One of the most common reasons to use an RF enclosure from the Gquipment Mini series is to house some RF/microwave PCB to create an RF module with some specific function. To evaluate the design criteria of this function, one must measure the RF module—for example, by measuring its S-parameters with a VNA using the RF ports of the RF Enclosure Mini. Then, the RF PCB is the DUT, and the RF Enclosure Mini is used as a test fixture.
Of course, the influence of the RF ports should be minimal; however, in practice, there will always be some degree of signal degradation. In this post, we will examine the RF port construction of the RF Enclosure Mini and assess the signal degradation created by it. For this purpose, we will mainly focus on return loss.
We start by taking a detailed look at the RF port construction of the RF Enclosure Mini so we can create a model of the RF ports. An RF port is constructed with a female SMA connector with an extended isolator that penetrates the RF Enclosure Mini’s aluminum wall. The extended isolator exactly matches the wall thickness. The contact pin that travels through the extended isolator forms a coaxial transmission line with the hole in the aluminum wall. The contact pin itself is somewhat longer than the wall thickness, so it can be used as the contact to solder on a trace of the PCB. Of course, all of this is much more easily understood by looking at the images below. The construction—using an SMA connector with an extended isolator—makes it possible to easily replace a test PCB, which is one of the main features of the RF Enclosure Mini.
The RF Enclosure Mini supports three different SMA connectors, as shown by table 1.
Table 1: SMA connectors that are supported by the RF Enclosure Mini
|SMA connector||Frequency (GHz)||Material|
|6 GHz SMA connector||6||Brass body|
|12.4 GHz SMA connector||12.4||Brass body|
|18 GHz SMA connector||18||Stainless steel body|
From an RF point of view, the 12.4 and 18 GHz connectors behave very similarly. Therefore, in this post, any reference to 12.4 GHz connectors includes 18 GHz connectors.
When we take a look at figure 1, we can identify three reference planes that are important to the performance of the SMA RF port. Reference plane 3 also defines the start of the SMA contact pin land pattern. This area and its design play an important role in the overall port performance that cannot be neglected.
The first reference plane is at the spot where the coaxial feed line meets the SMA female receptacle. This is the reference plane that defines the entry point of the RF port. The second plane is located where the SMA connector flange meets the enclosure wall. And the third one lies at exactly the outer end of the extended isolator, which coincides with the other side of the aluminum enclosure wall.
Performance between the first and third reference plane primarily depends on the SMA connector that is used. The land pattern, however, depends on many factors and is therefore rather difficult to control. The most important parameters that influence the design of a proper land pattern are the following:
Figure 2 shows all these parameters.
Finally, for best results, one should understand that the SMA connector and its land pattern together form a system.
We are looking for the best possible RF port performance. In S-parameters, that means an insertion loss of 0 dB and a return loss that is as low as possible. This can be achieved only if we pay close attention to the following general guidelines:
To achieve all of this, one must pay close attention to the SMA connector selected and the design of the land pattern. However, before we dive into the design process itself, we will examine the performance of the SMA port without taking the land pattern into account (that is, looking only at the signal path between the first and third reference planes). This will tell us what the maximum achievable performance will look like.
To find the performance of a single isolated SMA port of the Enclosure Mini, we first looked at the performance of the receptacle only (the signal path between reference planes 1 and 2). We took two SMA connectors with an extended isolator cutoff and affixed them back-to-back (figure 3). That way, we could measure the influence of the receptacle only. Of course, this gave the performance of these two parts together. However, we are interested in the performance of a single connector. Therefore, we made a simulation of the two parts, fine-tuned and validated the model with real measurements, and then used the model parameters to simulate the performance of a single connector part (the receptacle, in this case). This approach of validating a model to find the performance of an isolated SMA port is taken throughout this post.
In the next two figures, you will see the results of the SMA receptacle measurements. They were done for both the default 6GHz SMA connector that comes with the RF Enclosure Mini and the optional 12.4GHz connector.
The difference between S11 and S22 in figure 5 is believed to be caused by imperfections in the back-to-back construction. The inferior performance of the 6GHz connector is caused by the fact that it has a slightly thicker inner contact pin diameter, which lowers its impedance. The data were used to fine-tune a model of the receptacle (figure 6 and 7), which is nothing more than a transmission line, after which the performance of a single receptacle was simulated (figure 8 and 9). The data show that an individual receptacle of the 6GHz SMA connector adds a return loss of 25dB at 6 GHz.
