SMPS and dealing with ground noise

[tinska.h]

Hello,

I have never dealt with SMPS and I'm now planning to use a SMPS-module from TI (PTMA401120) to power my system which has noise sensitive areas. According to datasheet there is Vo ripple 85mVpp, and I have read that SMPS can cause some amount of ground noise.

I'm thinking about using LC-filter to filter out ripples and then use low noise linear voltage regulator. However, how should I deal with ground noise? I guess common chokes could be an answer? Or should I create a virtual ground for AGND(Or as I'm doing transimpedance amplifier, I could reference op-amps non-invert to virtual ground). I'm also using power rail to power my digital interfaces.

Which one would be better or what you would recommend? Any more neat solutions?


Thanks,

 

[goldsmith]

Dear tinska.h
Hi
It is simple . use a simple differential filter at your out put , thus , you can attenuate the noise as high as possible .
Best Wishes
Goldsmith

 

[mtwieg]

I'm not exactly sure what you mean by "ground noise," but I'm assuming your referring to common mode noise/interference. How such interference is viewed depends greatly on how the SMPS fits into the overall system (is it isolated, do you have a earth/chassis ground, what are you defining as your reference plane, etc).

With a SMPS, the severity of the introduced common mode noise is highly dependent on the details of how the thing is constructed. In isolated converters, the construction of the transformer is especially critical in determining common mode noise. The proper way to reduce common mode interference is to use faraday screens between the transformer windings, but I have never seen this implemented in small DC-DC modules like yours. Having faraday shields on the transistor heatsinks also helps in high voltage/high power supplies, but that shouldn't apply to small modules like this.

If the SMPS isn't designed for low common mode noise, then suppressing it with external filters may be possible, to some extent. But without knowing the details of your system and your requirements, I can't really say much about your case. Mainly my point is that noise/interference is best dealt with at the source, not with external filters.

 

[tinska.h]

Hello,

Thanks for your posts.

My current arrangement with ground planes is that SMPS is converting Power-Over-Ethernet power as isolated DC-DC. I have chassis ground which is connected to SMPS ground via 1.5kV 1nF capacitor. I would like to keep the isolated ground as separate from chassis ground. Chassis ground will act as RF-shield and be connected to real earth.

I'm thinking of utilizing Murata's common mode choke to filter out common mode noise and then smoothen out the differential mode with ferrite beads and capacitors. Also in my analog circuitry I have utilized virtual ground and chosen components with good PSRR numbers.

I'm just not sure about whether this is the most elegant solution here. I could try different setups but dont want to waste time and resources

 

[mtwieg]

Do you plan on having any kind of Y capacitors from the ethernet cable to the shield? I'm not very familiar with POE.

You should make a detailed diagram of all connections between the POE cable, the isolated circuit, and the chassis (and anything else that connects to the device), including parasitics. From there you can predict what loops the common mode signals will want to follow, and how to attenuate them.