Connector effects and reference plane influence in the N-Type connector

In this paper we outline the impact of mixing slotted N-type and slotless N-type connectors during a vector network analyzer calibration. We want to point out that making precise measurements of typical transfer standards, while using both N-type jack connectors is very difficult, because of the reference plane shift. We performed simulations of the two connector jack geometries to estimate the impact on measurements of devices under test with either plug or jack interface. The influence of mixing slotted and slotless connectors is then validated through simple measurements, which can be easily repeated in many labs. The results show that mixing of slotted and slotless jacks adds errors in the same order of magnitude sometimes reported as measurement uncertainties by national metrology institutes.


Introduction
To achieve low measurement uncertainties in coaxial measurements of scattering parameters (S-parameters) [1] with a vector network analyzer (VNA), the connector effects have to be considered [2,3] during calibration.For some standardized coaxial interfaces different designs for the jack connector, with the most significant difference being slotted or slotless jacks, exist.These two connector types behave differently as simulation and measurement results show in [4].For many coaxial connectors the faces of the outer conductors are aligned with the gap between jack and plug of the inner conductor.This plane is defined to be the reference plane during calibration.
The reference plane in the N-type connector interface is also defined by the outer conductor mating surface, as depicted Original Content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. in figure 1.However, the inner conductor interface is shifted, so that the jack protrudes into the plug side of the connection.Depending on the applied standard [6][7][8] this section has a length between 5.180 mm and 5.260 mm.The plug offset varies between 5.258 mm and 5.360 mm.This not only leads to possible connector damage when standard definitions are mixed, it also leads to measurement inaccuracies, whenever both slotted and slotless jacks are used in a series of measurements, either as calibration standards or as device under test (DUT).Ideally, all connector geometries would be identical.But even for this case the differences in used material (DC conductivity), surface roughness and plating (e.g.gold) remain.
The influence of the standardized diameter tolerances for coaxial connectors is studied in [9] and not further discussed here, though it would obviously increase the impact of mixing different N-type connectors even more.
The effect of the shifted inner conductor interface is quite difficult to characterize precisely during measurements.However, a simple measurement is carried out, to show the impact on results.In section 2 the used connector models are described and simulation results are presented.The approach to the problem is outlined in section 3 and the theoretically inferred impact on measurement results is discussed in section 4. Finally, in section 5 a measurement is carried out to show with a simple example that this impact on actual measurement results exists.

Connector model and simulation
This work is based on simulated S-parameters of slotted and slotless connectors of Type-N.Both connector pairs are simulated using CST microwave studio [10].A sketch of the assumed connector geometries is shown in figure 2. The pin gap definition is similar for both jack types and a value of 25 µm is chosen to obtain a pin depth of 12.5 µm.For the slotted connector a design with six so-called fingers is used.One could just as well do the simulations and comparison for a slotless and a slotted geometry with four fingers, or between two different slotted versions.The geometries of both jack types are not taken from commercial designs and kept rather simple for this investigation.Especially, the inside of the slotless jack is more complex in reality.However, this is more important for fabrication than for simulating the electrical performance.Hence, for the purpose of simulation, a contact is assumed to occur at the position characterized by the hole length.
Table 1 summarizes the connector dimensions and corresponding uncertainties used for the simulation.For each parameter a separate simulation is performed to be able to determine the corresponding sensitivity and therefore the resulting uncertainty component for the overall result.To make a comparison of the simulation results possible, the dimensions for the plug and the slotless jack are the same as for the connector models published by METAS [11].The dimensions of the slotted jack are chosen without a reference (not equal to any existing jack geometry).To make exchanging both connector models after the simulation feasible, the same overall length for jack and pin combined was used for both connector models.Even though the slotless model could be shorter, an overall length of 5.286 mm is used for both complete connectors.In CST a coaxial line section is used at both ports as port extension.However, this does not influence the simulation result.Figure 3 shows S-parameter results for connector pairs with slotted and slotless jacks.It can be seen that with the chosen dimensions the nominal S-parameters of the slotted and slotless connector pairs are very similar.This of course determines the result of this investigation and it should be noted that these connectors are not at all a worst-case choice.The uncertainty with 95% coverage interval is indicated by the transparent areas and originates only from the dimensions in table 1.The uncertainty budget for the reflection coefficient magnitude |S 11 | is shown in detail in figure 4. It shows that the significantly larger uncertainty for the slotted connector is mainly caused by the dimensions of the slots, while the main contribution to the budget of the slotless connector is the pin gap.To avoid further uncertainties and changes due to differences  in material parameters, perfect electric conductor and vacuum are assumed in this work.
To verify the simulation result, the slotless connector model is modified to be shorter and with a pin gap of 15 µm, to exactly match the model in [11].With this, a difference in input reflection of less than 6 • 10 −5 (|S 11 CST − S 11 METAS |) compared to the results obtained at METAS using a custom tool and less than 4 • 10 −4 (|S 11 CST − S 11 µWaveWizard |) compared to results obtained using the µWave Wizard TM [12] are achieved.For the slotted connector no comparison was available, so a The agreement for the transmission coefficients of both models is even better than for the reflection coefficients reported here.

