Massive EMI issues with TPS61235

MR Flashlight

Other Parts Discussed in Thread: TPS61088

We went to EMI pre-scan in the past two days with a new product. We are aiming for class B certification, and fell very short. One major issue was with the TPS61235. It's throwing off a ton of noise in the 100-140 MHz range.

We have already tried switching to a better-shielded coil and a larger output ferrite bead, but this was still not enough to bring our device into spec.

I've begun the process of finding a new part for a 5 V boost converter under the assumption that there is nothing that can be done to save this part, but if there is a way to get into spec it would save a lot of effort. The schematic is shown below:

L3 and L4 were initially a Coilcraft XFL4030-472 and a Wurth 742792602. We tried switching them to a Wurth 744025001 and a Wurth 742792651.

Below is an image of the EMI scan. The largest 2 spikes are radio stations, but the rest is coming from the TPS61235.

over 8 years ago

0 JohnTucker over 8 years ago

TI__Guru**** 191445 points

Do you have ringing on the SW node at those frequencies?

0 MR Flashlight over 8 years ago in reply to JohnTucker

Intellectual 420 points

Not particularly, no. I have attached images from my scope. One is an FFT with a center frequency of 100 MHz and the other is just a single pulse. The ringing dies out very quickly.

0 JohnTucker over 8 years ago in reply to MR Flashlight

TI__Guru**** 191445 points

Typically the first step to reducing EMI is to find the source. Typically for dc/dc converters it is SW node. Either the fundamental switching frequency and harmonics or ringing frequencies on the rising and falling edges.

0 Jasper Li over 8 years ago

TI__Mastermind 46905 points

Refer this application note "www.ti.com/.../slva790.pdf" about the reduce the EMI of the TPS61088. Although different device, the method to optimize the EMI should be similar.

0 MR Flashlight over 8 years ago in reply to Jasper Li

Intellectual 420 points

This document does seem to be helpful. The high frequency path is already very short, but we've added a 1 uF capacitor on top of the 22 uF output capacitor to serve as a better high frequency path. We've put in a snubber circuit which has gotten rid of a large portion of the ringing on the SW node. We also put in a 1 uF as a high frequency bypass on the input side.

We're heading to the lab tomorrow morning, so I'll update then on whether this has managed to solve our issues.

0 MR Flashlight over 8 years ago in reply to MR Flashlight

Intellectual 420 points

I forgot to update this after our testing.

In short, it didn't work. We still had significant EMI, well above the class B threshold. It looks as if we took care of some noise at the lower frequencies while potentially pushing some noise into the higher range. The graph for the scan is below.

I had another engineer suggesting that there might be some issues with my grounding (for example, AGND is shared by my microcontroller - I was under the impression I had to just keep AGND away from the high power ground). But to move things forward we have dropped the feature controlled by this device, since we don't seem to have a way to make it work well.

0 Helen chen over 8 years ago in reply to MR Flashlight

TI__Mastermind 40080 points

Do you put a whole ground under the TPS61235? What is the output current of the TPS61235? try to add one bead in series with the Vo+ and one bead in series with the Vo- to see whether it could be helpful.

If that still doesn't help. Please send your pcb layout in pdf file to us. My email address is helen-chen@ti.com

0 MR Flashlight over 8 years ago in reply to Helen chen

Intellectual 420 points

There is a full ground on the layer underneath the TPS61235, but due to space constraints the ground is not complete under every other component, and some components had to be moved to the opposite side of the board. At this point it is too late to be randomly adding beads trying to get the noise down as we are not going back for additional EMI scans. Short of someone able to tell me "this will definitely solve your EMI issues, guaranteed" we are not going back to the TPS61235 at this point.

I'll look at sending in gerbers (need to get it okayed to send that information out), but we have limited space in which to make corrections.

0 Helen chen over 8 years ago in reply to MR Flashlight

TI__Mastermind 40080 points

As there're other circuits in your system. The EMI problem may not be caused by the TPS61235 boost converter.

0 JohnTucker over 8 years ago in reply to Helen chen

TI__Guru**** 191445 points

Yes, you will have to [positively identify the source of the unwanted HF content before you can eliminate it.

0 MR Flashlight over 8 years ago in reply to JohnTucker

Intellectual 420 points

My regular application code turns circuits on and off depending on the operating mode. While these tests were being run, the only active power circuit was the TPS61235. The noise seen in this test was not seen in my other tests. So I can say with confidence that this noise is, in fact, coming from the TPS61235 circuit.

0 Helen chen over 8 years ago in reply to MR Flashlight

TI__Mastermind 40080 points

Its interesting! could we discuss by email. my email address is helen-chen@ti.com

Power management

Power management forum

Massive EMI issues with TPS61235