CASE STUDY - PICA

In 2014 one of my long-time clients needed 24W of isolated 5V in a tiny 1” square with the smallest transformer possible and smallest overall cost. The request sounded not all that difficult, but the further we looked at the problem, the more difficult it got. Without breaking the bank while maintaining maximum efficiency, we needed to go with a boundary-mode mosfet-switched flyback. We didn’t feel like we could afford synchronous rectification for the efficiency bump, so we were also stuck with Schottky diodes on the output. The advanced quazi-resonant controller we chose sets a minimum frequency limit based on the input and output parameters. This minimum frequency ultimately dictates the transformer design envelope, but key circuit parameters play together in non-obvious ways; leading to a circuitous search for an optimum design balancing the competing requirements of size, weight, performance, and cost.

In spite of the 3-month timeline, I was able to survey all available small high-frequency ferrite core types, prototype them, and test them relative to the model. After iterative optimization for efficiency, size, controller operating frequency, mosfet losses, diode losses, power delivery, etc., I downselected from around a dozen variably optimized prototype designs. The overall winner of the transformer shootout competition was a magnetic design that achieved all the requirements in a tiny RM5 core size with an elegant yet highly optimized interleaved litz winding. This, along with specific choices for output diodes, snubbers, mosfets and filter capacitors, led to an overall design that met or exceeded all the requirements. The chosen transformer design was shared among Nielltronix’s extensive network of custom magnetics winding houses, and ultimately 2500 units of the fully-custom machine-manufactured, dipped, and hipotted transformers were manufactured in under 30 business days for the first pilot build with zero defects or returns. For reference, that was 24W with a 550mm^3 core. That equates to a transformer power density of 40MW/m^3, and overall power density of 150W/in^3 which for 2014 was quite an achievement for an isolated flyback converter at minimum cost. Smaller and higher density is possible with more elaborate synchronous rectification and higher frequency designs. However, in this case simplicity and BOM cost pressure were paramount, and a remarkable design was achieved under budget and within the required timeframe.

This design is still being used nearly a decade later.