|
What
considerations should be used when specifying a connector?
There
are several factors that will determine the choice of connector
series and style. Cable and frequency range are the primary
factors. It is good engineering practice to try and match
the connector size (diameter) and cable diameter as closely
as possible in order to minimize reflections. The larger the
difference between cable diameter and connector diameter,
the worse the performance will be. Reflections will generally
increase as a function of frequency, and smaller connectors
will generally perform well at higher frequencies. For very
high frequencies (above 26 Ghz), precision, airline connectors
may be needed.
Frequency
range will determine the connector series used. Click
here to link to see the various connector series and their
standard frequency ranges. In general, it is advisable to
use push-on or bayonet style connectors at low frequencies,
typically below 6 Ghz. Threaded connectors should be used
for high performance, low noise applications.
The
cable that is specified will generally determine the impedance
of the connector used. 50 and 75 ohms are the standard impedances
used by most designers, and several connector series come
in both 50 ohm and 75 ohm versions. Click
here for a listing of common cables and their characteristics.
At frequencies below 500 Mhz or so, 50 ohm connectors can
be used on 75 ohm cable (and vice versa) with acceptable performance
levels. The reason for doing this is that 50 ohm connectors
are generally less expensive due to their greater usage. Click
here to see a list of useful transmission line equations.
Aside
from trying to match the cable and connector as closely as
possible in size to minimize reflections, connector interface
and dielectric materials are also important considerations.
Line to line and air interfaces such as the SMA and Type N
give excellent high frequency, low reflection performance.
Overlapping dielectric interfaces such as BNC and SMB are
usually limited in performance vs. frequency. The usual figure
of merit for a connector’s performance is its Reflection
Coefficient. This is a measure of how much signal is reflected
back from the connector. It can be expressed in terms of Reflection
Coefficient, VSWR (Voltage Standing Wave Ratio) and Return
Loss. Click here
for a Reflection Converter.
Due to FCC Part 15 regulations that require non-standard interfaces
on spread spectrum wireless devices, many designers have chosen
to specify common connector types such as BNC and TNC but
with reverse polarity interfaces. Reverse threaded interfaces
are also sometimes used.
Power
or voltage requirements are also a factor determining the
connector to be used in a particular application. High power
applications will dictate the use of large diameter connectors
such as 7-16 DIN and HN for example. Average power handling
is normally limited by the cable’s power specification
and is usually determined empirically. The voltage breakdown
level of the connector limits peak power. Power handling capability
will diminish as a function of frequency and altitude.
What
is VSWR and how is it specified?
VSWR
is a measure of the amount of signal that is reflected back
from a connector. It is a vector quantity in that it has both
amplitude and a phase component. This is important when considering
the impact of multiple connectors in a transmission line (see
How do I specify the performance characteristics of a cable
assembly?, below). Mismatched impedances cause the reflections.
If the cable being used has a characteristic impedance of
50 ohms, then the connector must also maintain 50 ohm impedance.
The transition from the cable to connector transmission line
sizes and captivation of insulators and contacts are the main
causes of mismatches within the connector.
There
are generally 2 ways in which the VSWR of a connector is specified.
The first is a “flat line limit” over the entire
frequency range. For example, for a straight BNC Plug on flexible
cable, the VSWR limit is 1.3:1 maximum (usually just written
as 1.3 max) to
4 Ghz. The second method is to allow for the fact that VSWR
is typically a direct function of frequency. A straight SMA
plug on RG-142 B/U cable can have a maximum VSWR of 1.15 +
.01 * F (Ghz) to 12.4 Ghz. For example, at 2 Ghz, the maximum
allowable VSWR would be 1.15+2*.01 or 1.17 max. At 12.4 Ghz
it would be 1.15+12.4*.01 or 1.274 max. Naturally, these values
can be converted to Return Loss or Reflection Coefficient.
Click here for
a Reflection Converter.
What
is Insertion Loss and how is it specified?
Insertion
Loss, expressed in dB is defined as 10*log (Po/Pi) where Po=
Power Out and Pi=Power In. There are 3 main causes of Insertion
Loss: Reflected losses, Dielectric losses and Copper losses.
Reflected losses are those losses caused by the VSWR of the
connector. Dielectric losses are those losses caused by the
power dissipated in the dielectric materials (Teflon, rexolite,
delrin, etc.). Copper losses are those losses caused by the
power dissipated due to the conducting surfaces of the connector.
It is a function of the material and plating used.
In
general, the insertion loss of a connector is on the order
of a few hundredths to a few tenths of a dB. As with VSWR,
it can be specified as a “flat line limit” or as
a function of frequency. Using the same examples as the VSWR,
a BNC is specified at .2 dB maximum when tested at 3 Ghz.
