👉 In the table above, the measured values and two simulations with a slightly different gap between the core set are shown. Observe how a 20μm difference can affect the magnetizing inductance of the core.
The initial design specs asked for a 3.4mH inductance. I got a value less than this, at 2.68mH. I have seen varnishing having an effect as well, in a no gap situation like this, but this sample wasn’t varnished anyway.
The leakage inductance is a bit lower than predicted.
For the PSFB topology the leakage inductance has an advantage and a disadvantage.
Pros:
👍 Remember that the ZVS condition is only achieved for a range of output loads. It’s common for the ZVS range to extend from 100% down to 50%-20% of the max power output. The more we want to extend ZVS down to lighter loads the more leakage or external inductance we need to add in series.
Cons:
👎 This extra inductance is causing a “duty cycle loss” phenomenon, which effectively means that the design might not be able to reach the max output voltage, and in that case the turn ratio must reconfigured all over again.
In a couple of real applications I have designed, I used an external inductor with inductance values form 10-30uH. A 1-2uH of a difference between prediction-reality wouldn’t made a difference there anyway😉 In others cases that leakage inductance maybe is all you need to achieve the needed ZVS range and the “shim” inductor is removed from the BOM altogether saving cost and weight.
BTW:
⭐️ It’s important to get the primary magnetizing inductance right in case current mode control is chosen as a control method.
I mentioned in a previous newsletter that the AL value of the actual core batch (for “zero” gap) designs will determine the final AL value.
🔑 Only through measuring the actual assembly for a sample, final AL value can be found, and then the transformer wound bobbin can be inserted to achieve the exact magnetizing inductance value. If the core set doesn’t match the AL value wanted, another set has to be chosen. This is time consuming…
✅ We can always design with the worst scenario in mind and be covered under all circumstances, and that’s probably a wise decision when you move from prototypes to production where we want to avoid human labor as much as possible.
A thermal comparison of the 2 transformer samples made:
