franklatona-r49313

RF layout design considerations for manufacturing

Discussion created by franklatona-r49313 Employee on Sep 30, 2014

When it comes to RF layout board design, consider the areas that will have the most critical effects to board variation due to the manufacturing process.

 

One of the areas that should be considered is RF trace line width versus thickness to the ground plane layer.  Most antennas are designed as a 50 Ohm load impedance so the RF characteristics of the traces must match this for maximum power transfer.  Clearly designing the most cost effective board may mean RF trace widths and layer thickness tolerances will have to change with the manufacture process.  If having a standard 1 to 5% tolerance is typical then changing to a 10% tolerance may change product performance since manufacturing tolerances can change.

 

In general most applications transmit at the highest output power since long range is preferred requiring the lease number of nodes for a particular system.  At higher RF powers harmonics can be an issue for certification compliance but can be minimized by adding a simple notch filter to the antenna feed trace.  As an example, at RF frequencies consider designing an RF trace antenna feed using Microstrip rather than Coplanar Waveguide (CPW) topology.  Microstrip allows for small tolerance variations as well as later in the design phase adding harmonic trap components much easier than CPW.

 

Microstrip topology uses the trace width and thickness between the trace and lower ground area to build a 50 Ohm impedance RF trace.  As mentioned above, the antenna impedance is 50 Ohm so designing a trace as close as possible to 50 Ohms will allow for optimal RF performance.  This is shown in Figure 1:

Microstrip Dimensions.PNG.png

Figure 1:  Microstrip board design topology

Since only two dimensions are used for microstrip topology, manufacturer variation due to process or material can have less impact on board performance if tolerances change.  Microstrip also has less performance impact if harmonic filter components are added if certification requirements call for them.

 

Conversely CPW uses the RF trace width, the width between the RF trace and surrounding ground and the thickness between the trace and lower ground area to build a 50 Ohm impedance RF trace.  This is shown Figure 2:

CPW Dimensions.PNG.png
Figure 2:  CPW board design topology

As shown in Figure 2, three dimensions are required for CPW topology therefore manufacturer variation can effect performance if tolerances change from board to board.

 

Example impedance effects from a 1 and 2 mil tolerance variation using microstrip topology:
For a board stackup of 10 mils, the trace width must be 18 mils to maintain 50 Ohm impedance.  With a variation of 1 mil, the impedance varies from 49.3 to 52.5 Ohms.  With a variation of 2 mils, the impedance varies from 45.9 to 54.3 Ohms.  For a board stackup of 20 mils, the trace width must be 38 mils to maintain 50 Ohm impedance.  With a variation of 1 mil, the impedance varies from 49.3 to 50.9 Ohms which is not that critical.  With a variation of 2 mils, the impedance varies from 47.9 to 51.7 Ohms.

 

From this example, trace width variations do not change the impendance of the RF trace to the point the overall RF performance will be impacted.

 

Example impedance effects from a 1 and 2 mil tolerance variation using CPW topology:
For a board stackup of 10 mils, the trace width must be 13 mils with a gap width to ground of 5 mils to maintain 50 Ohm impedance.  With a trace width variation of 1 mil and no gap width change, the trace impedance varies from 48.9 to 52.3 Ohms.  If gap width is factored in, the impedance varies from 49.7 to 54.4 Ohms and 46.6 to 50.7 Ohms.  With a variation of 2 mils and no gap width change, the impedance varies from 47.4 to 54.2 Ohm.  If gap width is factored in, the impedance varies from 47.6 to 58 Ohm and 42.2 to 50.4 Ohm.  For a board stackup of 20 mils, the trace width must be 20 mils with a gap width to ground of 5 mils to maintain 50 Ohm impedance.  With a trace width variation of 1 mil and no gap width change, the trace impedance varies from 49.3 to 51.2 Ohms.  If gap width is factored in, the impedance varies from 48 to 53.6 Ohms and 464 to 51.7 Ohms.  With a trace width variation of 2 mils and no gap width change, the trace impedance varies from 48.4 to 52.2 Ohms.  If gap width is factored in, the impedance varies from 44.9 to 56.9 Ohms and 42 to 52.6 Ohms.

 

From the CPW example, trace width variations by themselves do not impact the impendance that much but when two dimensions change for larger tolerance variation, the RF trace impedance changes enough to impact the overall RF performance.

 

In conclusion, Microstrip is preferred over CPW to avoid potential manufacturing tolerance variations as demonstrated by the two simple examples.  In addition microstrip has less impact on performance if harmonic traps are required after the board has been built and certification requirements need to be met.  This is very important since higher output power application requirements can impact system harmonics.

 

Note: Calculations herein assume copper material with a thickness of 1.4 mils and a dielectric constant of 4.2.

Outcomes