Design Synergy: Combining LINC2 Circuit Synthesis with Sonnet EM Simulation

automatically creating the layout and exporting it as = a DXF file to be imported into pcb layout software. Figure 10 shows the finished print= ed circuit board.

The circuit is 100% microstrip and ground vias so ther= e are no components to mount except for the SMA board edge connectors. Figure 11 shows the filter being t= ested using an HP-8720D vector network analyzer.= The results of the test are shown in the Measured column of Table 1 = (and the dark traces plotted in Figure 12).&nbs= p; Measurements taken with the vector network analyzer verify that the filter is passing the design goals on all points from the first build.

Figure 8, New LINC2 Notch Filter Specifications

Figure 9, Final Version of the Notch Filter (EM Simulation Enhanced)

Figure 10, Notch Filter Printed C= ircuit

Table 1, LINC2 Designed Notch Filter - Measured Vs. Simulated Results

Filter Parameter<= /span>	Design Goal	EM Simulated	Measured
F= c (Notch Center Frequency)	1= 487 - 1493 MHz	1= 489.8 MHz	1= 489.5 MHz
U= pper 3 dB Cutoff	&= lt; 1575 MHz	1= 533.25 MHz	1= 559 MHz
-= 30 dB bandwidth	&= gt; 13 MHz	1= 7.75 MHz	2= 1.375 MHz
N= otch Depth	N= ot Specified	4= 1.78 dB	4= 2.83 dB

Figure 11, Notch Filter Measureme= nts on the HP-8720D VNA

Figure 12, LINC2 Designed Notch F= ilter Response - Measured (dark traces) Vs. Simulated (light traces)

Calibrating the Simulator<= /p>

For a given filter order and specified bandwidth, the = depth of the filter’s notch is primarily related to the filter’s losses. The notch will be dee= per if the losses are low. Thus, get= ting a good simulation of the notch depth depends on making sure that the EM simul= ator is modeling not only the dielectric losses but also the metal losses (copper losses in this case). There a= re a number of metal models that can be selected in the Sonnet EM simulator including Normal, General, Rough Metal, Resistor, Thick Metal Model and others. The default, Normal, model was used in this example. The Normal model requires the specification = of conductivity, resistivity or sheet resistance at DC and the metal thickness= . Using the N= ormal metal model and specifying the conductivity for copper, the simulation predicted a notch depth within 1 dB of the measured 43 dB.

The EM simulator encloses the circuit in a conductive box. The top and/or bottom of= the box can be open (free space) or closed (lossless or any of the specified metals). If the circuit being simulated is to be completely enclosed in a metal shield box then the EM box provides an excellent simulation of the actual product. However, if it is intended that the actual physical circuit is not to be shielded but instead is to reside on an open unshielded pcb then the EM box top should be specified as free space (= the bottom of the closed EM box represents the ground plane of the pcb). The walls of the EM box are now an= issue because they do not exist in the actual circuit but are necessary to perform the EM simulation. In this ca= se some experimentation may be necessary to find the height of the EM box walls and the distance of the walls from the nearest circuit metal such that the = EM simulation will most closely agree with measurement.

For the filter in this example, the dimensions of the = EM box primarily affected the notch frequency.&nb= sp; The dimensions of the box can be adjusted to get good correlation wi= th measurement (as was done here) and then used as a template for other like circuits. The Sonnet simulator offers several via models including circular, rectangular and edge vias. The type, size and location of the ground vias used to short the coupled lines = also affect the notch frequency. F= or example, if the coupled lines terminate at the center of their shorting vias then the notch frequency will be slightly lower than for the case where the ground via is wholly contained within the end of the trace.

One way to “calibrate” the simulator is to= lay out more than one circuit on the printed circuit board with each circuit va= rying by some simulation parameter (such as EM box size) then noting which simula= tion parameter value created the circuit that measures most closely to the design goals. Figure 11 shows two ve= rsions of the notch filter built on one pcb. The lower filter circuit, created from the schematic in Figure 9 (pcb layout section shown in Figure 10), produced measurement results most close= ly in accord with the simulation.

