CC1110 AVdd and DVdd decoupling Capacitor

Emanuele said:

I note different values and case size of Avdd and Dvdd capacitor. What is the reason? 

Both designs are valid and to some extent driven by designer preference. The CC1110 PA for Codec design is somewhat more rigorous in its bypass approach and may have more margin in certification testing. It uses 0603 caps for low values (eg 22pF, 220pF, <.1uF) where there may be an advantage to using a 0402 as stated below. In short if you went to all 0402s you would be fine at these locations. All 0603s may lead to unacceptably long trace lengths and coupling due to component crowding.

Trades:
A smaller 0402 cap at higher C values (>0.1uF) uses higher dielectric material which has more loss, is less stable over temperature, has more layers which are inductive, and thinner layers which lower the operating voltage. The tradeoff is poorer high frequency performance traded for reduced pad sizes and shorter lines to the pins.

A small value in a larger package (eg 22pF 0603) suffer from greater package and PCB parasitics, possibly longer traces to the pin, and less space between components depending on board density. It is my experience assembly yields are slightly better as compared to the 0402 and notably better than 0201s. Small values (<.1uF)  in the larger 0603 size are slight more expensive.

Bypass cap values: A 4 cap solution would approach ideal. A 50pF 0201, 1000pF 0402, .1uF 0603 and a 1uF 0805. The smallest closest to the pin. Each cap is optimized for its portion of the spectrum moving lower in frequency as you move away from the pin. As the frequency of interest is reduced, distance from the pin is  less critical, and package parasitics have less impact even with the larger package. 4 caps per bias pin is generally not practical or desired. A two cap pair of .1uF | 1.0uF is common at lower frequencies and 220pF | .1uF is often seen in higher frequency designs.

Thank you Stewart,

H Stewart said:

A 4 cap solution would approach ideal. A 50pF 0201, 1000pF 0402, .1uF 0603 and a 1uF 0805.

this approach is right also for 433MHz configuration? Capacitor value and case are the same?

Using 4 caps is overkill. Two caps are generally all you need.  A  .1uF  0402 and a 220pF 0402 is a good start.

Caps do two things: Decoupling and Bypassing  The decoupling capacitor is a charge storage device; and when the IC switches state and requires a spike of additional current, the local decoupling capacitor  supplies this current through a low inductance path. The cap needs to be large enough to provide enough current during the switching that the voltage at the pin does not droop. For this a .1uF 0402 is sufficient, you can use a 0603 if you prefer.

The bypass caps provide a short to ground for the high frequency RF signals to keep RF off the DC lines like Avdd and Dvdd. Bypass caps also provide a AC ground to circuitry inside the device, the device performance will degrade if this ground is not sufficient. The decoupling cap acts like a bypass cap up to the frequency where its internal inductance starts counteracting the capacitance or about 500MHz.  For a 433MHz design you want a part good to at least the 5th harmonic or about 2GHz. A 200pF to .01uF 0402  is good. This cap should be placed closer to the package pin, (or the same distance),  than the .1uF since it targets the higher frequencies.

There  are a few places where I recommend a 1.0uF 0603. These includes the DCOUPLE pin, one or more of the Avdd pins since they are related to the ADC, and near voltage regulators. These pins can have signals in the kHz range.

Stewart,

I need to correct you.  The reason for two caps in RF world is one cease to be a cap at high frequency.  Did you ever look at the self resonance of a capacitor.  I turns from capacitance to inductance.  But this varies with freuqency and layout.  So you can actually have a capacitor or an inductor act just the opposite of its intent.  Be careful making reommendaiton you are not clear what the correct answer.    For TI parts the .1uF capacitor's self resonance doesn't provide enough frequency range. 

Regards,

RSS,

RRS said:

For TI parts the .1uF capacitor's self...  

??? Did you read my post?  I suggested .1uF AND a carefully considered 220pF, spoke to the inductive component of the higher value caps and outlined an ideal 4 cap solution for the reasons you state.

RRS said:

Be careful making reommendaiton you are not clear what the correct answer.

I helped pioneer the modern chip caps from the early 1970s and have spent countless hours characterizing various caps and creating detailed models from a few kHz to over 200GHz for very broadband applications (eg .01 to 20GHz) and applications such as 60GHz space cross links, 97GHz radars and 183GHz receivers used for NASA weather sounders (ATMS). I understand MLCCs.

HS