Development of a Reduced-Cost CP-FTMW Spectrometer Using

Direct Digital SynthesisIan Finneran

Daniel HollandBrandon CarrollGeoffrey Blake

California Institute of TechnologyDivisions of Geological & Planetary Sciences and Chemistry & Chemical

Engineering, California Institute of Technology, Pasadena, California 91125,USA

Motivation:Maximize Bandwidth/Sensitivity for Given Budget

B. C. Dian et al., Science 320, 924 (2008).G. G. Brown et al., Review of Scientific Instruments 79, 053103 (2008).

G. G. Brown et al., Review of Scientific Instruments 79, 053103 (2008).

D. A. Obenchain et al., Journal of Molecular Spectroscopy 261, 35 (2010).

B. Reinhold et al., Journal of Molecular Spectroscopy 270, 89 (2011).

Main Costs forCP-FTMW Spectroscopy at 8-18 GHz

AWG/AFG Amplifier Digital Sampling

~$10k-$40k per GHz

bandwidth

~$1k/W – low end$100-$500/W – high end ~$10k-$20k

per GHz bandwidth

Total Cost: ~$120k 2 GHz BW

1 W – 200 W per GHz

www.tek.com

www.arworld.us/

www.tek.com

Direct Digital Synthesis (DDS)Fr

eque

ncy

Time

Chirped pulse specs needed:• ~1 μs pulse length• High bandwidth• Phase stable• Low trigger latency

Dan Holland: Asynchronous Optical Sampling (ASOPS)

Time Domain THz Spectrometer

http://www.analog.com

A Small, Lightweight, Low-Power 2 GHz BW DDS Solution for <$1000

4 GS/s AWG DDS/PLL Solution

DDS Percent of AWG

Bandwidth 2 GHz 2 GHz 100%

Price ~$50,000(as low as $10,000 used)

$775 ~1.5%(new vs. new)

Power ~200 W 3 W ~1.5%

Size 1,400 cm3 42,000 cm3 ~3.3%

Weight 14.1 kg 0.24 kg ~1.7%

www.tek.com

http://www.analog.com

PLL board DDS board

DDS Drawbacks• Not arbitrary • 1 chirped pulse/trigger• Requires low end

computer/arduino to load control word

• Less bandwidth (for now)

4 GHz PLL Sample Clock Source

6 GHz PLL LO Source

DDS Chirp Source

Super-Nyquist content with 1 GHz chirped pulse 2 GHz sample clock

~$70,000, 40 W, 3.8 GHz (DSB) solution2.1 kHz repetition rate (70 FIDs per gas pulse)

I. Peña et al., The Journal of Physical Chemistry Letters 4.1 (2012): 65-69.

LO = 11.9 GHz 31 seconds of data collection or 65,000 averages

PLL is internally clocked (for now)

S:N = ~350:1(CH3)2CO

211- 202 EE

(CH3)213CO

170 million Averages (22 hours), LO = 11.9 GHz(CH3)2

13CO

(13CH3)2CO

(CH3)2CO

S:N = ~18,000:1 211- 202 EE

321-312

211- 202

541- 532

321-312

P. Groner et al., The Astrophysical Journal Supplement Series 142, 145 (2002).J. Swalen and C. Costain, The Journal of Chemical Physics 31, 1562 (1959).F. Lovas and P. Groner, Journal of Molecular Spectroscopy 236, 173 (2006).

Very Tentative Super-Nyquist (CH3)2C18O Detection

11900+2000+(2000-1617.73) = 14282.27 MHz

14282.26 MHz 111 - 000 EE

S:N = ~0.2% of super-Nyquist normal species

R. Nelson and L. Pierce, Journal of Molecular Spectroscopy 18, 344 (1965).

Chirp Source: DDS/PLL - $775LO source*: PLL - $150

Chirp Source: AWG ($50k-$100k)LO source: MW Synthesizer (~$7k)

*Doubler = $360, not needed for newest PLL

Chirp Source: AWG ($50k-$100k) LO source: MW Synthesizer (~$7k)

Chirp Source: DDS/PLL - $775 LO source*: PLL - $150

*Doubler = $360, not needed for newest PLL

(CH3)2CO LO=11.9 GHz 10 million averages (1.3 hours)

Phase Stability: Averaging on 321-312 AA Acetone

(10762.6 MHz rest frequency, LO = 11900 MHz)

Conclusion• DDS+PLL combination for 2 GHz chirped pulse: $775

– ~$170 if you make your own circuit board– Bandwidth limited to 2 GHz (for now)

• PLL for LO: $150– $10+VCO if you make your own circuit board

• Super-Nyquist content could be used for bandwidth extension• Size, weight, power reductions for in situ applications• Future applications in segmented chirp spectroscopy are

possible

Future directions:LO source for millimeter-wave segmented chirp pulse spectroscopy?

Acknowledgements

• NSF (grant CHE-1214123 and the Graduate Research Fellowship Program)

• NASA (grant NNX09AM84G)• The Pate lab• Steve Shipman• Jeff Groseth• Bob Dengler• The rest of the Blake lab

– Brett McGuire