#71๐ 20 W Flyback with LM5155 from Texas Instrument
SuperPowers for Electrical Engineers ๐ฆธ๐ปโโ๏ธ๐ฆธ๐ผโโ๏ธ
๐ Hello, friends! Dr. Molina here ๐จโ๐ง
Welcome to Dr.Molina Newsletter, where I break down Magnetic design for you every week, in
4 minutesthe time needed.
Texas Instruments is a company I have always admired. They have been here ever, creating the chips for everythingโฆ, even they appeared in the famous series of startups called Halt and catch fire. They bought Unitrote, and with them, most of the great engineers at that company, including my friend Laszlo Balogh, who has helped me a lot in building Frenetic. TI is also very good doing application note. Maybe their publications arenยดt as easy to follow as the Infineon ones, but they have their Workshops, which are fantastic learning opportunities.
I found a new article about โHow to design a Flybackโ from TI and couldnยดt miss the opportunity to add value to their publication, which is already great.
The publication explains how to design the complete Flyback, but I will only talk about the magnetic. The specs used for the example are:
The article has a section for selecting the transformer, which explains how to specify the magnetic for a vendor. Later, they chose an off-the-shelf magnetic from Wรผrth. This is their process for creating the specs:
Calculation of the maximum duty cycle for the input voltage range
Turn ratio, considering 40% as a max duty cycle
Round the turn ratio. In this case, to 0,5 (Np/Ns=2)
Update final Dmax with the selected turn ratio
Calculate the auxiliary winding turns ratio
Calculate the magnetizing inductance
Calculate resultant waveforms (Ipeak and Irms)
Write the selection criteria for the magnetic
Dr. Molinaโs section
This is the section I would add if Texas Instruments call me tomorrow ๐.
First, at all, I will add some constraints:
Type of isolation: Functional
Isolation Voltage: 1000 V
Creepage: 2 mm
Clearance: 2 mm
I would draw the complete waveforms and the FFT, checking Irms and Ipeak in the voltage range.
I love to use magnetics off the shelf; however, engineers love to build their prototypes. Therefore, I will share several options to achieve a similar design.
For 250 kHz, I would choose a 3C97 material. This is a Flyback. Therefore, the core selection should consider the saturation limits. In this case, the Ipeak determined by TI is 6A. If we limit Bmax to 200 mT and take EFD and E core families, we find this.
In the size of EFD15 and E13 (0,5 cm3), with around 25 turns, we can get the 20 uH using 3C97. If we increase the size a bit with EFD20 and E16, we will have more options.
With these four options, Iโm sure engineers can get a solution. Of course, we must play with the number of turns to avoid saturation doing a trade-off between core and potential winding losses.
Choosing 4 options in that range, this is what I get at Frenetic, taking the Core Optimizer solutions.
I have seen designs for this power level with all the different cores; however, if I recommend one option for building a sample in your own lab, I would choose EFD20, itโs not the smaller one, but the design will be more straightforward. If you are a Frenetic user, be bold and go for the EFD15, and we will build the complex winding for you.
The winding losses are the main problem even in this low power cases because just 1 W of losses is tough to dissipate.
For the winding, I have done a simple interleaving to reduce the Llk to 100 nH, and I have used Litz wire in primary and secondary; the auxiliary wire is round.
I will run our new 2D model for winding losses, where I can see the distribution of the losses due to the gap and the magnetic field per wire. The accuracy is better than FEM so far. (This data will be accessible to Frenetic Users at some point in the next months)
I have selected an SMD coil former to complete the mechanical definition.
And thatโs all. If someone is interested in designing a Flyback with the chip LM5155 from TI and following their guideline, I already have several options for his transformer ๐ค.
Reference:
[1] How to design a Flyback, https://www.ti.com/lit/an/snva866a/snva866a.pdf?ts=1677154713341&ref_url=https%253A%252F%252Fwww.google.com%252F