Engineering Note: Interconnection of Coaxial Components

Many present day electronic sub-systems utilize multiple RF functions in integrated assemblies. However, there are still many subsystems where single function RF and microwave components interconnected with coaxial calbes at the next higher assembly (NHA) are better suited to the application. Off-the-shelf components can be used, maintainability and repariability is supported, and design changes and upgrades are more readily implemented. When selecting single function components, engineers should resist the tendency to use a common connector series throughout RF component manufacturers select the optimum connector series for a given component based on the circuit function and topology. This yields the most cost-effective design utilizing readily available components with the best performance and reliability. These standard parts are what engineers should design around.

Assume a converter design as shown below. An input signal at 3.5 GHz is required to be downconverted to provide a 75-ohm real-time output at 1 GHz to an operator manned console. The end user has specified that type N bulkhead connectors provide the input interface, but the output interface is under the subsystem designer's control. Research of readily available components would show that the coupler, input amplifier, and filters are most readily available with SMA jacks. Standard mixers are available with either type N or SMA connectors. However, a mixer with SMA connectors is smaller, and provides flexibility when interfaced with the filters. This is because the filters selected are tubular designs, which allows them to be directly interfaced with adjacent components without cables (cables can be used for more robust environments). Readily available detectors use a SMA plug on the input, and a BNC jack on the video output. The output amplifier is readily availalbe in SMA and BNC. BNC is selected to interface (again can be done without a cable) with a tubular 50/75 ohm matching transformer that is standard with a 50 ohm BNC connector on the input side, and a 75 ohm BNC jack on the output.

In summary, the approach outlined will result in a cost-effective design with a short implementation time. The interconnecting cables serve as the between series adapters.

Illustrative Subsystem
Input

Engineering Note: Semi-Rigid Cable Design

MIDISCO has been manufacturing semi-rigid cable assemblies for over 25 years. During this time, we have developed a number of guidelines for the design and dimensioning of cable assemblies that will provide the user with assemblies that are cost-effective, perform well, and are reliable. Some of them are summarized here, and are illustrated in the drawing below. Additional details can be obtained from the factory.

1. Make drawings full scale.

2. Use tight dimensioning, but loose tolerances. Tolerances of +/-0.030” are generally recommended. Cables (except very short ones) are slightly flexible, and can be “walked” into next higher assembly.

3. Dimension lengths as follows: (a) Straight plugs and jacks: To the reference plane. (b) Bulkhead jacks: To the bulkhead mounting surface. (c) Right angle connectors: To the centerline of the mating surface.

4. Avoid right angle connectors when ever possible. They are more expensive than straight connectors and perform poorly at high frequencies. MIDISCO can bend cables tightly enough so that a straight plug will have the same profile as a right angle connector.

5. Avoid bulkhead and panel mounted connectors if possible, to preclude the need for expensive tooling to ensure correct connector orientation.

6. Dimension straight lengths to the start of bends. A radius to the inside of the cable, and associated angle should be used to specify bends. Radii to the center- line of cables are not measurable. Dimensions that cannot be measured should be avoided.

7. Design bend radius as large as possible, and make them identical if you can. Minimize the number of cable bends wherever possible. Design cable paths for utility, not appearance. Use gradual S bends, rather than straight runs followed by 900 bends.

8. Use stress relief loops whenever possible on short cable assemblies.

9. Avoid direct marking. Use hot-stamped, heat shrink tube.

Engineering Note: 50 Ohms – How Come?

INTRODUCTION

Most electronic systems have evolved around a characteristic impedance of 50 ohms since the 1930’s. This standardization to 50 ohms began with attempts to develop coaxial cables to handle the higher power systems that were rapidly evolving. The quick answer to the “How Come?” is that the 50 ohm “quasi-standard” developed as a compromise between power handling capability and low insertion loss and was based on evaluation and testing on the air-dielectric coaxial cables that were in use at the time.

BACKGROUND

Insertion loss per unit length of a cable is a function of a number of factors including conductor area and the ratio between the diameters of the inner and outer conductors. It was shown mathematically in the 1930’s that the insertion loss for an air dielectric cable was at a minimum for a characteristic impedance of about 77 ohms. Similar calculations showed that the power handling for the same cable type was limited by voltage breakdown and it was then determined that 30 ohms was the best characteristic impedance for optimal power handling.

The arithmetic mean (sum of n data points, divided by n) between 30 ohms (best power handling) and 77 ohms (lowest insertion loss) is 53.5 ohms and the geometric mean (the nth root of the product of n data points) is 48.1 ohms. The geometric mean will indicate the central tendency or typical value). Thus, the 50 ohm choice developed as a result of a optimization tradeoff between power handling and insertion loss for an air dielectric cable. Does this rationalization still stand up today?

LATE 1930’S WORK

The best way to increase the power handling capability of an “air” cable is to fill it with a solid dielectric which has a much higher breakdown voltage than air. This usually causes the connectors to become the limiting factor for voltage breakdown and power handling. The most common solid dielectric in use today is PTFE (Teflon) with a minimum loss occurring near 52 ohms. So it was quite by accident that when we use 50 ohm semi-rigid cables with solid PTFE (Teflon), they give nearly the lowest loss possible. The interesting thing is that PTFE was not invented until 1938 by Roy Plunkett¹, well after the 50 ohm standard was in place.

