Designing a four layer stack up PCB


#1

I am designing a four layer stack up PCB having nrf52832 SoC with chip antenna (2450AT18B100: http://www.kynix.com/Detail/744089/2450AT18B100E.html).It is BLE Board (2.45 GHz). This is the first time I have been designing a PCB with an antenna. I am using kicad EDA Software. The stackup of my design is:

  • Signal and RF traces with ground copper fill (Green Colour in Picture below)
  • Ground Plane
  • 3.3V Plane
  • Bottom Plane (Basically Solid Ground Plane) (Red colour in Picture below)

I am going to add many vias (at 1/12 of RF frequency which will be 2.45 GHz). The antenna part is not completed yet because I am yet to receive some specification from my PCB fabrication house. I will change the width and shape of antenna feed considering the distance between signal plane and ground plane. I have one question regarding the calculation of width of antenna feed line. Since, I have four layers, Can I use formula of coplanar waveguide with bottom ground plane? Won’t power plane and bottom layer will effect the impedance of coplanar waveguide? Do I need to take into account presence of power layer and bottom layer (actually solid ground plane).

NOTE: The other sensor in the left part of my layout is a CCD (TCD1304) which is basically a linear image sensor.

What I want from the community is suggestion on the pcb and review my design for RF. Since, this is the first time I have designed such a board, I don’t want to design a PCB which won’t work or work with very short range. The pictures of my PCB are given below:

Picture 1:


Picture 2:

Picture3:

After reviewing the datasheet of my chip antenna and suggestions from Johanson Technology, I can see they have suggested a large ground plane for which I don’t have space.You can check pictures of my layout above as I have marked area for antenna (7mm X 23mm).

The picture below are taken from Johanson Technology’s layout suggestion:
LAczw
wI1iJ


#2

For something like this I would probably refer you to the manufacturer. I know Johanson so some design assistance and review services on their products. Here is a link to that page.


#3

OK, thanks. I’ll have a read~


#4

I’d recommend Signal Integrity Simplified by Eric Bogatin (ISBN 0130669466) as a resource worthy of acquisition and study for folks that are interested in electronics development beyond the trivial. It’s not the cheapest text in the world, but a bargain if one looks at it from the perspective of saving board spins and time spent in that unpleasant “it doesn’t work and I have no idea why” stage of development. Haven’t read the updated version (ISBN 013451341X) that supposedly includes more topics and costs more money, but I suspect it’d be worth the scratch also.

As to the original question, two foundational items to check on would be ensuring that the feedline to the antenna has the proper characteristic impedance, and to clear away ground planes from beneath the antenna, for the same basic reasons that folks doing mobile radio installations use coax cable instead of an old string of Christmas lights and mount the antenna on the outside of the vehicle instead of inside the trunk. The impedance question is a function of trace width, layer placement, board material, and stackup. Consult one of many online calculators available for a quick check, or Bogatin or similar references if you’d like to also understand a bit more of the ‘why’ behind it.

Notice also that any trace length extending beyond the ground plane effectively becomes part of the antenna rather than the feedline, and will affect results. Notice how in the sample layouts above, the little SMT component in series with the antenna is placed precisely on the edge of the ground plane? That’s the point where feedline ends and antenna begins, and the SMT placement allows a smidge of fine-tuning capability through component selection.

Insofar as the layouts shown don’t provide some key info, offering a thumbs-up or -down with any confidence is difficult. There does appear to be a tee network of some form in the ground-free area though, which suggests to me that that region might not be as ground-free as perhaps it ought.