Problem and evaluation approach
To calibrate a VNA, known standards are connected and measured at the VNA reference plane.The so-called error box parameters are calculated using a suitable algorithm, and the calibrated VNA is modeled as a series connection of an ideal VNA and the error box.
If the test port adapter (TPA) is a plug, the jack of the measured objects protrudes into the error box, as depicted in figure 5(a).Therefore the error box changes with each measured object.Obviously, a changing error box causes inaccuracies of the calibrated measurement result.
In case of a jack TPA, as depicted in figure 5(b), the error box remains unchanged.However, the TPA protrudes into the measured object and thereby also influences the measurement result.The result will therefore depend on the TPA used in the measurement and will change, if another TPA is used.
The simulations to evaluate the impact on prospective results are done separately assuming ideal measurement objects with either a plug or jack port.The effect in both cases is caused by the shift of the inner conductor reference plane.An accurate simulation of both the slotted and slotless connector pair is sufficient to characterize the issue.This separation is performed in order to better illustrate the impact on prospective measurement results.Measurement uncertainty calculations were performed in Matlab [14] using the UncLib [15].
In a first step an artificial error box is created, which is then cascaded with the simulated connector pair with both slotted and slotless type in both orientations for plug and jack test ports.The result is then cascaded with six ideal terminations (Reflection coefficients: Open = 1, Short = −1, Match = 0.01, DUT 1 = 0.95, DUT 2 = 0.2, DUT 3 = 0.05), which are constant over frequency and without uncertainty.Thereby, twelve so-called simulated measurement results are obtained, which can be used for an evaluation.No additional uncertainties (drift, linearity, noise, repeatability. ..) are included here.

Results
To better show the impact on theoretical measurement results obtained by simulation, the effect on jack and plug DUTs are shown separately and for each result only the difference to the nominal value of each respective DUT is plotted.The magnitude of the discussed problem is not independent of the assumed DUT and tends to be less significant for high reflects than for low reflects.The uncertainties shown are only caused by the dimensions of the connectors discussed above.No other influences (VNA, cables, etc) are taken into account here.

Case: plug test port
For measurement objects with a jack interface, as depicted in figure 5(a), the jack protrudes into the TPA and is therefore not part of the DUT or calibration standard.The connector is in this case a part of the error box.In the simulated calibration ideal values for the calibration standards are used.However, in this case the connector type is assumed to be different for the measured calibration standards and the DUTs.Hereby, the error box is changed as it would be during a measurement when slotted (slotless) standards are used in the calibration while the measurements using this calibration are obtained with the respective slotless (slotted) object.The result shows by how much the measurement of a DUT with a jack should be expected to change, when using standards with the other type of jack connector during calibration.It is important to observe that, when mixing connector styles, this problem is unavoidable during the measurement.The error box changes and thereby falsifies the results.Figure 6 shows the simulated results of the calibrated measurements on the ideal jack DUTs evaluated as described in section 3. The jack cascaded with the ideal DUT changes the error box from calibration to measurement and thereby causes change of close to 0.0014 in magnitude of the simulated result and close to 0.55 • in phase of the reflection coefficient.However, these deviations strongly depend on the characteristics of the in this case pretty similar connectors and the reflection coefficient of the DUTs.and only three examples are shown here.Figure 6 shows three examples where the results contain no uncertainties, because the simulated measurement results include no uncertainties within this work and the jack type calibration standards do not contain a connector.It should be noted that even though the connector with associated uncertainties are known, it does not show up in the result for jack type DUTs.

Case: jack test port
For measurement objects with a plug interface, as depicted in figure 5(b) the error box remains constant, however the measured value changes depending on the TPA used during the measurement, because it protrudes into the DUT and therefore is a part of the measurement result.For the simulated calibration the ideal standards cascaded with a connector are used and two calibrations are simulated using the slotted and slotless connectors.Subsequently, the resulting error boxes are used to deembed the simulated measurement result of the DUT obtained with the other jack connector type.
Therefore, the result shows by how much the measurement result for a DUT with a plug should be expected to change, when it is measured using a TPA with the other connector style.It is important to notice that this problem only shows up, when the TPA is changed.Therefore, it occurs when results are exchanged between laboratories for example with the characterization data for calibration standards.This data for an N-type plug calibration standard always includes the TPA connector jack, which was used when the characterization was performed.Figure 7 shows the simulated results of the calibrated measurements on the ideal plug DUTs evaluated as described in section 3. The jack of the TPA protrudes behind the reference plane of the simulated measurement result and causes a change of close to 0.0013 in magnitude and 0.77 • in phase of the reflection coefficient for the exact same DUT.So, in this result the difference between the traces with the same color shows the discussed problem.However, as for the plugs, the deviations strongly depend on the characteristics of the in this case pretty similar connectors and the reflection coefficient of the DUTs.Furthermore, only three examples are shown here.The results here contain uncertainties, however these are not the ones of the connector pair used during the measurement, but the uncertainties from the connector pair used during the characterization of the calibration standard.
For both cases, plug and jack test ports, the database for calibrated measurement capabilities reports measurement uncertainties that are of the same order or even lower [16].

