5305 Spectrum Drive, Frederick, MD 21703-7362 TEL: 301-846-9222, 800-638-2048 Fax: 301-846-9116
web: www.aeroflex-weinschel.com email: sales@aeroflex-weinschel.com
84
Revision Date: 12/15/04
Programmable Attenuators
Weinschel has been a major supplier of
programmable attenuators to the RF industry for over
25 years. Historically the most demanding specifica-
tions for these components have been low insertion loss and SWR,
combined with a reasonable life expectancy of several million
switching cycles. This was usually adequate for RF instruments
like spectrum analyzers and signal generators, wherein the
attenuator bandwidth rather than the switching speed was of prime
concern. To achieve wide bandwidths the programmable attenua-
tors were mostly of electromechanical design and the linearity of
these passive components was not only assumed but never
questioned by any customer. Intermodulation distortion discus-
sions and problems were usually limited to components such as
amplifiers, mixers and filters.
In recent years, however, wireless communication systems
employing complex digital modulation schemes, increased chan-
nel capacity, high transmit power and extremely low receiver
sensitivity have put into question the linearity of passive compo-
nents. Even very low level multi-tone intermodulation products
generated by attenuators can seriously degrade the efficiency of
a system/ instrument if these products fall within the user pass-
band. For two closely spaced tones at frequencies f1 and f2, the
third order IM products at 2f1 - f2 and 2f2 - f1 , are the most
harmful distortion products. They are harmful because they are
located close to f1 and f2 and virtually impossible to filter out. In
today's base stations the multicarrier power amplifier (MCPA) is
replacing banks of single-channel amplifiers and their corre-
sponding power combining network. MCPAs have the capability
of carrying a number of modulation schemes simultaneously and
can also employ schemes such as dynamic-channel-allocation
(DCA) to use the allocated frequency spectrum more efficiently.
The in-band intermodulation distortion (IMD) performance of
these amplifiers is extremely critical and needs to be measured
using low distortion programmable multi-tone generators whose
IMD performance must be quite superior. This is discussed in the
two case studies cited here.
Electromechanical programmable attenuators obviously
provide a far superior IMD performance than
their corresponding solid state
counterparts employing semi-
conductor switching elements.
However, their slow switch speed, in the order of millisec-
onds, and short switch life in the order of 5-10 million cycles
make them unattractive in some applications like cell phone test-
ing and other ATE systems. Solid State programmable attenuators
do overcome these two problems and are therefore included here
for IMD performance comparison. It is not the intent of this brief
article to go into the theory of intermodulation distortion. The goal
here is to provide some good basic IMD test data for a
variety of commercial programmable attenuators and let the end
user select the most appropriate type for his application.
Measurement System and Parameters...
All test data presented here was generated using a
commercially available Passive IM Analyzer, Summitek Model
SI-800A which provides a fully integrated system for characteriz-
ing distortion produced by cables, attenuators and other passive
devices. Although the system is capable of measuring both,
through and reflected IM3, IM5, IM7 and IM9, the focus here is
only on through IM for the most troublesome third order product,
IM3. To carry out a meaningful comparison between different
attenuators all measurements were carried out using two equal
amplitude input tones at 869 MHz (f1) and 891 MHz(f2) , the IM3
frequency being 847 MHz (2f1-f2). Input carrier power was
stepped in increments of 1 dB from -7dBm to +27dBm. All exter-
nal adapters and cables were carefully selected to maintain the
system's residual IM level of around -120 dBm. Although the sys-
tem permitted receiver measurements between -70 to -120 dBm
we restricted all measurements between -85 to -110 dBm by
using a calibrated low IM coupler and attenuators at the output
port of the DUT. One must be aware that the accuracy of such
small signal measurements can easily be off by 2 to 3 dB so
restricting the measurement dynamic range helps reduce the
receiver non-linearity error. Measurements were done over sev-
eral days to ensure stability and repeatability.
Distortion Comparison for Basic Types of
Programmable Attenuators...
The programmable attenuators discussed here are the
switched type with a discrete number of `cells'. Switching
between the zero and attenuate state on each cell is achieved by
a DPDT switch configuration. The cell values are usually in a
binary sequence. For example a 6 cell/6 bit unit could have 1, 2,
4, 8, 16 and 32 dB sections providing a 63 dB dynamic range in
1dB increments. Four basic families of programmable attenua-
tors are compared, each family being identified by the switch
element used to achieve the transfer from zero to attenuate state.
For the purposes of distortion comparison it was deemed
necessary to select units with similar electrical length and/or
programmability. Both the electromechanical units, TO5 relay
and edge-line type, had an electrical length of about 20 cms. The
two solid state units had 6 cell programmability yielding 63 dB in
1 dB step size. All IM3 vs Pin measurements were done with the
attenuators programmed to be in their characteristic zero insertion
loss state. The zero state was selected because it
generated the highest IM3 levels. The graph below shows the
Intermodulation Distortion in Programmable Attenuators....