For the SMA, the requirement is .06*SQRRT Frequency in GHz
when tested at 6 Ghz. For example, at 4 Ghz, the requirement
would be .06*2 or .12 dB max. Although the connectors are
specified to operate over a wide frequency range, they are
only specified for testing at particular frequency because
the test procedure required to obtain accurate measurements
of such small losses is a very precise, and time consuming
process. The procedure is defined in MIL-PRF-39012 and can
be found at:
http://www.dscc.dla.mil/Programs/MilSpec/ListDocs.asp?BasicDoc=MIL-PRF-39012
How do I specify the performance characteristics of a cable
assembly?
There
are two performance characteristics of cable assemblies that
are of interest: VSWR (or Return Loss) and Insertion Loss.
For
all but the shortest cable assemblies (less than 6 inches)
using extremely low loss cable, the Insertion Loss will be
overwhelmingly due to the attenuation of the cable itself
and can generally be determined from the manufacturers data
sheets.
On
the other hand, the VSWR will normally be overwhelmingly due
to the connectors. Remembering that VSWR is a vector quantity,
as the frequency is swept, the VSWR of each connector will
add and then subtract in and out of phase giving a swept VSWR
plot that goes up and down. Where these maximums and minimums
occur will depend on the length of the cable and its dielectric
constant. As a general rule, the maximum can be calculated
by determining the Reflection Coefficient of the connectors
on each end. Worst case will be the addition of the 2 reflection
coefficients. Although small, some amount should be added
for the cable. In addition, the attenuation of the cable,
if significant, will reduce the VSWR. For this example, we
will ignore the attenuation of the cable as a factor in the
calculation. As an example, let’s say we have one connector
with a VSWR of 1.2 at the frequency of interest and the other
connector is a 1.25. The cable VSWR is 1.05. Click
here for a Reflection Converter. Converting the VSWR to
reflection coefficient gives .091, .111 and .024 respectively.
The maximum Reflection Coefficient=. 226. This converts back
to a VSWR of 1.584 maximum. A quick way of getting the result
is to multiply the 3 VSWR values. In this case it would be1.2*1.25*1.05=1.575.
This is very close to the previously calculated result. For
Return Loss, VSWR can be converted to dB. For Return Loss,
if the connectors have the same Return Loss value and the
cable Return Loss is at least 6 dB better, then the cable
assembly Return Loss can be determined by subtracting 6 dB
from the connector Return Loss. For example, if the Return
Loss of each connector is 23 dB, then the cable assembly can
be expected to have a maximum Return Loss of 17 dB. If the
Return Loss of each connector is different or if the cable
Return Loss is not insignificant, then each of the Return
Losses would have to be converted to Reflection Coefficient,
added and then converted back to Return Loss. It is very important
to realize that the VSWR of the connectors and cable add vectorially
and the resultant VSWR of the cable assembly will be considerably
higher than each individual component’s VSWR.
How
does Amphenol RF make certain that the VSWR of a connector
will meet the specification?
Two
steps are during the design and development phase of connectors
at Amphenol RF. Once the 3-D mechanical drawings are completed,
the first step in the RF design process is to model the connectors
using High Frequency Structure Simulator (HFSS) software.
This is a state of the art computer program that allows us
to model the 3-D structure of the connector and simulate its
RF performance. Click
here for a HFSS Analysis This software has no frequency
limit and will allow us to view the VSWR (Return Loss) and
Insertion Loss of the connector, or any microwave device.
In addition, it has the capability if performing Time Domain
Reflectometry (TDR). TDR is a technique that allows us to
see reflections as a function of distance. This enables us
to see “inside” the connector and determine exactly
where the discontinuities are located. Click
here for TDR Analysis. Corrections can be made, and a
new analysis can be performed. In addition, the HFSS software
has an “OPTIMETRICS” module that allows us to automatically
vary dimensions within the connector and solve for the best
VSWR. This process greatly reduces the engineering design
cycle time. Click
here for Optimetrics Analysis.
After
the modeling is complete and the connector has been prototyped,
it will be tested on a Network Analyzer. This is a piece of
test equipment that measures the S parameters (VSWR and Insertion
Loss) of the connector or cable assembly. Differences between
the simulation and actual test data can be evaluated. In general,
the simulated data results in an optimized VSWR with enough
margin to allow for manufacturing and assembly variations
and still meet the customer requirements.