Measurement
To confirm the theoretical results measurements are conducted, though since the discussed effect is small, it is difficult to design an experiment to definitively prove it.Here, one port measurements are performed using plug DUTs (case from figure 5(b)) and two different TPAs.Please note, since fictional connectors were used for the simulation, it is not the objective here to achieve agreement of simulation and measurement, but to demonstrate the impact of mixing N connectors.The experiment is conducted in a laboratory with excellent temperature and humidity control, using metrology grade equipment and there are no cable movements in this one port measurement.
Two type N calibration kits are measured, and one is used for calibration, while the standards in the other kit are considered DUTs.This experiment is performed twice using two different high quality TPAs with good repeatability made by different manufacturers.One TPA is slotless with a pin depth of 13 µm while the other one has six slots and a pin depth of 23 µm.Since the same definitions for the calibration standards are used, and the same DUT kit is measured, the results are expected to be identical within the limit of repeatability, if the TPA would be of insignificant influence.
The obtained results for the characterized load calibration standard considered as a DUT are shown in figure 8. Here, eight results are shown.All are using the same calibration standards (and definitions).Four are measured using the slotted TPA (dashed lines) and four are obtained using the slotless TPA (solid lines).A clear difference in the results is observed, which is larger than the connection repeatability indicated by the results using the same TPA.
The obtained measurement results for the short DUT are depicted in figure 9.Here again, the difference between the results obtained using the two TPAs is larger than the observed connection repeatability, while using the same TPA.
As depicted in figure 5(b), the inner conductors of the two different TPAs protrude beyond the reference plane of the measurement and are therefore part of the DUTs.These two different inner conductors of the TPAs have different dimensions and material properties, which explain the differences observed in the results here.The impact on reflection magnitude is in the order of 0.003 for the load and short standard.The reflection phase differs by more than 10 • for the load and by 0.25 • for the short.These values are close to commonly reported measurement uncertainties.However, it is worth noting that the effect of the shifted reference plane is not currently accounted for.

Conclusion
The results in this paper obtained from measurement and simulation show that mixing slotted and slotless calibration standards and DUTs causes inaccuracies, which should be considered, when low measurement uncertainties are desired.All presented results are based upon the specific connector geometries, which are assumed for the simulation and indeterminate during the measurement.In practice all differences in the material properties, such as conductivity and surface roughness, of all used devices and calibration standards can cause similar inaccuracies.These are however not investigated within this work.The observed inaccuracies might be somewhat different (reduced or increased), if the inner conductor mating plane is used as the reference plane, but this has not been investigated here.
To avoid the inaccuracies shown in this work, one could characterize the mechanical dimensions and material properties of all connectors used in a measurement and compensate for the differences by using simulation results of the connector effect.However, this would include all DUTs and thus be a challenging task which would require too much resources for most applications.
Another approach would be to not mix N-type interfaces for calibration standards, DUTs and TPAs.However, this would effectively create separate N-type standardizations and laboratories would need many times the amount of N-type equipment, to be able to perform all N-type measurements.
One further option would be, to always use the same TPA.This adapter would then need to be a fixed part of the used calibration kit and it would always be applied during the measurement of DUTs, reducing the lifetime dramatically.Also, the original problem would of course reoccur, when measurement results are transferred to the next user in another laboratory, who will then need to use their own TPA.

Figure 1 .
Figure 1.Reference plane defined by the outer conductor mating surface (red dashed line) in a mated, slotted N-type connector.Shifted inner conductor interface (blue dotted line).Reproduced from [5].CC BY SA 4.0.

Figure 2 .
Figure 2. Sketch of the inner conductor connector types.(Not to scale.).

Figure 3 .
Figure 3. S-parameters of both connection types with uncertainties (95% coverage).The phase of the transmission coefficients is normalized to the mean.The jack is at Port 1 and the plug at Port 2.

Figure 4 .
Figure 4. Uncertainty budget for the magnitude of the reflection coefficient |S 11 | from figure 3 for (a) the slotless and (b) the slotted complete connection.

Figure 5 .
Figure 5. VNA reference plane illustration with (a) pin test port adapter and jack DUT and (b) jack test port adapter and pin DUT.

Figure 6 .
Figure 6.Effect on the measurement of slotless (slotted) jack DUTs using slotted (slotless) standards during calibration, with dashed (solid) line style.The results represent the difference to the nominal value of the respective DUT.

Figure 7 .
Figure 7. Difference to the respective nominal value for each plug DUT measured with slotless (slotted) test port adapter with dashed (solid) line style.The phase is normalized to the mean for each reflection value.

Figure 8 .
Figure 8. Measurement results for a Load with plug connector.Four measurement results are obtained using a slotted TPA.Four results are obtained using a slotless TPA.The phase is normalized to the mean of all eight results.

Figure 9 .
Figure 9. Measurement results for a Short with plug connector.Four measurement results are obtained using a slotted TPA.Four results are obtained using a slotless TPA.The phase is normalized to the mean of all eight results.

Table 1 .
Dimensions of the connectors used in simulation.