All,
I’m attempting to design and build an AM/FM tuner/amp board utilzing the Silabs Si4731 tuner, and TI TPA2016D2.
Long story short is I’m on my second PCB as I found the TPA2016D2 amp (class D) created a lot of interference in the FM range, and learned the need to add some ferrite beads on the amp output and power input…
Well, I thought this would fix things, but unfortunately it did not…even with the amp completely off, I’m observing poor reception (SNR values of mid to upper 20s).
I’m controlling this board with a Raspberry Pi 3, and my best guess is I’m getting interference/noise from the RPi. Any thoughts on this? I know that’s a pretty broad question.
I tinkered with the Sparkfun Si4703 board in the past and had good luck with it connected to an RPi…I just noticed the Sparkfun Si4703 board has series resistors on the Si4703 inputs, whereas my design does not. Does anyone know if Sparkfun added these resistors (330ohm) to deal with this?
Have you compared your layout to Sparkfun’s Si4703 board? I would look very closely at you power supply decoupling around the Si4731, and even more closely at the grounding for the Si4731 (e.g. are you using the part with the exposed pad and is it grounded?, are you using a good ground plane underneath the IC?).
You can also take a look at SI Labs [AN383.](https://www.silabs.com/documents/public/application-notes/AN383.pdf)
Thanks for your reply languer…I’ve compared my layout to Sparkfun’s to some degree, but I’m likely too much of a noob to understand the subtleties.
I’m using the SSOP package, but think I need to switch over to the QFN package.
I’ve studied the AN383 integration guide, but it only shows the QFN…I mainly referenced it for antenna stuff.
The whole idea of power supply decoupling is new to me…but if you’re talking bypass caps and stuff like that, yes I’ve integrated them.
I don’t think the QFN package will provide much better results just because. Getting that damn pad at the bottom of the package properly mounted is not trivial. The main thing is that you have a good ground plane, that the copper underneath the IC (i.e. the next layer down) is ground and you don’t have it like swiss cheese from all the signals you are routing through it. That all the power pins have good decoupling to a good ground plane (like the next layer down and not like an inch away). And that you take care of the I/O. The app note references the antenna but if you actually look through it you will find plenty of information on decoupling and I/Os (e.g. Table 6, Section 2.5, Section 2.6).
Thank you languer…I do have some traces routed under the chip, breaking up the ground plane, so I’ll get to work on avoiding that. Perhaps a QFN will help me with that? I’m already dealing with a QFN on the amplifier, what’s one more!? 
I’m actually soldering the components onto the PCB with a custom hot plate I designed and built. Utilizing it, with a solder paste stencil, seems to make the QFN soldering tolerable.
I’ve been studying the app note references, and yeah, I have other problems. Oh well, I’m learning, and having fun…although it has been frustrating. I probably should have picked an easier first PCB project.
That is a very cool hot plate setup. If you already have another QFN, and are doing smt stencils, then I think QFN is the way to go.