New SMA Launch Model Michael s Work on the SMA Launch as of 2/04/2009
Table of Contents The Latest Model Michael Enders recently received new information with regards to the viability of the SMA Launch design. This new information required changes to be made to the model. These changes have been reflected in this newest model and will be shown and explained Tuning Stub length, air cylinder radius, fifty ohm pad radius, thru-hole locations
The Model A summary of the changes of this model: 1. The SMA connector is placed on the bottom of the board (it hits the fifty ohm line first) 2. There is no center thru-hole 3. There is no copper pad where the center pin enters the dielectric In many respects, this model is simpler than the previous model. There are less variables so tuning should be easier, assuming the response is receptive to changes.
The Model This screen capture shows the signal flow. The fifty ohm line now has a bulb at its end which is directly connected to the center pin. The center pin actually goes deeper into the air/epoxy block a stub length that is anywhere from 32.3mil to 15mil (it must be large enough to solder to the trace). The air cylinder is shown around the center pin. This keeps the signal from shorting to the ground on the top of the board. The air/epoxy block must have a diameter of 144mil.
Tuning Stub length (in the air/epoxy block) 59646-0001 01 Initial Model - Swept Stub Name X Y m1 100-19.1653-5.00 m2 11.0000-19.1455 m3 12.0000-18.0333-1 stub='15mil' stub='16.73mil' -15.00-2 stub='21.92mil' stub='27.11mil' m1 m2 m3 stub='32.3mil' -25.00-3 -35.00-4 5.00 1 15.00 2 The variable stub was varied. This variable is made so that whatever the value, the stub length beneath the trace is exactly that value. As it is swept, the results don t change much. I kept the 15mil value for the rest of the optometric sweeps, as it seems the best for our range.
Tuning Air Cylinder Radius 59646-0001 03 Swept Air Cylinder Radius -1-2 air_radius_extra='10mil' -3 air_radius_extra='17.5mil' -4 m1 m2 m3 air_radius_extra='25mil' air_radius_extra='32.5mil' air_radius_extra='40mil' -5 Name X Y m1 100-38.4825 m2 11.0000-41.6801 m3 12.0000-34.9111-6 5.00 1 15.00 2 The variable air_radius_extra denotes the distance of air between the ground plane and the center pin of the launch. As this value increases, results get much better. Clearly, the larger this air gap, the better the results. What is the largest it can be without problems? The 40mil value is used for future optometric sweeps as it gave the best results for 10-12GHz
Tuning Fifty Ohm Pad Radius 59646-0001 04 Swept Fifty Ohm Pad -1-2 -3 m3 extra_fifty_pad='0mil' extra_fifty_pad='5mil' -4 m1 m2 extra_fifty_pad='10mil' extra_fifty_pad='15mil' -5 Name X Y m1 100-38.4825 m2 11.0000-41.6801 m3 12.0000-34.9111-6 -7 5.00 1 15.00 2 The bulb at the end of the trace has, as default, a diameter equal to that of the width of the fifty ohm line. The variable extra_fifty_pad increases the radius of the bulb by its value (in mils). As the pad is increased beyond 0mil, results degrade. 0mil is used for further optometrics.
Tuning Thru-Hole Locations (more negative gets closer to trace) 59646-0001 05 Swept Thru Hole Locations -1-2 -3-4 -5 close='-40.3mil' close='-38.3mil' close='-36.3mil' close='-34.3mil' close='-32.3mil' close='-30.3mil' -6 5.00 1 15.00 2 As the variable close becomes more negative, the closest distance from the edge of the thruholes to the trace decreases (therefore, higher order modes should move to higher frequencies). We see this happens with little degradation to the rest of the response.
Tuning close = -50.3mil 59646-0001 06 'close' = -50.3mil -1-2 Name X Y m1 100-55.7209 m2 12.0000-26.5442 m3 11.0000-30.4759 m4 18.0000-14.1836 m4 m2-3 m3-4 -5 m1-6 5.00 1 15.00 2 I tried a single run with the best parameters (so far) but with close = -50.3mil; this is 10mil closer to the trace than the closest one on the last slide. We see the return losses pop up over the 1-15GHz range, but stay below -22dB-ish. No higher order modes, okay at 18GHz.