If we follow the same procedure for the 12.4 GHz, we get the following result for the single 12.6GHz receptable.
Of course, an SMI port is more than a receptable, so, to find the performance of a single SMA connector, we followed a similar approach but used two SMA connectors, including the extended isolator. We created a test setup by cutting two walls off an RF Enclosure Mini (each with an SMA connector hole), placing two connectors in it, and again placing them back-to-back (figure 11). Note that we removed the contact pin in this setup. This construction was again modeled (in a way similar to what is shown in figure 6). After calibrating the model with measurement data (figure 12), we ran a simulation with just one port to estimate its performance. These tasks were again done for both the 6GHz and the 12.4GHz connectors. This gave us an indication of the best possible SMA port performance, as seen at reference plane three (figure 1). Because the procedure to isolate a single port is done with a simulator, we can, without any extra effort, simulate the response over an extended frequency range as well. That explains why the single port results show the entire range from DC to 18GHz.
The results show that the default 6GHz connector port delivers a better than 20 dB return loss. The 12.4GHz SMA has an excellent performance of better than 25 dB return loss for a single connector.
Finally, there is one important remark to be made about the construction from figure 11 to generate the data. It was discovered that there were some slight alignment mismatches in the construction on the contact pin area as in the parallel walls. This probably degraded the return loss measurements. However, there was no room left to improve the construction. Maybe later, in a follow-up.
Although the maximum achievable results look great, they are hard to achieve in practice. This is due to the many complications that arise when taking the SMA connector’s land pattern into account. The construction of this zone, together with the design of the RF Enclosure Mini with its table mount systems, sets the upper limits of the RF port. We already had a quick look at the construction of the land pattern with its critical parameters in section 2.
Taking the same approach, we again singled out the performance of an SMA connector using a back-to-back construction and simulation. However, this time, the back-to-back construction was implemented by using an RF Enclosure Mini with a PCB acting as a “through” (just a straight coplanar wave guide).
Because this “through” has some length and the SMA connectors reflect RF energy, the return loss charts will look quite different. This effect is explained in the next section. After that, we will be ready to take the effects of the PCB land pattern into account.
The PCB transmission line has a strong effect on the return loss chart. This effect exists because the return loss of a port will be influenced by the energy bounced back from port two (and vice versa). These signals will act like vectors. Because of the length of the PCB transmission line separating the two ports, the signal from port two is delayed, forcing its phase to increase with frequency.
This means that from DC upward, the total reflected energy will have maxima and minima across the spectrum. The spacing is determined by the length of the PCB transmission line and the wave velocity of the substrate (i.e., its dielectric constant). Note that this is no different from the arrangement where the connectors were placed back-to-back, except in that case the length of the transmission lines was much shorter, which spaced the minima and maxima much farther apart in the frequency domain.
A model of the two SMA connectors, the aluminum walls, and the PCB with its transmission line is shown in figure 14. Again, a linear model is calibrated using measurements for both types of SMA connectors. Please note that if the PCB transmission line is not well matched, it will generate a bad return loss by itself, which would mask the SMA connector’s influence on the return loss and its land pattern.
Finally, the performance of a single SMA connector is simulated using the linear model of figure 15. As stated earlier, the SMA land pattern is not taken into account in this model.
The plots below show the results of the simulation of the linear model. We chose a GCPW with an FR4 substrate to model the transmission line, but another model would also work.
From these plots, one can observe that the results are a superposition of the SMA connector response (the “overall” trend, as seen in figure 13) and the maxima and minima created by the interaction through the transmission line. The plots also show that the 12.4GHz SMA connector manages to deliver a return loss of 20 dB or better across the entire frequency range, while the 6GHz connector cannot do so. Please note that the performance shown is that of both SMA connectors interacting with each other, as pointed out in previous sections.
Until now, the focus was on isolating the performance of one SMA port and explaining what to expect when you connect two SMA connectors with a transmission line as a through using the RF Enclosure Mini. But in reality, the SMA pin’s land pattern plays an important role in the overall performance of the system. Things like how close the signal and return currents are, how the EM fields are aligned, and, of course, the inevitable introduction of impedance mismatches will all be important.
