Solution Finder Get Tech Support

Ultra Flat Design 60W Peak

Posted by: Tomeco on

Hello.

The customer has a requirement for a power supply that must be very flat (the populated PCB must not exceed 10.5mm).

230V input, 24V output, 15W continuous power and 60W short term (max 3 seconds, max 5% of the time).

I thought I might base this on the sample project "RDR-928 - 60W Ultra Compact USB PD 3.0 Charger"

I tried to modify the design in PI Expert, but I'm running into problems.

The first problem is the transformer.

I was looking for some exact instructions on how to calculate the transformer power of a switching power supply based on the core size.

I found some kind of guide here:

https://www.mag-inc.com/Design/Technical-Documents/Ferrite-Core-Documents

"Magnetics 2021 Ferrite Cores Catalog" (page 67)

However, this guide seems to me to be more empirical than accurate.

There are products like "cm4" and "mills/Amps" in the formula, which I don't really understand.

I might not understand it and could just blindly use it.

"Trust, but verify."

So I tried to calculate the transformer output from the RDR-928 using this formula.

I entered all the parameters according to the core datasheet and according to the RDR-928.

I entered a Current Density (Dcma) value of 600, which is somewhere between 750 "conservative" and 500 "aggressive".

I got 58.46W, which seems OK.

So you could say that the formula works, but...

Supplemented Bmax according to RDR-928 is 3464. According to the chart in the document mentioned, it should only be about 1000.

However, that wouldn't even get the output to 17W.

Maybe the core from "HWAHE Electronics" (the manufacturer of the core in RDR-928) has much better parameters than the one from "Magnetics International" (the catalog with the formula), but I don't really believe that.

More likely, I'm missing something and my knowledge is not enough.

The only core I have found that could hopefully meet my requirements is the ATQ25/9.3.

http://www.hwahe.com/page93?product_id=48

Even if I add some insulation and shielding thread on it, I could fit the required 10.5mm. The PCB would house the transformer in a cut out window.

In the calculation I left the Bmax at that 3464 and the power should be 75W.

Nice.

So I entered the core and skeleton parameters into PI Expert Online and... another problem.

The skeleton from "TBI" has a width of Bw=3mm.

http://tbi-tw.com/tbi_dwg/bobbin_th/25-atq/30_atq25-9.3e/TBI-238-09111.12XX(ATQ25-9.3E-8P-THT-1A-12)Rev.1.pdf

The datasheet even says the winding space is only 2.4mm.

However, PI Expert ignores this parameter and adds its own width (here 7.4mm), which is unusable.

Compare the photos: PIE_DesignPIE_Library



Is this a bug in PI Expert, or am I doing something wrong?

Can you help me with this design?

Thank you.

Files

Attachment Size
PIE_Design 54.49 KB
PIE_Library 67.16 KB

Comments

Submitted by PI-Wrench on 10/05/2022

If you run the design using PIXLS, you can select the "custom" option and enter the parameters for your core and bobbin.. The design in PIXLS is calling out for fairly large wire, but since the 60W peak is low duty cycle, you can use smaller wire size. I ran the PIXLS spreadsheet as if the design was running continuously at 60W to get an Innoswitch that would be sufficiently large to deliver the peak power. I used the PIXLS spreadsheet for Inno EP.. I've appended  a PDF file of the PIXLS spreadsheet I ran for the the design. You may need to correct some of the bobbin parameters, as I had to guess at some of the values. However, it will give you an idea of the Innoswitch size you need to deliver your peak power.

Attachment Size
InnoSwitch3-EP Flyback Design1.pdf 829.38 KB