Amphenol RF- Products

Radio, it seems, is going the direction of television. Today in the US, two companies are vying for consumer dollars in the race to put Satellite Radio in the car, on the boat and everywhere in between. For several years now many high-end automobiles have been available with satellite radio factory installed. Today, more models, crossing the entire price range, are available with factory installed systems.

In the aftermarket, consumers can buy satellite radio from retailers across the county. Packages vary widely from special custom installation designs to simple do it yourself packages.

Satellite Radio has its proponents and detractors. Advocates of the technology talk about the virtual commercial free content, coast to coast reception and the list of celebrities that have joined the ranks of SJ's (Satellite Jocks). Opponents talk about the monthly subscription cost, the special equipment that is required and the sometimes limited availability of coverage.

Transmission of the satellite signal passes through a specially designed Antenna to an RF Receiver, and then passed through to the Tuner module installed inside the car. Amphenol RF offers a wide range of inter-connect solutions for the satellite radio market, connecting the antenna to the RF receiver. SMB slide on connectors are the most prevalently used interface, providing a cost effective yet strong performing interface available in a wide variety of configurations. Car manufacturers use the FAKRA line of connectors, an SMB interface combined with a mechanical and color coded housing suitable for the automotive production line.

The long term success of Satellite Radio is still to be determined, though the continued growth of both companies of late seems to indicate the technology is here to stay.

Technology Update: Surface Mount Coaxial Relays

Recently Amphenol RF has developed several coaxial relays for applying directly to a printed circuit board (PCB) utilizing the SMT process. These designs enable the user to move from a frame mounted package design necessitating a labor intensive final assembly into the supporting product. In addition to the physical mounting of these earlier designs, hand soldering of the solenoid power leads was required as well as cabling of the NC/C/NO contacts to the main PCB.

The cavity where internal switching occurs provides for a 50 or 75 ohm transmission line depending on the application. In addition the product can be supplied as either a single pole-double throw (SPDT) or single pole-single throw (SPST) designs.

With an understanding of the pcb characteristics, Amphenol RF can work with the customer to develop a pcb footprint capable of optimizing RF performance at the launch area of the signals. These characteristics include pcb material, thicknesses and knowledge as to whether the pcb designs represents a microstrip, coplanar or stripline configuration.

With this knowledge careful consideration is then given to provide superior results in the areas of isolation, insertion loss and return loss.

The comparative size benefit demonstrates a 50% savings of available real estate. The product can be supplied in tape-and-reel for pick and place equipment resulting in significant total installed cost reduction. All internal components are designed to consider the necessary process temperatures for no lead processes.

Dave Q: "How does Amphenol RF Test Receptacles?"

A: In an earlier issue of Ask Dave, I described how we can compensate for the mismatch created by a receptacle when it is launched onto a board or into a device. However, before we can do that, we need to design the connector for optimum Return Loss performance and test it to confirm that it meets the specifications.

How does Amphenol RF test receptacles? Methods for testing adapters and cable mounted connectors have been around for many decades, and the techniques are accepted throughout the industry. However, receptacles can present problems due to the fact that their output may be anything from a diameter to a flat tab and standards for these varied outputs do not exist. Normally, when designing a connector, we model the connector and optimize the design without considering the extension past the back of the connector. The reason for this is that we do not control how the customer mounts the connector to their circuit and depending on how this is accomplished, various mismatches can occur. So we endeavor to optimize the design of the coax part of the connector. The question remains, how do we confirm that our design is good?

We occasionally have customers specify that we test a receptacle using chip resistors; either a single 50 ohm chip resistor, two 100 ohms chips or even four 200 ohm chips in parallel. Naturally, we will test it as the customer prefers, but for design verification purposes, or if there is any question as to the validity of the results, we prefer to use a specially designed coaxial test fixture that mounts onto the end of the receptacle and provides a continuous and well matched 50 ohm impedance. This may require that we need to make up additional test pieces for test and verification, as this will not work with a tab style connector. The Return Loss of a typical SMA receptacle using a coaxial test fixture is shown below.

You can see that the performance of this receptacle is better than -30 dB up to 12 GHz and -27 dB to 18 GHz.

The Return Loss of this SMA receptacle using one 50 ohm chip resistor is shown below..

You can easily see that using a single chip resistor results in a poor match (reflection coefficient = +80 mU) and really isn't a useful method for terminating a receptacle much above 500 MHz. Note that the Time Domain (S11 Real) scale has been increased to 20 mU/division as compared to only 5 mU/division for the coax fixture.

The Return Loss of this SMA receptacle using two 100 ohm chip resistors is shown below. You can see that the mismatch is now slightly capacitive, with two capacitive discontinuities approximately -5 mU each. This is fairly comparable to the coax test fixture, so it could be useful as a termination up to 18 GHz. However, in this case, the coax test fixture is still about 3 dB better up to 18 Ghz and exhibits less phasing in the Return Loss plot.

A plot showing a comparison of the two 100 Ohm chip resistors vs. the coaxial test fixture is shown below.

The Return Loss of this SMA receptacle using four 200 ohm chip resistors is shown below. The resultant mismatch is highly capacitive (-120 mU) and once again no longer useful as a termination above 500 MHz. It certainly presents no advantage to using one or two chip resistors.

In addition, soldering chip resistors is a very time consuming technique, requires patience and a steady had or special fixturing, and is not an easily repeatable process.

Naturally, for different size receptacles (BNC, N etc.) different coax test fixtures are needed in order to achieve optimal test results. Generally, the mismatch introduced by the chip resistor is so large as to mask the actual reflection of the receptacle under test and it becomes difficult to determine if your design or the termination is at fault. Gating techniques are not very useful as it is virtually impossible to separate the chip resistor reflections from the receptacle reflections.

Also, remember that even though a connector is properly designed, problems can arise at the launch area. And as I discussed in an earlier edition of Ask Dave, it is necessary to match the launch geometry based on the connector back end configuration and the board parameters.

So at first glance it appears that using two chip resistors in parallel can be a reasonable substitute for a properly designed coaxial test fixture. Don't be fooled. Sometimes, using two chips can give results that appear better than the device really is. In the next issue of Ask Dave, I will present an example showing how testing a 75 Ohm BNC receptacle gave very good looking results using two 150 Ohm chips in parallel, but the HFSS simulation and Network Analyzer testing using a coaxial load showed that in reality, the performance was very poor.