October 1, 2005
Arbitrary waveform generators (AWG) are signal sources that
derive their analog outputs directly from digital information.
For a general description of AWGs please refer to the backgrounder:
"The Importance of Arbitrary Waveform Generators",
http://www.agilent.com/about/newsroom/tmnews/background/2004/20sep2004_bg.html
Wideband AWGs are commonly used for simulating signal
scenarios for receiver testing and transmitter up-conversion
blocks. The wideband AWG's ability to generate arbitrary signals
is important in aerospace and defense and emerging communications
where a wide range of signal formats are used. Many of these
are either proprietary or do not comply with any accepted
commercial standard. In addition, the high dynamic range of
the wideband AWG makes it possible to simulate the effects
of system impairments such as distortion and noise, and to
verify overall system integrity under these stresses.
A Wideband Challenge: Wide Bandwidth and Long Play Time
A typical communications channel may have information bandwidth
as narrow as tens of kilohertz, and typically less than a
few megahertz. This relatively narrowband signal is created
using a variety of modulation schemes, such as QPSK or OFDM.
For wireless transmission, this narrowband signal is up-converted
on a carrier with a frequency of hundreds of megahertz or
even several gigahertz. A wideband AWG makes it possible to
perform the first stage of up-conversion digitally. Digital
up-conversion has a number of advantages: the results are
predictable, known channel impairments can be easily added,
and it is simple to vary the carrier frequency to test all
the available channels in a multi-channel communications system.
In a traditional AWG, producing the carrier digitally comes
at a price, however. Creating a carrier of, for example, 500
megahertz requires a sample rate of 1.25 gigasamples per second
to avoid aliasing the signal. As an example, to store the
waveform data for one second of unique play time of a compressed
voice channel (10 kilobits / second) on a 500 megahertz carrier
requires approximately 1.25 billion samples. Most of these
samples are used to capture the repetitive carrier. The number
of samples required to produce the voice information alone
is only approximately 20 thousand samples.
The N6030A is a high-performance AWG that provides engineers
with both wide bandwidth and wide dynamic range, simultaneously.
Its dual, differential output channels operate at 1.25 gigasamples
per second and offer 15 bits of resolution giving 500 MHz
of instantaneous analog bandwidth and more than 65 dB of spurious-free
dynamic range in each channel. The unit has an option of up
to 16 million samples of waveform storage for each channel.
This memory depth is one of the largest in the industry at
this sample rate. Still, at the full sample rate, this corresponds
to only 12.8 milliseconds of unique play time. Download times
for the N6030A are also the fastest in the industry, loading
the full sample memory in approximately 1 second over the
internal PXI backplane. Scaling these results show that it
would take 5 gigabytes of high speed memory to directly store
one second of unique play time. Aside from the cost, it would
take approximately 80 seconds to load the waveform (as well
as additional processing time to compute it initially). Many
common applications, such as bit error rate (BER) testing
in the presence of Gaussian noise or modeling Doppler shifts
due to wobble in satellite systems, require play times of
ten or more seconds. For the reasons mentioned above, it is
not practical to directly store the carrier and modulation
data in memory, in these cases.
The Sequencer: Memory Compression in a traditional AWG
Many waveforms produced by an AWG are repetitive. The sequencer
in most commercial AWGs makes use of this characteristic to
reduce the amount of unique memory required to store the waveform.
For example, a radar pulse consists of a pulse with some defined
pulse shape followed by some off time. This pulse is repeated
a number of times with a certain duty cycle. The traditional
AWG sequencer makes it possible to take sections of the waveform
memory and play them with a defined number of repetitions
in a predefined order. Thus a short section of memory with
the pulse off can be repeated, this is followed by the pulse,
and finally this complete set is repeated some number of times.
Traditional Sequencer
The N6030A has the industry's most powerful traditional sequencer,
enabling two levels of nesting and a wide range of hardware
triggers. It is still of limited use in producing signals
having narrowband modulation on a high frequency carrier.
In this case the carrier is repetitive -- but due to for example
inter-symbol interference, Doppler frequency shifts, or added
noise -- the modulated carrier is different each time a particular
narrowband modulation symbol is played. The waveform data
(and carrier) must all be written to unique waveform memory,
and the available playtime is very short.
