LM5180: Advice Request - High Conducted Emissions Peaking at 56 MHz

Hi Jason:

Have you put a y-cap between GND and S-GND which may helpful for the high frequency EMI, you can try 1nF see whether it helpful.

One more question is which standard you have to pass? normally CIPSER 22 CE will be under 30Mhz, when it above 30Mhz, commmon noise may become the major noise souce. 

Thanks

Daniel Li

Daniel Li14 said:

Have you put a y-cap between GND and S-GND which may helpful for the high frequency EMI, you can try 1nF see whether it helpful.

Hi Daniel,

Yes, I have a Y-cap on the board already. Sorry I forgot to show that on my schematic. It's on a different page.

Daniel Li14 said:

... which standard you have to pass? normally CIPSER 22 CE will be under 30Mhz, when it above 30Mhz, commmon noise may become the major noise souce. 

I need to pass CISPR 32/EN 55032 Class A

I will look at a few more things today:

1. Minimum load conditions. Both of my outputs have very little load. I'll give them a minimum load of 50 mA and see if that makes an improvement.
2. Layout. I'll compare my layout to the LM5180 evaluation board LM5180EVM-DUAL and see if I made any poor layout choices.
3. I might buy the LM5180EVM-DUAL and test its conducted emissions to compare it to my design.
4. I'll plot my conducted emissions measurements and show how they compare to the CISPR 32/EN 55032 Class A limits

I should clarify some aspects of my testing. I am using a current probe to measure the common mode current as a proxy for radiated emissions. This test procedure is described in Henry Ott's book in 18.3.1

TITLE: Electromagnetic Compatibility Engineering
by Ott, Henry W
ISBN: 0470-18930-4
ISBN 13: 978-0470-18930-6
Publisher: Wiley
Publish Date: 2009-08-24
Binding: Hardcover

The frequency range I selected is 30 MHz to 500 MHz. I didn't go lower since my resolution bandwidth is 100 kHz. I would need to do a second sweep at lower frequencies with more resolution. I didn't go higher than 500 MHz since my current probe's transfer impedance gives ineffective results above 500 MHz.

I tested with a higher load and the results look worse

Green - the device under test is not powered. Current measured is from the cable acting like an antenna picking things up in the room.

Blue - The device is powered normally, minimal load on the +15 VF and +3.3 VF rails.

Red - Same as blue but higher load on rails. Added 333 Ohm to +15 VF ~50 mA. Added 56 Ohm to +3.3 VF ~50 mA

Hi Jason:

some idea from my side:

1. your input voltage in schematic is 12V , for EVM it fitting from 10V to 65V, what's you max input voltage? and what's your input votlage when you test the EMI?

2. have you change or raise the input voltage see whether the 56Mhz noise will shift?

3. can you change lower voltage diode of secondary see any help 

4. change the Y cap to smaller value see any improvement.

5. any poisslbe to shield the transformer?

6.any resonant votlage you can see in the primary SW or secondary diode has 20nS period?

Thanks

Daniel Li