This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LMZ31710: Relevance of AGND in LMZ31710 and TPSM84A22

Part Number: LMZ31710
Other Parts Discussed in Thread: TPSM84A22

In the datasheet of LMZ31710, it is told not to connect AGND (analog ground) with PGND (common ground) because they are connected internal to the IC.

[1] What are the reasons for recommending not to connect AGND in PGND?

[2] Why are Radj and Rrt_clk referenced to ANGD and not to PGND?

I normally maintain one solid ground plane on layer 2 instead of dividing the ground planes. If I follow the recommended layout, I need to divide this solid ground plane. Having different grounds in a circuit board has its own set of problems; so I don't prefer to go to that path.

In the Zynq 7000 SoC EVM (zc-706; reference device - U98), AGND is routed as thick trace in layer 1, while underneath it in layer 2 is a solid PGND ground plane. In such a case, the EM fields of any signal in between Vadj - AGND connection and RT_CLK - AGND connection are contained within the traces on layer 1 and PGND ground below. This means that the PGND ground is referenced anyhow.

[3] Do you think the way it is routed in Zynq SoC okay?

[4] I have the exact same question for the routing of Radj between Vadj and AGND in the new power module TPSM84A22.


  • Hi Binayak,

    Thank you for your request regarding the LMZ31710.

    1) AGND is connected to PGND internal to the device to keep the quiet, reference ground (AGND) from the noisier, switching ground (PGND). They are connected at one point internal to the module to avoid ground loops.

    2) Radj and Rrt are referenced to AGND to give the most accurate reference and to avoid switching noise from affecting the tolerance.

    3) I would recommend connecting AGND as a trace on a single layer while making PGND a solid ground plane. There will be no need to connect AGND to PGND since the connection is already made internal to the device.

    4) Same for TPSM84A22.

    Regards, Jason

  • Hi Jason,

    If we look at energy flow (signal or power), the electromagnetic fields will flow in the dielectric space bounded by the trace and the plane in adjacent layer (a transmission line structure).

    Consider: a trace on top layer runs from Vadj pin to the series resistor Radj (also on top layer), then the trace continues to run  upto AGND pin - all in top layer. On the adjacent layer below this top layer is a solid PGND plane. If a signal exists in this configuration, it exists as electromagnetic field waves that travels between the trace on the top layer and the PGND plane on layer2. Hence, the notion that the PGND plane will not be referenced in this scenario is a mistake.

    Everything will be referenced to PGND plane in such scenario - whether we like it or not.

    So I still don't see how we can avoid referencing to the PGND plane.




  • Hi Binayak,

    Yes, I agree signals will be coupled between layers.

    We do need AGND to be referenced to PGND in order for the circuit to have the same 0V reference. This is why we make the connection internal to the device. This ensures that the connection is made without the heavier power conversion current flowing through the IC reference ground.

    Regards, Jason

  • Hi Jason,

    I'll be routing the AGND as a trace on the top layer with a solid PGND plane on layer 2 (just like in the zynq soc EVM).  I find it as a much better way to route than create a separate AGND plane as shown in the recommended layout guidelines of the data sheet.

    Thanks for your help.