Direct Digital Synthesis: Efficient compression of Narrowband
Modulation on a Carrier
Direct Digital Synthesis (DDS) makes it possible to define
a carrier in the Sequence memory as a single instruction.
The start and stop carrier frequencies as well as a phase
offset are specified for each sequence stored in the waveform
memory. Only the narrowband baseband I Q modulation data is
stored in the waveform memory. The narrowband waveform memory
is clocked at a much reduced rate and interpolated before
placing it on the carrier in real-time. This greatly extends
the play time. Also stored in the sequence memory is a gain
factor that can be used for fading applications. These DDS
enhancements make it possible to store much longer unique
signal scenarios. In the example described above, more than
20 minutes of unique 10 kHz compressed voice data could be
stored in the dual 16 megasample modulation waveform memory,
and applied in real-time to a 500 MHz carrier. Moreover this
same waveform data could be used to modulate multiple differing
carrier frequencies. This enables the testing of all the frequency
channels using separate instructions in the sequencer memory,
without reloading the waveform memory. Finally, gain can be
varied in the sequencer and true Gaussian random noise added
to enable BER testing down to arbitrarily low error rate values
(10^ -12 or lower BER). Without DDS, the minimum BER that
can be achieved is set by the maximum number of unique modulation
symbol bits that can be stored in the waveform memory (~10^
-4 BER), and is not adequate for rigorous system tests.
The DDS algorithm and sequencer enhancements are implemented
in a Field Programmable Gate Array (FPGA) in the N6030A AWG.
Unlike the static, non-reconfigurable logic (custom ASICs
and board-level ICs) used in many commercial AWGs, the FPGA
has the advantage that its function can be redefined as a
part of the instrument software. When there is a need to efficiently
test narrowband signals on a carrier, the DDS personality
can be purchased (Option 330) and upgraded in the field.
Each N6030A module has two analog output channels. Both can
be used to drive an I Q modulator to give 1 GHz of modulation
bandwidth. Alternatively only a single channel can be used
for IF up-conversion on a digital carrier, in this case 400
to 500 MHz of modulation bandwidth is possible. In either
case, the DDS personality gives a powerful method of putting
narrowband modulation on a high frequency carrier.
Applications of Direct Digital Synthesis
The DDS personality as well as the sequencer in the N6030A
are designed to be generic and versatile, addressing a wide
range of possible applications. Some examples are described
below. As standards and modulation formats evolve in the future,
many can be simply captured by modifying the data in the waveform
and sequencer memories of the AWG.
- Narrowband modulation can be created at baseband at a
reduced sampling rate. This is interpolated and up-sampled
to the full DAC clock rate in the FPGA. This results in
a large increase in the unique play time that can be stored
in the waveform memory. For example, baseband QPSK data
consists of one of four fixed positions in the I Q plane
representing the different QPSK symbols. In the case of
OFDM, multiple closely placed subcarriers are created near
DC. These are modulated with the symbol data using phase
shift keying or quadrature amplitude modulation. In all
these cases, the baseband modulation is then digitally up-converted
to the desired carrier frequency programmed into the DDS.
DDS Sequencer with added Noise
- Additive Gaussian noise (AWGN) can be combined with the
output of the DDS multiplier. This is invaluable in testing
the reliability of a communications channel in the presence
of noise. Traditional methods of adding AWGN to reference
waveforms can create problems in setting the desired energy-per-bit
to noise power spectral density ratio (Eb/No). Traditional
solutions typically require the reference waveform to be
manually summed with an external AWGN generator, resulting
in the need to combine two different RF signals to achieve
the overall ratio. Besides the added cabling complexity,
the user must also account for cable loss through the combiner
to achieve the desired Eb/No (or carrier to noise). The
N6030A series' Option 250 gives users the ability to seamlessly
add controlled amounts of AWGN noise to their test waveforms
eliminating the need to use a dedicated external noise source.
Besides saving cost and rack space this also gives users
a more accurate approach to receiver testing because cable
drift and noise power calibration have largely been eliminated.
The noise has a bandwidth of 500 MHz (15 dB crest factor)
that is uncorrelated with the AWG sample clock. When using
both DAC outputs driving an external I Q modulator, 1 GHz
of noise bandwidth is obtained. This gives users the ability
to perform true BER measurements as a function of the input
signal in the presence of very broadband Gaussian noise.
