nRF9160 reference design VDD decoupling caps (physical positions seem reversed)

[Joe Butler]

On the reference designs for the nRF9160 DK (and also the 91-Thingy), the 3 x VDD decoupling caps per VDD pin have the largest cap value closest to the nRF9160. This seems to be the opposite way around to what I would expect.

Is there a reason for this?

Can you confirm that this is the correct placement (i.e. that it should not be the smallest cap closest to the nRF9160 VDD pins)?

Also: with the ferrite bead, its placement between the caps and VDD1 implies that it is filtering noise originating from the nRF9160 rather then filtering out noise from the power supply?

What are the downsides to not using a ferrite bead in this position?

 

[mathiaso]

Hi, Joe.

Is there a reason for this?

Can you confirm that this is the correct placement (i.e. that it should not be the smallest cap closest to the nRF9160 VDD pins)?

Yes, this is the correct placement. The reason might be to have the lowest total ESR, due to shorter trace length between the pin and the larger caps, as there is a possibility of larger current going between the pin and the larger caps than between the pin and the smaller caps.

Also: with the ferrite bead, its placement between the caps and VDD1 implies that it is filtering noise originating from the nRF9160 rather then filtering out noise from the power supply?

Yes, the ferrite bead is filtering noise originating from the nRF9160.

What are the downsides to not using a ferrite bead in this position?

As the nRF9160 creates some noise on VDD, you will have more noise in the VDD net if the ferrite bead isn't included. You can see this has been added from v1.2 to v2.1, if you compare the reference circuitries.

Best regards,
Mathias