These RC snubbers can be finetuned during tests.
Figure 4. Half Bridge driver circuitry
⚙️ I don’t think I have to explain much here. The BNC connectors are connected to a dual channel signal generator. Two square waves with some degrees of phase difference, thus a couple hundred nanoseconds of delay time between the signal, will serve as dead time between switching of the mosfets. About 1W will be sufficient to drive moth mosfets. Usually without going into math, from experience 1-3W is all that’s needed for half and full bridges at 100-200kHz operation. The floating PSU selected, that drives the switches, can supply 3W and therefore, we are covered!
📩 If any of you readers is interested in the Altium project (not completed yet!) or has a suggestion or correction, feel free to answer this Newsletter and get in touch with me!
🎙️New live Webinar dedicated to "Exploring Frenetic and Maxwell options for an Optimal Performance"
Hosted by Sotiris Zorbas and Alfonso Ramirez, the lecture will start from a LLC transformer design in Frenetic Online and followed by the Ansys 3D simulation for its visualization and optimization.
