0
\$\begingroup\$

I've been designing my own custom RP2040 board, but with only a minimal number of components to do a UART and I2C peripheral, and no USB. However, the design guide https://datasheets.raspberrypi.com/rp2040/hardware-design-with-rp2040.pdf in Section 2.1.2. "Decoupling capacitors" recommends that you include nine 0.1 uF capacitors very close to each power pin on the RP2040.

My design is meant to be hand soldered, and I was forced to use 0402 capacitors for anything to fit in the PCB layout. I managed to get it working, but the 0402 capacitors were quite tricky to solder and there were a lot of them. Are there any other alternatives I can do for these capacitors? I normally prefer 0805 capacitors as easier to place and solder by hand. Is it possible to take two 0.1 uF capacitors next to each other and replace them with a single 0.2 uF cap in 0805 form factor? Or perhaps use a better quality and more expensive 0805 capacitor that is shared across multiple pins?

The design guide mentions there are design tradeoffs, but I'd be curious to know more about what the other alternatives are? I'd like to reduce the part count, and larger and more expensive is ok.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ hm, I find the layout from that guide to be rather strange; they seem to have a non-contiguous ground plane close to the IC, they seem to be effectively adding slots to the ground plane for the 1.1 V supply lines, which hence have no adequate return current path. In other words: I'm not a high-speed design issue, but if they are holier than thou about decoupling capacitors, maybe they should start at baseline sensible return current path layout. \$\endgroup\$
    – Marcus Müller
    Commented Sep 20, 2023 at 19:07
  • \$\begingroup\$ What soldering setup do you have? I find 0402s are difficult to solder with a basic Hakko FX-888 or similar, but quite straightforward with a Weller WX station with its smaller tips and shorter tip-to-grip distance. \$\endgroup\$
    – Hearth
    Commented Sep 20, 2023 at 22:45
  • \$\begingroup\$ I have a Weller with a very fine tip, and I'm able to do the 0402, it worked, but it is super finicky and there are nine of them. On my previous ATMega328p board, I had maybe two 0805 capacitors and that was a lot faster to put together a quick PCB. \$\endgroup\$
    – Wayne Piekarski
    Commented Sep 20, 2023 at 23:11
  • \$\begingroup\$ The document you links recommends using one capacitor near each power pin, not nine. \$\endgroup\$
    – The Photon
    Commented Sep 21, 2023 at 5:02
  • \$\begingroup\$ Yes, there are also nine VDD inputs on the RP2040, and each requires a capacitor very close to the pin, hence the nine capacitors. Was hoping to avoid that many capacitors and simplify the design if there is something I can tradeoff. \$\endgroup\$
    – Wayne Piekarski
    Commented Sep 21, 2023 at 5:35

1 Answer 1

Reset to default
2
\$\begingroup\$

Is it possible to take two 0.1 uF capacitors next to each other and replace them with a single 0.2 uF cap in 0805 form factor?

The reason multiple small capacitors as close as possible to pins are used is to have the lowest possible impedance at high frequencies by paralleling many capacitors. If you want to parallel fewer capacitors, the impedance will go up. If you want to use larger capacitors or more distance between the load and capacitor, impedance will go up. Eventually the design will stop working.

If the datasheet says X capacitors, I wouldn't recommend trying to do fewer. You could try larger (although 0402 is already large enough that hand soldering should be straightforward). Have you tried routing it with 0603? How bad does it look?

Another option would be to pay for assembly, which for small boards is actually relatively cheap. If the problem is that you're making 100 of these and you don't have the setup to do stencils and reflow, then probably you want to pay for assembly regardless.

Edit

Marcus points out that the example routing doesn't try very hard for tight current loops and constrains decoupling to be on a single side of the board. Are you also trying to put everything on one side of the board? If so, consider putting the decoupling on the backside, you should have plenty of room to use larger capacitors on the back while still having shorter current loops than the reference design.

\$\endgroup\$
5
  • \$\begingroup\$ well the linked design guide already admits that on their reference board they couldn't do as many. Not sure what that says about the quality of the guidance… \$\endgroup\$
    – Marcus Müller
    Commented Sep 20, 2023 at 19:02
  • \$\begingroup\$ Look at how their design looks from the bottom: imgur.com/a/4FM6GOi They have (possibly to add capacitance?) elected to do a 3.3 V fill on top copper, decided that 1.1 V needs no current return path, and decided to rather add C9 far away from the supply pins than to pring C11 closer. Pretty sure this is not an EMI-friendly board when there's 200 MHz switching noise on the 3.3V rails… \$\endgroup\$
    – Marcus Müller
    Commented Sep 20, 2023 at 19:14
  • \$\begingroup\$ I'm only making a few of these at home by hand, using a solder iron to put the capacitors in place, so not using reflow or stencils or anything like that. I'm just used to being able to solder my equivalent ATMega328p board very quickly since it only has maybe two 0805 capacitors for the voltage regulator and that's it. Makes quick prototypes easy to put together. \$\endgroup\$
    – Wayne Piekarski
    Commented Sep 20, 2023 at 23:17
  • \$\begingroup\$ @WaynePiekarski I've done many boards with 0402 decoupling on everything and hand soldered them. With a lens/microscope, a < 1 mm chisel tip with a cheap iron, and a $10 pair of tweezers, they're not much different than 0805. If you're finding a huge difference in difficulty, I wonder if maybe you have an inappropriate tip or are trying to do this with your bare eyes? Regardless, looking at that layout I would think you could do 0603 on the backside and still have a better layout than the example had with 0402. \$\endgroup\$
    – user1850479
    Commented Sep 21, 2023 at 0:50
  • \$\begingroup\$ I've done it successfully under a microscope and the board worked, it just wasn't a lot of fun to put those 9 capacitors down. The board was a lot more work than the ATMega328p. I was just curious to know if there was a way to simplify the PCB design if that is possible, since the documentation mentioned design tradeoffs indicating there might be another way. \$\endgroup\$
    – Wayne Piekarski
    Commented Sep 21, 2023 at 5:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.