What
causes differences between simulated and actual test data?
| 1. |
Actual
interfaces during testing will be different from the model. |
| 2. |
Dielectric
constants and dimensions of cables are variable and cable
is modeled as "perfect". |
| 3. |
Dielectric
constants of different materials used as insulators (i.e.:
delrin, lcp) are not precisely known. |
| 4. |
Calibration
kits for all types of connectors are not available and
therefore gating is often used in taking data. |
| 5. |
Small
air spaces are usually not modeled as they greatly increase
the complexity of the model.. They are typically filled
with dielectric. |
| 6. |
Actual
assembly of the device can result in components compressing
(ie: press fits) which may cause the longitudinal and
relational position of the components to be different
from the model. |
| 7. |
Generally,
small radii are not modeled. They are replaced with chamfers
or even eliminated completely. |
| 8. |
Connectors
are modeled using nominal dimensions. Actual parts have
tolerances. |
| 9. |
The
adaptive frequency is a single frequency usually about
80% of the upper frequency limit, but the analysis is
usually swept over a broad frequency range. |
| 10. |
Crimping
can deform cables in an extremely variable manner and
this is usually not modeled. |
Click
here to see examples of some of varied devices modeled
and analyzed in HFSS.
What
is Passive Intermodulation Distortion?
Passive
Intermodulation Distortion (PIM) is a phenomenon that occurs
when 2 signals present on a transmission line mix in a non-linear
manner. This mixing creates additional frequency components
(where the fim = ± mf1 + ±nf2 and m and n are
integers) that may fall within a cellular uplink band, causing
interference. Poorly designed or assembled connectors and
cable assemblies can generate PIM. Amphenol RF has a complete
line of low PIM connectors and cable assemblies. Click
here for PIM. This paper discusses the basics of PIM and
describes how to design low PIM connectors as well as detailing
the latest test procedures used in the industry.
What Is the Difference Between Average and
Peak Power?
The amount of power a connector can handle can determine the
long term (or even short term) reliability of your system.
Using a connector that cannot adequately dissipate the power
applied can cause serious problems and failure of your system.
There are 2 types of power handling (expressed in watts) that
must be considered: Average Power and Peak Power.
Average Power is the input power to a cable/connector which
will produce a maximum safe center conductor temperature under
steady state conditions when terminated with a matched load.
A safe center conductor temperature is one that will not melt
the dielectric. When considering Average Power the following
points should be noted:
- Average
Power is inversely proportional to frequency and must be
de-rated accordingly
- Average
Power is equal to a Power Rating @ 1 Mhz/√ (Frequency in
Mhz)
- Connectors
generally have higher power ratings than the cable to which
they are attached
- Connectors
have metal shells whereas cables have braids covered by
plastic jackets
- Connectors
can be attached to bulkheads which help dissipate heat
- Connectors
usually have lower attenuation per unit length due to air
sections within the connector
Peak
Power is limited by the voltage rating of the connector, and
is determined by the equation V2/Z where V=the peak
voltage rating and Z is the characteristic impedance. When considering
Peak Power the following points should be noted:
- Peak
power generally has a very short duty cycle, but you should
calculate the average power of a peak pulse to be certain
it does not exceed specifications
- Peak
Power is not a function of frequency
- Peak
Power is an inverse function of VSWR and modulation schemes
and must be derated
- Peak
and Average Power are functions of altitude and must be
derated
- Maximum
power ratings will always be the lesser of the cable/connector
combination
What is the Difference Between Return Loss and VSWR?
This question addresses VSWR (Voltage Standing Wave Ratio)
and Return Loss, which are both a measure of the same parameter.
That is, the amount of signal reflected by a connector. It
is the major factor contributing to the total signal efficiency
of the connector.
Return
Loss is the portion of a signal that is lost due to a reflection
of power at a line discontinuity. Return Loss is similar to
VSWR and is generally preferred in the cable industry to a
VSWR specification. Since it is a logarithmic measurement,
it is very useful when displaying very small reflections.
VSWR
is an acronym for Voltage Standing Wave Ratio. VSWR is the
ratio of voltage applied to voltage reflected. VSWR is similar
to Return Loss and is generally preferred in the connector
industry to a Return Loss specification. Since it is a linear
measurement, it can be useful when displaying larger reflections
due to the fact that small differences are not compressed
as they are in a logarithmic measurement.
Q: How can I be sure a PCB connector will perform when mounted on my design?
A: When designing surface mount or PC style connectors, the connector manufacturer generally does not
know on what type of environment the connector will be launched. The connector is designed as a coaxial device
and its performance is optimized using High Frequency Structure Simulator (HFSS) software. However, depending on
the type of PC board (microstrip, stripline, coplanar, etc.), and board parameters (dielectric constant, thickness,
trace width, etc.), different compensation schemes will be necessary at the launch because a discontinuity
capacitance is introduced at the point of launch, and it can severely degrade system performance.
