Multilayer LC filter GND connection

[aaron_do]

Hi all,


I want to use a multilayer LC filter in my design, but I'm worried about how to hook up the ground. These are the filters I'm referring to.

**broken link removed**

My issue is that the ground pins are on the sides, whereas I'm using microstrip lines in my board. I can move the side ground planes closer to hook up this filter, but the question is, what is the return path for the current? Unfortunately, I don't have the resources to simulate this.

The second question is, do I join the two side ground planes together underneath the chip? I assume it would improve the input-output isolation, but I'm worried it would again mess with the return path. Any thoughts, or links would be helpful.


thanks,
Aaron

 

[FvM]

Some design details are probably open for discussion, but there can't be any doubt that you need multiple vias to connect the filter ground pads to the micro strip ground plane.

 

[aaron_do]

That's a good point which I will keep in mind. I'm still interested to know whether I should be connecting the two side ground planes. Also, the only way I see this design having a return current flowing in a "nice" manner is if it is connected to a coplanar waveguide. Any thoughts?


thanks,
Aaron

 

[volker@muehlhaus]

My issue is that the ground pins are on the sides, whereas I'm using microstrip lines in my board. I can move the side ground planes closer to hook up this filter, but the question is, what is the return path for the current?

Aaron, this is a really good question. If you have the ground at the bottom of the PCB, and connect the filter ground with vias, the via inductance will impact the filter response. This is especially important with high isolation requirements. Think of a filter that has a "local" ground, and then some inductance from that local ground to the circuit ground. Now shunt current in the filter input will flow through that ground path inductance, and induce a voltage at the filter output. So we really need to minimize the ground path inductance.

For very best isolation, you would connect the "in" and "out" grounds at the filter ground pin, and force ground current to flow through up to that level and back down. Then, the signal ground path is directly at the filter ground pad - it can't get any better than this. The bottom (back side) ground would need to be slotted, so that ground current will indeed go the easy way up and down.

Or for "good enough" isolation, just keep the vias path to ground as short as possible.

 

[aaron_do]

For very best isolation, you would connect the "in" and "out" grounds at the filter ground pin, and force ground current to flow through up to that level and back down. Then, the signal ground path is directly at the filter ground pad - it can't get any better than this. The bottom (back side) ground would need to be slotted, so that ground current will indeed go the easy way up and down.

I hadn't thought of that. Thanks for the insight. Would be interesting if I could somehow simulate this.

thanks,
Aaron