Microwave Journal
www.microwavejournal.com/articles/2674-dc-to-110-ghz-measurements-in-coax-using-the-1-mm-connector

DC to 110 GHz Measurements in Coax Using the 1 mm Connector

A look at the 1 mm connector design, the proper connection technique and examples of how 1 mm connectors allow good ultrabroadband performance

July 1, 1999

DC to 110 GHz Measurements in Coax Using the 1 mm Connector

The 1 mm coaxial connector was designed to provide an enabling technology for test systems that work up to 110 GHz. It was introduced as an open standard to the RF and microwave industry through the IEEE 287 Precision Connector Standards Committee in 1989. After nearly a decade of process and product development, a wide variety of test accessories using the 1 mm connector are now available in the marketplace. This article presents a thorough look at the 1 mm connector design, some insight into process and product design of 1 mm components, a survey of the products available from industry and a discussion of the proper connection technique. Finally, examples are shown of how the 1 mm connectors allow good ultrabroadband performance.

Keith Howell and Ken Wong
Hewlett-Packard Co., Microwave Technology Division
Santa Rosa, CA

Ten years have gone by since the 1 mm connector, shown in Figure 1 , was first designed, yet it is just now coming of age. The device is a key building block for constructing an ultrabroadband (DC to 110 GHz) test environment. With the application of the connector and the supporting technology behind it, basic high frequency device research can be conducted with a single connection and a single sweep instead of the multiple setups required in the past. Back in 1989 there was not a great deal of millimeter-wave business, but the technology trends were evident. Julius Botka and Paul Watson of Hewlett-Packard had the vision to design the 1 mm connector as a key building block for future test systems. Upon completion of the design it was proposed to the IEEE 287 Standards Committee on Precision Connectors as an open connector standard. The committee thoroughly reviewed and later approved the connector design.1

Two grades of 1 mm connector are covered in the IEEE standard: the laboratory precision connector (LPC) for applications where lab-grade precision is required (such as calibration standards), and the general precision connector (GPC), which finds use in connectors and cables. The main difference between the two grades is precision of the connector interface - clearance, flatness and pin depth are more tightly controlled on the LPC grade. Users should take note that because of the frequency coverage (DC to 110+ GHz), there is no room to cut corners on the GPC grade - it too must be considered a precision connector. In order to obtain satisfactory broadband performance, particularly in applications where a number of connect/disconnect cycles will occur, 1 mm components should be specified with at least a GPC level of precision.

As can be seen in Figures 2 and 3 , which show jack and socket interface dimensions for 1 mm GPC connectors, the 1 mm connector is named for its outer conductor diameter. The interface is defined with an air dielectric, 50 W impedance and corresponding 0.434 mm center conductor diameter. The jack pin has a 0.250 mm diameter. A slotted socket design is used in both the GPC and LPC grades. With a 0.434 mm center conductor diameter and a 0.250 pin diameter, approximately 0.09 mm of wall thickness remains on the slotted socket (depending on the compensation scheme and clearance used, which vary with manufacturer).

 A side benefit of the connector design is compactness. Coupling between connectors is achieved by means of M4 × 0.7 threads and a nut with a 6 mm hex. This small size is advantageous in millimeter-wave package design because the connector dimensions more closely match circuit geometries.

DC to 110 GHz Calibration Kit Development

Accurate measurements require high quality and accurate calibration standards. Typically, vector network analyzers use opens, shorts and loads as calibration standards. Because low reflection, < 0.03 (-30 dB return loss) broadband fixed loads are very difficult to attain, sliding loads have been used for more accurate high frequency applications. For the 1 mm connector size, a sliding load is impractical. Such a device would be extremely expensive to manufacture and the sliding load element would be too fragile to endure any reasonable usage.

Very accurate high band calibration can be attained using offset shorts. The size, flatness and surface finish of offset shorts may be accurately controlled with high precision machining technology. Four offset shorts were needed to cover the 50 to 110 GHz frequency range. Below 50 GHz, accurate calibration is attained using open, short and load standards.

At 110 GHz, the wavelength for transmission in air is only 2.72 mm. A length change of only 3.78 mm will cause a 1š phase change in a reflected signal. Thus, in order to obtain good accuracy with an offset short calibration, extremely precise control of both the offset length and pin depth is required. The offset length control demands tight tolerances (±0.005 mm) on center and outer conductors. However, accurate measurement of pin depth requires a new approach. The parts on 1 mm connectors were small enough that the deflection and increased uncertainty due to contact measurement techniques consume a major portion of the tolerances required. Optical methods for measurement of pin depth have been developed recently, and this technique was employed extensively for the 1 mm connector development. A Zygo white light interferometer was used to measure the free-state pin depth value. Separate analysis and measurements were performed to characterize the contact deformation of both male and female outer conductors to determine the minimum specified pin depth.

