Hi I’m building a board for a 7.9 GHz frequency synthesizer using microstrip transmission lines but since I have no prior experience building multilayer boards I’m not quite sure if I made my board correctly. I’m planning to submit this to a manufacturer with a requirement of 8 mils trace width and spacing. All the help and advice I can get to improve my layout is very welcome.
Thanks in advance and Happy Holidays to all.
Hi Worrier,
7.9 Ghz, not a trivial task!
What is your board’s layer spacing and the substrate you’re designing for?
A couple of suggestions that jump out:
For your shunt matching elements on your transmission line, I’d avoid making the microstrip into “T” shapes. Leave the microstrip straight, and just place the top pad of the shunt elements onto it.
Put more ground vias on the the ground side of the shunt elements. You’ll want these connections to have as low of a resistance to ground as possible.
Either get rid of, or thicken your thermal relief lines to your SMA connector’s ground pins. It’ll make it harder to solder, but you want as few constrictions in your ground return path as possible.
What’s the purpose of the ring of vias around each of the SMA connectors? (Given that you don’t have a top layer ground pour)
The highest two-pin jumper in the picture is a bit too close to the RF trace to make me comfortable.
Assuming the bottom IC has a ground slug, you’ll want a lot of vias to ground in that ground slug.
Thanks for the reply Roko. I’m gonna use the usual double-sided FR4 with 1.6 mm thickness.
Regarding your suggestions:
I’ll just adjust the shunt elements. Thanks for pointing that out.
Honestly this is my first microwave board so I don’t have any knowledge regarding placement of vias. Is there a required number of vias to achieve a low resistance? Does this only apply for the shunt elements or should I add vias for the different ground (Analog, Digital and Charge Pump) pins of the IC as well?
Regarding the ring of vias around the SMA, I thought it was necessary to improve isolation. Is it overkill?
I’ll adjust the highest 2 pin jumper. I was also doubting if that would work. Thanks again for pointing that out.
My apologies for the multiple follow-up questions. Just want to minimize problems with the board before I can submit it to the manufacturer. Thanks again.
Hey Worrier,
To answer your questions
Generally with RF ground, the more vias the better to try and get a solid ground return path. I’m sure there’s a point with too many, but a few for each grounded end of a shunt element is good.
If there’s no ground layer on the top of the PCB, I can’t imagine the via’s doing too much.
FR4 isn’t the best choice for higher frequency circuit boards, though it should still work. The trick to RF design is that your first board will likely have problems with it. If you have access to a good spectrum analyzer (ideally with sweep generator and a directional coupler) or a VNA, you’ll be able to tune it. Sometimes I’d run prototypes with 8 slightly different PCBs in a panel to try and find out which design worked the best (for things like microstrip widths, 1/4 wavelength stub tuning, PIFA designs, etc… Things where the “theory” doesn’t always play nicely at microwave frequencies)
Best of luck… If you have any more questions, don’t hesitate to ask.
Hi Roko. Thanks for the reply. It’s been pointed out to me that the Microstrip lines of the original board layout is not really Microstrip (the reason for which I don’t get since I assumed I should just apply the width computed by txline) and with FR4 as my material, I should make the traces as short as possible.
Here’s my updated layout. My only problem with this is the long distance of the trace from the capacitor to FOUT. Hope to get some feedback if this is good. Thanks in advance and Happy Holidays.