nRF7002 decoupling capacitors

Looking the nRF7002 reference schematics in the datasheet and I'm wondering if you really have to use so many different capacitor values for decoupling. I would like to harmonize the values to reduce the number of lines in the BOM, which would reduce cost in manufacturing.
Are there any specifications about the minimum and maximum capacitance values?

The MLCC capacitance under DC bias is a related matter. As we all know, for example 100nF 0201 capacitor under 3V3 might be actually something like 60 nF. For bigger caps, the loss of capacitance is even worse. Therefore, it would be nice to know what are the real required capacitances under DC bias, rather than some nominal values.

Another thing I would like to protest, is that you show smaller capacitors connected in parallel with a larger one. It's based on wrong conclusion. All the caps are small SMD chips, and they all actually have ESL in the same range. A smaller cap isn't any better at high frequencies. In fact, having different capacitance values parallel connected may be dangerous: with the cap ESL and trace inductances they may build up harmful parallel resonances. My practice nowadays is to use a single value for parallel capacitors - it's much safer regarding unpredicted resonances.

Parents
  • Hello,

    The short answer is follow the reference schematic and layout, including BOM, to guarantee operation as stated in the datasheet. That said, I have forwarded your input internally and will get back to you with feedback once I receive it, this normally take some time.

    Kenneth

  • Hi,

    I checked with our hardware team, and here's a summary of their input.

    Nominal vs. derated values
    You're right that DC bias matters. The values that count are the effective (derated) capacitances. That's why the reference design uses a specific BOM: those parts have known derating behaviour.

    Why there are so many different values
    The nRF7002 has several functional blocks, and the decoupling network is designed around them. The capacitors decouple the internal LDO outputs and the supply pins, keep noise apart between the supply domains, and supply local charge for the current peaks during Wi-Fi TX. Because the supply pins are spread around the package, most capacitors serve a specific pin, so their placement matters as much as their value.

    Changing the BOM
    The reference design is the validated configuration, and we strongly recommend following it as it is. Changing values, swapping parts or removing capacitors can affect regulator stability, noise isolation and RF performance, and we can't guarantee performance for designs that deviate from it.

    Kenneth

Reply
  • Hi,

    I checked with our hardware team, and here's a summary of their input.

    Nominal vs. derated values
    You're right that DC bias matters. The values that count are the effective (derated) capacitances. That's why the reference design uses a specific BOM: those parts have known derating behaviour.

    Why there are so many different values
    The nRF7002 has several functional blocks, and the decoupling network is designed around them. The capacitors decouple the internal LDO outputs and the supply pins, keep noise apart between the supply domains, and supply local charge for the current peaks during Wi-Fi TX. Because the supply pins are spread around the package, most capacitors serve a specific pin, so their placement matters as much as their value.

    Changing the BOM
    The reference design is the validated configuration, and we strongly recommend following it as it is. Changing values, swapping parts or removing capacitors can affect regulator stability, noise isolation and RF performance, and we can't guarantee performance for designs that deviate from it.

    Kenneth

Children
No Data
Related