The only way to predict all these effects is to use a 3D EM field solver. For this purpose, an accurate 3D model was created (figures 2 and 14). This model covers all the x, y, z dimensions of the aluminum walls, the PCB, the contact pin with its land pattern, the SMA connector with its extended isolator penetrating the aluminum enclosure wall, and the different materials such as the PCB substrate, the aluminum of the enclosure’s walls, the copper traces, the vias, the Teflon isolator, and the surrounding air. The EM simulation was used to predict the S11 and S21 parameters over a frequency range from DC to 12 GHz.
5.2.1 Results for the 6GHz SMA Connector
The next two figures show the results for the RF Enclosure Mini using the 6GHz SMA connectors and a grounded coplanar wave guide on a FR4 substrate with a thickness of 0.86 mm (image 3). The impedance of the transmission line is controlled within +/- 10%. As can be seen from figure 14, the SMA pin’s land pattern deviates from the standard transmission line layout in an effort to keep the impedance discontinuities at a minimum. The simulation results are shown in figure 18. These results are validated by the measurements from figure 19 for the frequency range from 0 to 6 GHz.
The results from the EM simulator show quite a different story then the linear model (figure 16). Although they both correctly predict the frequency response with its maxima and minima, the EM model also shows that the return loss performance at higher frequencies (> 6 GHz) will deteriorate a lot. The aforementioned parameters play an important role at higher frequencies!
It is instructive to look at an animation of the actual EM wave traveling through the system (animation 1). In this animation, an EM wave is injected from the left side into the system. All energy that reaches the far-right side is completely absorbed. This animation shows that there is EM leakage at reference plane 3 where the SMA connector ends (figure 1). Also, after the initial excitation has finished, the reflections at the two SMA ports become visible.
5.2.2 Results for the 12.4GHz SMA Connector
In the next figures, the results are shown for the 12.4GHz SMA connector.
The results show that the performance once again follows the same pattern as predicted in the linear simulation from figure 17. Also, the EM simulation (figure 20) predicts that the performance will decrease at the higher frequencies (> 8 GHz) range. But this time the degradation is much less severe. This is due to a much better impedance match of extended PFTE insulator in the aluminum enclosure wall of the SMA 12.4GHz connector, among other factors (such as differences in the exact pin diameter and the clearance in the z-direction between the contact pin and the PCB). Figure 21 shows the results of the measurements of S11 and S21 over a frequency range from 0 to 6 GHz.
Animation 2 shows how the EM wave travels through the system.
We already said in section 2 that S11 and S21 performance depends on many parameters. In this section, we will examine one of them in more detail: the clearance in the x-direction (figure 2). We define this clearance as the sum of the clearances between the aluminum wall and the PCB (air gab) and the copper clearance at the edge of the PCB board. The air gab is measured from the aluminum wall (which coincides with reference plane 3 in figure 1) to the PCB edge. The PCB copper clearance is the distance between the edge of the board and the beginning of the copper trace (or plane, for that matter).
We created two different situations, one with a total clearance of 0.5 mm and another with a clearance of 0.2 mm. The 0.5 mm situation is a configuration one would encounter when using (typical) default settings for the copper clearance at PCB edges and some safe distance for the air gab (allowing for PCB board tolerances). The 0.2 mm situation was created by eliminating the PCB edge copper clearance altogether (which is actually quite a common practice in RF board design).
The air gab clearance was set to 0.2 mm although the dimension tolerances of the RF Enclosure Mini are tight enough to design for a 0.1 mm air gab.
The EM simulation results are shown in figure 22 and 23. These are the results for the 6GHz SMA connector type. Intuitively, one would expect that a larger clearance would degrade the return loss performance due to the impedance mismatch that is created by the contact pin that is now running “on its own” between the coaxial part that stops at the aluminum wall and the transmission line on the PCB. It will start to act more like a small wire adding inductance than like a transmission line.
The results show that the expected return loss degradation indeed happens but only for the higher frequency ranges (approximately > 6 GHz). Below that range, the return loss performance improves. The influence of adding clearance is stronger for the 0.5 mm case. This all makes perfect sense. The effect can also be produced by running the linear simulation from figure 15 while setting L1/L2 to some value that mimics the inductance of the SMA contact pin bridging the enclosure wall and the PCB.
So, the 6GHz SMA type gains a lot by using the 0.5 mm clearance for its designed frequency range from 0 to 6 GHz. Above this range, the opposite is true, but this connector should not be used at that frequency range anyway.