Traceability of a calibration kit ultimately goes back to dimensional standards. Measuring the outside diameter of the center conductors is relatively straightforward, and gages of adequate quality for calibrating the measuring instrument are readily available. However, this was not the case for measuring the inside diameter of the outer conductors. Commercially available 1 mm ring gages were determined to be of insufficient quality to meet the tolerance budget. In-house ring gage standards were developed to satisfy this need. This development was a significant challenge because the goal for the gage tolerance was 20 percent of the part tolerance, and the part tolerance in the case of the LPC grade was ±0.003 mm. However, once this goal was accomplished, the foundation in dimensional accuracy needed for traceable measurements had been established.

The socket center contact design was compensated to 50 W. Analysis was performed using HP HFSS. The center contact was quite a challenge to machine and measure accurately because it was a slotted design with a wall thickness of only 0.09 mm, slot width of 0.06 mm and slot length of approximately 1 mm. The fingers had to be burr-free, of equal size and evenly formed to obtain the desired performance. Once the part was produced, it had to be measured. Considerable thought and energy went into determining a method to measure over the slotted region and considerable testing took place with assembled units to verify the performance electrically.

Precision Offset Short Electrical Performance

The calibration scheme used in the calibration kit calls for an open-short-load calibration over the low band (up to 50 GHz) and an offset short calibration for the high band (from 50 to 110 GHz). The high band is split into two bands. Three offset shorts (with lengths of 1.30, 2.45 and 3.00 mm) are used over the 50 to 75 GHz range, and three offset shorts (with lengths of 1.30, 1.825 and 2.45 mm) are used over the 75 to 110 GHz range. Figures 4 and 5 show how the calibration is stitched together on a network analyzer to create highly accurate directivity, source match and reflection tracking terms in the calibration.

Connector Life Cycle Testing

While technicians have (over the course of time) exercised some 1 mm connectors for hundreds of cycles, it is unreasonable to expect them to deliberately test and disconnect to destruction. Other methods had to be used to test the connector's lifetime. Finite element modeling of the cyclical stresses on the socket fingers was performed for various magnitudes of deflection, which simulated a radial offset between center conductors during connection. The model showed that perfect alignment would lead to near-infinite life for the four-slotted design, but an eccentricity of just 2 mm reduced the theoretical life to 10,000 cycles. While illustrative, this model differed from the real world in several ways. For example, in the real world, the center conductors are supported by dielectric beads, which have some compliance. This added support tends to make the connectors more tolerant of small levels of misalignment. However, another real-world attribute is wear, which tends to reduce connector life. With an idea of what to expect, the next step was a mechanical test bed.

Testing was performed on a cam-driven connector insertion machine with varying degrees of center conductor alignment. With perfect alignment, the machine was operated up to 30,000 cycles. Although the gold plating was worn away, all the features remained intact. As expected, better than predicted life was observed for connectors with small amounts of misalignment because the center conductors could flex slightly on the beads to reduce the stresses on the contact fingers.

Nonetheless, center conductor alignment is critical on the 1 mm connector, both from the standpoint of connector life and electrical repeatability. Thus, the true position of the center conductors is stringently controlled in the connector standard, and users should exercise good practice when making connections (see sidebar). In addition, the connector design standard calls for 1.2 mm of outer conductor engagement before the center conductors make contact, thereby assuring alignment, as shown in Figure 6 .

Connector Repeatability Data

A high number of connector repeatability tests were conducted during the development of the 8510XF network analyzer system. In a representative test, the port was terminated four different ways and 15 connect/disconnect cycles were performed for each version. The data listed in Table 1 represent the worst-case drift up to 110 GHz. In cases where a short was connected, the Data/Memory function was used and, in cases where a load was connected, the Data - Memory function was used. The first two sets of data were recorded at the test port, while the second two sets were recorded at the end of a 10-inch cable. Note that the data taken with the short at the end of the cable were much worse than the data taken at the test port. This degradation was caused by the cable being held by hand during the connection process and the small amount of movement. The data taken with the load were less sensitive than this variation because the load configuration was large and easy to hold in a fixed position during connection and disconnection.