Summary and Conclusions

The notch filter performance as measured by the HP-872= 0D vector network analyzer is displayed in Figure 12 (dark traces). The EM simulated filter performanc= e is plotted on the same graph in Figure 12 using light traces. As can be seen from the measured v= ersus simulated plots, overlaid on the same graph, the first article prototype’s performance follows that of the simulation very closely.<= span style=3D'mso-spacerun:yes'> The depth and frequency of the not= ch at maximum attenuation cannot even be distinguished as different for the measu= red and simulated cases since their plots almost exactly overlay one another. <= /p>

The measured versus simulated values for depth of notc= h and frequency at maximum attenuation is -42.83 dB at 1489.5 MHz and -41.78 dB at 1489.8 MHz respectively. A co= uple of things (such as the height and width of the EM box) will affect the EM simulation slightly, but with a little experience (in addition to guidance = from the Sonnet manual) the user will know how to appropriately size the box for good accuracy.

The cost of prototype pcb services and fabrication ser= vices (if there are components to be mounted on the circuit board) are significant. Even more costly= is the time it takes to spin a new iteration of the design, including testing, laying out a new board with changes and going through another pcb/fab cycle. Thus, circuit designer= s are always on a quest for first-time-right design methods that really produce prototypes that meet the design goals the first time.

The example presented here demonstrated the possibilit= y that first article prototypes can meet design goals when Sonnet EM simulation is employed to enhance the analysis capability of circuits design with LINC2 circuit synthesis software. L= INC2 does not have built-in EM simulation capability and Sonnet software does not have the circuit synthesis capability of LINC2. However, together these two tools provide more power for the designer than either one alone.

Moreover, LINC2 leverages the power of Sonnet EM simul= ation through a seamless interface to Sonnet.&nb= sp; LINC2 automatically creates the filter schematic from user input. One button click automatically cov= erts the distributed filter elements to microstrip or stripline and one button c= lick automatically lays out the physical circuit in the LINC2 Geometry Layout wi= ndow (Figure 6).

From the Geometry Layout window, one menu pick sends t= he physical design to Sonnet. Th= e user does not have to start Sonnet or even have it running. The LINC2 interface will automatic= ally start Sonnet if it is not already running, transfer the design to Sonnet an= d automatically set up all the EM project, physical and analysis parameters including analy= sis frequencies, physical units, dimensions, dielectric and metal layers, EM box and cell sizes, reference planes and ports etc.

Any of the Sonnet simulation parameters can be adjusted within Sonnet (including the ones imported from LINC2). The EM cell size or metallization = loss models, for example, can be selected and adjusted. However, in many cases no further = action need be performed because LINC2 will have already set up the appropriate entities within Sonnet without the need for user intervention. The only action that is needed to = get EM simulation results for the filter is to select Analyze from the Sonnet Proj= ect Editor’s Project menu; the software is really that easy to use!

Many people, in all disciplines, hang on to their old = and familiar software for fear of having to learn how to use a new and often complex and difficult to use software package, even if the new software has powerful advantages. EDA (electronic design automation) software should have plenty of automation (t= asks that are actually performed automatically for the user). Real design automation is what the= LINC2 software suite provides in abundance.

The LINC2 Filter software used in this article is one = of the many circuit synthesis products that are part of the LINC2 suite of circuit design and synthesis tools from ACS (Applied Computational Sciences). LINC2 offers exact circuit synthesi= s, schematic capture, circuit and system simulation, circuit optimization and yield analysis in a single affordable design environment. More information about LINC2 can be= found on the ACS web site at www.app= liedmicrowave.com.

References

[1] “About Sonnet,” About Us, September 1,= 2014, http://www.sonnetsoftware.com

Sonnet is a tr= ademark of Sonnet Software, Inc.

All trademarks= and registered trademarks are the property of their respective owners.