When using the 12.4GHz SMA connector type, there is not much to be gained in return loss performance by holding to the 0.5 mm clearance. Also, the improvement is now limited up to 4 GHz. However, when we now use a clearance of 0.2 mm, the return loss performance improves significantly for frequencies above 4 GHz (figures 24 and 25), while the degradation in performance below 4 GHz is very modest. So, when using the 12.4GHz connectors, one would design for a clearance of (at most) 0.2 mm.
The return loss of the RF Enclosure Mini ports, implementing a transmission line (GCPW), are summarized in table 2 for the different SMA connector types available. This overview is helpful to understand what performance to expect in this particular setup. As explained in section 4, these return loss figures are not equal to the SMA’s individual performance, but they are good indicators and easy to verify.
Long story short: for frequencies below 6 GHz, the SMA 6GHz connector type gives the best performance and is best used with a clearance of 0.5 mm in the x-direction. For all frequencies above 6 GHz, only the 12.4 and 18GHz rated connectors are viable options. Also, in this case, the clearance in the x-direction should be 0.2 mm or less.
Table 2: Performance overview of a GCPW with different SMA connectors using the RF Enclosure Mini
|Enclosure type||SMA connector||x-clearance (mm)||Frequency span (GHz)||Return loss|
|RF Mini||don't care||0.5 or smaller||0 - 1||>= 25 dB|
|RF Mini||6 GHz||0.5||0 - 3||>= 25 dB|
|RF Mini||6 Ghz||0.5||0- 6||>= 20 dB|
|RF Mini||12.4 Ghz / 18 Ghz||0.2 or smaller||0 - 10||>= 20 dB|
|RF Mini||12.4 Ghz / 18 Ghz||0.2 or smaller||0 - 12||>= 18 dB|
Until now, we have focused mainly on a small number of parameters that influence the RF performance of the system. But, as already mentioned in section 2, there are many other parameters to consider as well. We discuss the most important ones in this section.
Clearance in the Z-direction.
As shown in section 5.3, smaller is not always better. It will depend on the frequency range you are designing for. The examples used in this post with the SMA 6GHz connector all use a PCB with a thickness of 1 mm and a pin diameter of 0.9 mm, which corresponds to a clearance in the z-direction of approximately 0.3 mm.
Again, when planning for higher frequencies, the SMA 12.4GHz connector is the way to go, and clearance in the z-direction ideally should be 0. In the examples in this post, the 12.6GHz SMA’s pin diameter of 1.3 mm in combination with the PCB thickness of (approximately) 1 mm indeed leads to a zero-clearance situation.
Select the right SMA connector type
Below 1 GHz, it all doesn’t matter much. Performance will be more than adequate. Up to 6 GHz, the 6GHz connector in combination with some clearance in the x-direction (0.5 mm) will give the best results. Above 6 GHz, the 12.4GHz is the only viable solution. Also, clearance in the x-direction must be small (0.2 mm or smaller).
Control your impedance
Obviously, if you are designing for an excellent return loss, you must control the impedance of your transmission line to be 50 Ohms with a tolerance of +/- 10% or better. For frequencies above 6 GHz, Rogers substrate (or the equivalent) is recommended.
SMA contact pin land pattern
It is hard to design an SMA land pattern that will work in all situations because it is dependent on the selected connector, PCB substrate, and geometry of the transmission line used. However, the following general remarks can provide some guidance.
A good design will minimize any impedance discontinuity. One way to achieve this is by treating the SMA land pattern as a transmission line itself and designing it for 50 Ohm impedance too. Best results are to be expected when the SMA contact pin diameter equals the transmission signal trace width. Often, there must be a little extra spacing added between the signal trace and the top ground plane to compensate for the additional capacitance created by the contact pin. Don’t overdo it because if you do, you will end up with an inductive system. If you have access to an EM simulator, it is a good idea to first simulate your design before submitting it to a PCB manufacturer.
It is obvious that one should keep soldering to a bare minimum because it will change the properties of the contact between the SMA connector pin and the PCB trace. But below 3 GHz, you will probably not notice any effect (well, unless you make a mess of it, obviously).
In this article, we showed how to estimate the performance of a single SMA connector and also studied the performance of a transmission line build using the RF Enclosure Mini system. The data and insights provided should give the RF design engineer a good idea of what performance to expect from the RF Enclosure Mini system.