- Complex radar simulation scenarios can be efficiently
created using the DDS engine. The DDS allows independent
control of waveform parameters. Instead of creating waveforms
using AWG memory, users can define waveforms by creating
independent profiles for the signal's carrier frequency,
frequency modulation (FM), phase modulation (PM) and amplitude
modulation (AM) characteristics which are stored in the
sequencer memory. The value of realizing waveforms using
the N6030A DDS engine is that realistic waveforms can be
realized for receiver testing in the lab instead of actual
in-flight testing; an expensive and time-consuming process.
Another key benefit is that these realistic waveforms can
be created without utilizing precious waveform memory. Some
simulations require several seconds of unique playtime to
evaluate a radar receiver's detection algorithms. For the
reasons mentioned previously, such long playtimes are possible
only with DDS.
- A good example of where DDS is useful might be to simulate
an aircraft (radar) flying past airport surveillance radar.
As the aircraft approaches the rotating surveillance antenna,
the amplitude of the radar return may vary from pulse to
pulse. The radar's transmit (and receive) frequency may
hop from pulse to pulse. In addition, the aircraft's speed
may be changing (Doppler shift) as it closes in on the antenna.
Such variations in the radar signature from instant to instant
are one reason why in-flight testing has been so heavily
utilized throughout the design process. However, flight
range operations are very expensive and time consuming and
can cost hundreds of thousands of dollars per operating
hour. The N6030A series' DDS option provides design engineers
a way to simulate these complex and relatively long simulations
in the R&D lab by allowing the engineer to program all
parameters of a waveform profile while conserving the AWG's
on-board memory.
- Satellite communications signal simulations can be performed
taking into account variations due to orbital anomalies.
Users can generate a reference QPSK waveform and modify
its characteristics with a slowly varying amplitude modulation
profile to simulate a satellite wobbling or tumbling in
orbit -- which needs to be stabilized. Doppler shifts can
also be added. If the amplitude or frequency variations
are not linear functions of time, they can be closely approximated
by a piece-wise linear approximation. With 500,000 entries
in the sequencer memory, the N6030A DDS can approximate
complex amplitude and frequency profiles with an almost
arbitrary degree of accuracy.
- The DDS capability can be used to simulate frequency hopping
radio systems. Many military and commercial radios employ
complex modulation schemes that hop their carrier frequencies.
Where commercial radios often hop to improve spectral efficiency,
military radios utilize frequency agility to improve communication
security (low probability of intercept). In both cases,
test waveforms are required to modulate and/or hop the modulated
signal across a wide bandwidth. In commercial CDMA systems
the hopping carrier information can be included in the sequencer
data. The waveform data consists of the lower rate modulation
information. By sampling the waveform data at a lower rate,
longer unique play times are possible.
Another example is the Link-16 tactical radios used by the
United States Navy, the Joint Services, and forces of the
North Atlantic Treaty Organization (NATO). Though the actual
modulated carrier is only a few MHz wide, it is dynamically
hopped across 255 MHz of frequency spectrum. New radio designs
based on the Wideband Networking Waveform (WNW) format have
even wider hop bandwidths. In this case the hopping sequence
is classified. In addition to waveform encryption, many
secure communication systems also employ frequency hopping
to provide an additional layer of security, as well as,
improved spectral efficiency. The transmitter and receiver
work together to coordinate the frequency use plan. Though
the actual hop frequencies may be known in advance, the
order of use for each frequency is determined dynamically.
For example, the US Navy's Link-16 communication system
utilizes 51 distinct (unclassified) frequencies during operations.
However, the use of this spectrum is determined by one of
the transmitters (classified). The transmitter periodically
tells the intended receiver where to hop to the next frequency
through an encoded frequency word. N6030A Option 300 is
a hardware interface which allows the user to dynamically
point to pre-defined carrier signals created with the DDS
engine. In this way, the N6030A series can simulate the
transmitter portion of the radio which performs the frequency
hopping so the designer can evaluate other sections of the
radio or performance of the intended receiver.
###
Contacts:
Janet Smith, Agilent
+1 970 679 5397
janet_smith@agilent.com
|
Contact Us