At Amphenol RF, we use ANSOFT HFSS to model a connector on a customer's PCB. After simulating the performance and
examining the frequency and time domain results, we make changes to the pads and/or trace to achieve a better
broadband impedance match. We will then advise the customer as to what changes will result in optimum performance.
The connector above is shown mounted on a microstrip board. There is no compensation, and the Return Loss for the uncompensated
launch is -20 dB and the Insertion Loss is -.04 dB. The Time Domain shows a capacitance at the launch, so additional inductance
must be added to compensate.
Several different configurations can be simulated in a matter of minutes, and the changes recommended result in a Return loss of
-35 dB (a 15 dB improvement) and a decrease in Insertion Loss from -.04 dB to less than -.01 dB.
This is an example of the types of services Amphenol offers to our customers that help design better products.
Q: What causes PIM in connectors and how does Amphenol design low PIM connectors?
A: In our May 2002 issue of The Amphenol RF Connection Newsletter, I introduced you to Passive Intermodulation Distortion (PIM). In this issue I will discuss its causes in connectors and cable assemblies and how to prevent it from becoming a problem in your system. As discussed earlier, PIM is caused by non-linearities in a passive circuit. The main cause of non-linear action in a connector or cable assembly is poor contact junctions, which can be caused by any one of the following:
- Insufficient contact pressure
- Irregular contact surfaces
- Oxidation-causes metal/oxide junction
- Contact impurities
- Corrosion
In addition, ferromagnetic materials such as nickel and steel, as well as contamination all contribute to the problem.
At Amphenol RF, our best performing connectors are designed with soldered center and outer contact connections using one-piece construction wherever possible. In addition, we have a line of N and 7-16 connectors that are over-molded to eliminate the effects of wind, vibration and thermal stresses. Contact forces and other aspects of the designs are such that sufficient pressure is assured in order to prevent PIM. Center contacts are plated silver or gold and connector bodies can be plated silver or white bronze. Connectors and cable assemblies are manufactured in clean environments and assembled by highly trained personnel who are aware of the causes and effects of PIM.
All assemblies are then tested to ensure that they meet all published specifications.
Q: How to ensure high performance with minimal
reflections and losses?
A: It is important to understand that a connector is
not just a mechanical connection between two transmission lines.
It is also an RF connection with the express purpose of transferring
as much of the RF energy as possible through the transmission
line from Point A to Point B. All of the mechanical dimensions
and materials used in the design of the connector will have
an impact on its RF performance, particularly internal diameters.
These diameters set up the transmission line impedances which
must be close to the characteristic impedance (Zo) of the system
(i.e. 50 ohms) or else the reflections in the connector will
be excessive. Where significant deviations from Zo are required,
perhaps for a barb or knurl on a contact, proper compensation
steps must be added to minimize reflections.
When we design a connector, how can we be certain that it will
perform well and with minimal reflections and losses? In order
to ensure that we meet all RF performance requirements, Amphenol
RF has been using a very powerful simulation tool called HFSS.
This software package is manufactured by ANSOFT and we have
been using it to design connectors since 1997. The tremendous
advantage of the tool is that we can design the connector and
simulate its performance without having to make actual samples
and test them. We can try various modifications and have results
within minutes or perhaps hours at the most for more complex
designs. Only after we are satisfied with the results will we
submit the design for samples to be made and then tested.
|
In this example, a BNC right angle plug for RG-58/U cable had
a VSWR requirement of 1.25 maximum from DC- 4 GHz.
In this example, a BNC right angle plug for RG-58/U cable had
a VSWR requirement of 1.25 maximum from DC- 4 GHz.
Mechanical
Connector Drawing
|
The connector above was modeled in HFSS and is shown below.
In HFSS, we need to make certain changes to the mechanical connector
drawing. For this particular connector, these involve removing
overlapping materials such as press fit insulators, adding the
RG-58/U cable and setting up a proper BNC interface. Each component
within the connector must be a separate identifiable entity
and assigned a material property. This includes any air spaces.
All external metal parts are unnecessary as HFSS defaults to
a perfect conducting background. Since electric fields cannot
penetrate metal surfaces, the metal body of the connector and
external hardware are not needed for the simulation.
|
| HFSS Model |
After setting up the simulation in HFSS to run from DC-4 GHz, the initial VSWR results
are shown below.
|
|
Initial Design VSWR Results
It can be seen that the VSWR meets the customer requirement up to 3 GHz, but from 3-4 GHz, it is above the spec and changes must be made to the design to improve the VSWR. In the next issue of Amphenol RF Connections, I'll describe how we use ANSOFT HFSS to help us modify the design and improve the VSWR performance.
DW 9-12-03
|
|
 |