AND NOW 75 OHMS

So, why 75 ohms now? Commercial cables like the type that bring CATV to your home don’t have to carry high power, so the key characteristic is low insertion loss. The answer to the “Why 75 ohms?” now becomes obvious. It was shown above that a 77 ohm cable gives the lowest loss, so the current 75 ohm CATV standard appears to be just a convenient round-off.

1: A research chemist @ E.I. du Pont de Nemours and Company who accidentally invented Teflon.

Technical Note: Do You Really Need Semi-Rigid Cable?

INTRODUCTION: Copper-jacketed semi-rigid cable is usually the first choice of RF and Microwave Engineers for critical high frequency transmission line applications. This is because of its low loss, repeatability and amplitude and phase stability. Unfortunately, the copper-jacketed semi-rigid has certain disadvantages. It is difficult to work with which also makes it expensive to use.

DISCUSSION OF OPTIONS: Copper-jacketed semi-rigid (Ref: MDC5250, 5141, 5085) uses a seamless copper tube as the outer conductor. Detailed drawings are often needed to define bends and special tooling can be required. Once these types of cables are bent, it is very difficult to straighten and rebend them without damage and/or performance degradation. This can then lead to the need to scrap and replace cables, further adding to the cost.

Aluminum-jacketed semi-rigid (add suffix A to copper semi-rigid cable model numbers to specify an aluminum jacket) is a viable alternate to copper-jacketed semi-rigid for certain applications. It is hand-formable, can be reshaped and it is 40% lighter in weight than copper. However, for small bend radii, aluminum will develop wrinkles and can shear. For cables with complicated bends, detailed drawings and tooling can still be required. Note also that soldering to aluminum is difficult and since aluminum has lower conductivity than copper, attenuation tends to be somewhat degraded.

There are also truly flexible high frequency cables that provide another alternate to semi-rigid cables (BlueFlex Ref: MDC8401, 8402, and 8085). They use an inner shield of silver plated copper/mylar laminate plus an outer shield consisting of silver plated AWG 40 copper braid. The outer jacket is blue FEP. The advantage of using this type of cable is that the user needs only to specify the length he needs and then install in his system. Cable to connector termination is somewhat complicated because of the loose braid. The cable has to be trimmed back to expose the braid and then carefully dipped in a solder pot to freeze the braid before soldering it to the connector. Standard connectors designed for semi-rigid cables are used for the equivalent BlueFlex cable. Flexible cables have little memory and they will not retain their shape when bent and often require hold down points when installed in the next higher assembly (NHA). In specific applications, these factors may negate the cost advantages of using them as replacements for semi-rigid cables.

Perhaps the best alterative to semi-rigid cable is the MIDISCO Ultra-Flex (AKA conformable or Semi-Flex, Ref: MDC8250, 8141 and 8085). The difference between Ultra-Flex and semi-rigid cable is primarily in the outer jacket material. In place of the seamless tube used in semi-rigid, Ultra-Flex uses a composite of copper foil tape and high strength braid material, which is then tin dipped to solder the foil and the braid together. The result is a cable that provides virtually the same performance as semi-rigid cable, but doesn’t require detailed drawings or special tooling since it can be shaped by hand. It is also less susceptible to damage and metal fatigue after flexing and bending. Ultra-Flex can also be used to 2000 C (semi-rigid is rated only to 1250 C) since the outer jacket serves as an expansion joint to resist outer jacket damage. Standard connectors designed for semi-rigid cables are also used for the equivalent Ultra-Flex cable. For applications requiring a non-conductive outer jacket, Ultra-Flex is also available with an FEP jacket over the metallic outer conductor.

CONCLUSION: Each of the cable types discussed herein has their advantages and MIDISCO can supply bulk cable and connectorized cable assemblies encompassing all of the options. However, for users looking for an alternate to semi-rigid, Ultra-Flex is probably the most attractive option. It offers the same performance as semi-rigid, the same ease of termination and a wider operating temperature range.

Engineering Note: Wilkinson N-Way Stripline Power Divider

The insertion loss of an Wilkinson N-Way Stripline Power Divider is specified as the additional loss above the theoretical split. As an example, shown below is a simplified block diagram of a Stripline 8-Way Wilkinson Power divider (MDC2890 reference) which accepts a single input and divides it into eight equal amplitude, in-phase outputs. It is a symmetrical device made up of a cascade of seven 2-Way Dividers [the number of 2-Way Dividers required for this type of design is N (-) 1], where N is the number of outputs required. The red highlights in the diagram trace the input signal through to one of the eight identical output paths.

Each 2-Way Power Divider has a theoretical loss of 3 dB. Based on this and the fact that there are three Dividers between the input and each of the eight outputs, the loss through each path is 9-dB. This is not insertion loss. The input has simply been divided into eight equal amplitude signals. Above this theoretical split, we have to add the “real” losses associated with the internal microstrip line losses, internal and interface mismatch losses and temperature sensitivity. These losses are why the MDC2890 is specified with an insertion loss of 2.5 dB maximum (typically less at ambient) above the 9-dB attributed to the cascading of three 2-Way Dividers.

As shown in the figure, a 4-Way Divider will have a theoretical loss of 6-dB, since only two of the 2-Way Dividers are in each signal path. Similarly, a 2-Way Divider has a theoretical loss of 3-dB. In each case, the “real” insertion losses are as described above. Obviously, the 4-Way Divider loss will be less than that of the 8-Way, and the loss of the 2-Way will be proportionately less.