IEEE Geoscience and Remote Sensing Magazine - June 2017 - 65

june 2017

ieee Geoscience and remote sensing magazine

To
Transmitter
I
L
×N

Phased-Locked
Oscillator
10 GHz

10-GHz
BPF

To ADC

fIgURE 10. The system block diagram of the 2-8-GHz DDS-based chirp generator.

BPF
Amplifier

Frequency
Multiplier (× 2)
12-18-GHz Output

×N

BPF

Mixer

2-8-GHz
BPF

DDS-Based Chirp Generator
BPF
BPF
BPF

Amplifier

Frequency
Multiplier (× 4)
6-9 GHz Output
DDS
1.5-2.25 GHz;
240 µs
1-GHz
Reference
Clock

DDS-BASED SYSTEM
As discussed in the previous section, VCO-based chirp
generator design is sensitive to temperature change, and
the inherent tracking speed of the PLL limits further reduction in the chirp's sweep rate to improve the coherent gain.
Thus, we developed a UWB digital-based chirp generator
for use in the Snow Radar in 2013 [49]. This DDS-based
Snow Radar was first flown on the NASA P-3B aircraft as
part of the 2013 OIB Antarctica Mission. Figure 10 shows
the block diagram of the chirp generator portion of the
radar. The chirp generator consists of three major subsections: a DDS, a frequency multiplier, and a frequency
down-converter. The DDS is implemented using a field
programable gate array combined with a 2.5-GS/s, 14-b
digital-to-analog converter. The DDS generates a 240-μs,
1.5-2.25-GHz baseband chirp using a phase accumulator that is fed into a frequency-multiplier subsection. The
frequency-multiplier subsection is constructed by cascading two stages of amplifier-filter-frequency multiplier-filter modules to multiply the bandwidth of the baseband
chirp by a factor of 8. The resultant 12-18-GHz chirp is
then mixed with a phase-locked 10-GHz clock to generate
the required 2-8-GHz chirp.
One major issue with this design is that the frequencymultiplier chain introduces frequency nonlinearity to the
chirp primarily because of the nonlinear group delay of the
filters and the signal reflections between nonideal components. The frequency nonlinearity of the chirp is also amplified at each stage of the frequency multiplier. A frequency linearization method is thus developed to minimize the
frequency nonlinearity [49]. This is done by first measuring the phase offset of the frequency-divided version of the
12-18-GHz chirp and then predistorting the corresponding
phase of the baseband chirp in the DDS. The purpose of the
frequency division is to reduce the sampling requirement of
the chirp signal while maintaining the phase information
of the chirp. Figure 11 shows the system response of the
DDS-based radar measured using a 3-μs (~450-m) optical
delay line before and after predistortion. The sidelobes
and sidebands are clearly reduced significantly after the
phase predistortion.
In 2015, the bandwidth of the Snow Radar was further
extended to cover 2-18 GHz to expand its ability to measure thin snow (<10 cm) on sea ice and to collect off-nadir
snow backscatter data at X- and Ku-bands for snow-density
estimation [33], [50]. The new radar has a range resolution
of about 1.13 cm in snow (with windowing). This radar
was test-flown on a Twin Otter aircraft in Barrow, Alaska,
as part of the 2015 NRL Determining the Impact of Sea
Ice Thickness on the Arctic's Naturally Changing Environment (DISTANCE) sea ice campaign. The radar employs
a similar design architecture as the previous 2-8-GHz
Snow Radar, except that the baseband chirp starts from
1.375 to 2.375 GHz. This baseband chirp then undergoes
a frequency multiplication of factor 16, which results in a
millimeter-wave chirp that spans over 22-38 GHz. Similar

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2017