Table I
Connector Repeatability Test Data

Repetition Number

3 mm Short at Test Port (dB)

50 W Load at Test Port (dB)

3 mm Short at end of 10" Cable (dB)

50 W Load at end of 10" Cable (dB)

1

0.0065

-44.109

-0.0480

-47.470

2

-0.0068

-50.123

-0.0230

-49.680

3

-0.0130

-54.951

-0.0243

-46.967

4

-0.0122

-43.873

-0.0507

-46.092

5

-0.0148

-52.018

-0.0854

-45.426

6

-0.0592

-46.357

-0.3655

-43.910

7

-0.0528

-42.727

-0.3577

-47.523

8

-0.0204

-45.225

-0.1695

-42.359

9

-0.0193

-45.514

-0.1161

-43.975

10

-0.0159

-43.166

-0.2127

-44.932

11

-0.0164

-46.117

-0.2164

-42.924

12

-0.0583

-44.848

-0.0819

-46.920

13

-0.0486

-46.066

-0.1633

-46.613

14

-0.0199

-43.113

-0.1859

-42.160

15

-0.0436

-50.781

-0.0970

-47.643

From the test data it can be concluded that the 1 mm connector is very repeatable. In addition, in a real-life situation where a cable is used, fixturing is recommended to minimize cable movement and achieve the highest accuracy.

A Survey of 1 mm Products

Use of the 1 mm connector grew slowly at first. The first 1 mm products were introduced in 1993 - a set of cables and waveguide-to-coax adapters for wafer probing applications.2 Rosenberger, a large German connector manufacturer and original member of the IEEE 287 Precision Connector Standards Subcommittee, introduced its first 1 mm connectors and adapters in 1995. The pace of product introductions increased in 1996 with a special ultrabroadband 85 GHz test system from Hewlett-Packard, coaxial probes from Cascade Microtech and additional test accessories from Rosenberger. Since that time, the usage of the connector has grown dramatically with the introduction of the HP 8510XF network analyzer and a family of 1 mm accessories and new products from a variety of other manufacturers. The 1 mm connector is now firmly positioned to address both the needs of researchers and manufacturers developing products for such markets as 20+ Gbps optical-to-electrical converters, automotive radar and wireless LAN.

Currently, 1 mm products are being used in Cascade Microtech's ACP 110 wafer probes (GPC socket connectors), GGB Industries' probes (available with either jack or socket 1 mm connectors) and Rosenberger's (RPC-1.00) connectors for 0.047-inch semirigid cables and in-series adapters. Figures 7 , 8 and 9 show examples of these products. In addition, Hewlett-Packard has a growing family of 1 mm accessories, including the 85059A calibration and verification kit, 11500I/J/K/L test cables, 11920A/B/C in-series adapters, 11921A/B/C/D and 11922A/B/C/D between-series adapters, 11923A sparkplug launches and V/W281C/D waveguide-to-coax adapters.

Conclusion

Over the last 10 years, the 1 mm connector has grown into its current position as the accepted industry-standard connector for mm-wave measurements. Considerable work has gone into laying the cornerstones of traceability, accuracy and repeatability in the development of the HP 85059A calibration and verification kit and the HP 8510XF network analyzer system. A wide variety of accessories are available from a number of manufacturers today, which enable precise and repeatable 110 GHz measurements in coax and on wafer. The connector design is inherently rugged because the outer conductors engage before the center conductors. However, due to their small size and broadband frequency coverage, the user must pay careful attention to proper connection techniques to realize the ultrabroadband performance that can be achieved using 1 mm connectors.

Acknowledgment

The authors wish to thank Chen yu Chi, Jane Huynh and Mike Pfendler for their contributions to the work reported in this article.

References

1.         IEEE Standard for Precision Coaxial Connectors (DC to 110 GHz), Std. 287-199x (Revision of IEEE Std. 287-1979), IEEE, 245 East 47th Street, New York, NY.

2.         E.M. Godshalk and J. Pence, "Low Cost Wafer Probe Scales 110 GHz Summit," Microwaves & RF, March 1993.

Keith Howell received his BSME degree from California State Polytechnic University, San Luis Obispo in 1980 and was hired by

Hewlett-Packard in 1981. He spent over a decade as a manufacturing engineer developing precision turning processes for the 8510 network analyzer and its calibration accessories prior to transitioning into product development. Howell is presently an R&D engineer responsible for microcircuit development.

Ken Wong received his BSEE degree from California Polytechnic State University, San Luis Obispo, CA and did some graduate-level work at the University of California, Berkeley. Since graduation, he has been with

Hewlett-Packard Company. Currently, he is a senior engineer and project leader responsible for the development, design, manufacturing and measurement of precision microwave calibration and verification standards for vector network analyzers and power meters. Wong's responsibilities also include process developments for compliance with ISO standards.