"; _cf_contextpath=""; _cf_ajaxscriptsrc="/cfthorscripts/ajax"; _cf_jsonprefix='//'; _cf_websocket_port=8578; _cf_flash_policy_port=1244; _cf_clientid='E4231E78386BD7346FEB9570C26E56A2';/* ]]> */
Optical Spectrum Analyzer Tutorials
Browse Our Selection of Optical Spectrum Analyzers
Optical Spectrum Analyzer Design
This tab describes the key concepts and implementation of the design used in Thorlabs' Optical Spectrum Analyzers.Contents
Click to Enlarge
Schematic of the optical path in Thorlabs' OSA, detailing the dual retroreflector design. We will refer to this schematic throughout this tutorial.
Thorlabs' Fourier Transform Optical Spectrum Analyzer (FT-OSA) utilizes two retroreflectors, as shown in the figure to the right. These retroreflectors are mounted on a voice-coil-driven platform, which dynamically changes the optical path length of the two arms of the interferometer simultaneously and in opposite directions. The advantage of this layout is that it changes the optical path difference (OPD) of the interferometer by four times the mechanical movement of the platform. The longer the change in OPD, the finer the spectral detail the FT-OSA can resolve.
After collimating the unknown input, a beamsplitter divides the optical signal into two separate paths. The path length difference between the two paths is varied from 0 to ±40 mm. The collimated light fields then optically interfere as they recombine at the beamsplitter.
The detector assembly shown in the figure to the right records the interference pattern, commonly referred to as an interferogram. This interferogram is the autocorrelation waveform of the input optical spectrum. By applying a Fourier transform to the waveform, the optical spectrum is recovered. The resulting spectrum offers both high resolution and very broad wavelength coverage with a spectral resolution that is related to the optical path difference. The wavelength range is limited by the bandwidth of the detectors and optical coatings. The accuracy of our system is ensured by including a frequency-stabilized (632.991 nm) HeNe reference laser, which acts to provide highly accurate measurements of beam path length changes, allowing the system to continuously self-calibrate. This process ensures accurate optical analysis well beyond what is possible with a grating-based OSA.
Each OSA model has a spectral resolution of 7.5 GHz, or 0.25 cm-1. The resolution in units of wavelength is dependent on the wavelength of light being measured. For more details, see the Resolution and Sensitivity section below. In this context, the spectral resolution is defined according to the Rayleigh criterion and is the minimum separation required between two spectral features in order to resolve them as two separate lines. These spectral resolution numbers should not be confused with the resolution when operating in the Wavelength Meter mode, which is considerably better.
The Thorlabs FT-OSA utilizes a built-in, actively stabilized reference HeNe laser to interferometrically record the variation of the optical path length. This reference laser is inserted into the interferometer and closely follows the same path traversed by the unknown input light field. To reduce the presence of water absorption lines in the MIR region of the spectrum, our OSAs feature two quick-connect hose connections (1/4" ID) on the back panel, through which the interferometer can be purged with dry air or nitrogen. Our Pure Air Circulator Unit, which uses hosing that can be directly inserted into these connectors, is ideal for this task.
Click to Enlarge
OSA Resolution vs. Wavelength of the Unknown Input
The resolution shown here was calculated using the formula to the left, using Δk = 1 cm-1 for Low Resolution Mode and Δk = 0.25 cm-1 for High Resolution Mode. Although the formula is valid for all OSA models, the usable wavelength range of each model is limited by the bandwidth of the detectors and optical coatings.
The resolution of this type of instrument depends on the optical path difference (OPD) between the two paths in the interferometer. It is easiest to understand the resolution in terms of wavenumbers (inverse centimeters), as opposed to wavelength (nanometers) or frequency (terahertz).
Assume we have two narrowband sources, such as lasers, with a 1 cm-1 energy difference, 6500 cm-1 and 6501 cm-1. To distinguish between these signals in the interferogram, we would need to move away 1 cm from the point of zero path difference (ZPD). The OSA can move ±4 cm in OPD, and so it can resolve spectral features 0.25 cm-1 apart. The resolution of the instrument can be calculated as:
where Δλ is the resolution in pm, Δk is the resolution in cm-1 (maximum of 0.25 cm-1 for this instrument) and λ is the wavelength in µm. The resolution in pm as a function of wavelength, converted using this formula, is shown in the graph to the right.
The resolution of the OSA can be set to High or Low in the main window of the software. In high resolution mode, the retroreflectors translate by the maximum of ±1 cm (±4 cm in OPD), while in low resolution mode, the retroreflectors translate by ±0.25 cm (±1 cm in OPD). The OSA software can cut the length of the interferogram that is used in the calculation of the spectrum in order to remove spectral contributions from high-frequency components.
The sensitivity of the instrument depends on the electronic gain used in the sensor electronics. Since an increased gain setting reduces the bandwidth of the detectors, the instrument will run slower when higher gain settings are used. The figures below show the dependency of the noise floor on the wavelength and OSA model.
The OSA is also designed so that it samples more points/OPD when the translation of the retroreflector assembly is slower. The data sampling is triggered by the reference signal from the internal stabilized HeNe laser. A phase-locked loop multiplies the HeNe period up to 128X for the highest sensitivity mode. This mode can be very useful when the measured light is weak and broadband, causing only a very short interval in the interferogram at the ZPD to contain all the spectral information. This portion of the interferogram is normally referred to as the zero burst.
Click to Enlarge
Noise Floor in Absolute Power Mode
Absolute Power mode is recommended for narrowband sources. The OSA203C noise floor was measured in low-temperature mode.
Click to Enlarge
Noise Floor in Power Density Mode
Power Density mode is recommended for broadband sources. The OSA203C noise floor was measured in low-temperature mode.
The vertical axis of the spectrum can be displayed as Absolute Power or Power Density, both of which can be displayed in either a linear or logarithmic scale. In Absolute Power mode, the total power displayed is based on the actual instrument resolution for that specific wavelength; this setting is recommended to be used only with narrow spectrum input light. For broadband devices, it is recommended that the Power Density mode is used. Here the vertical axis is displayed in units of power per unit wavelength, where the unit wavelength is based upon a fixed wavelength band and is independent of the resolution setting of the instrument.
The interference pattern of the reference laser is used to clock a 16-bit analog-to-digital converter (ADC) such that samples are taken at a fixed, equidistant optical path length interval. The HeNe reference fringe period is digitized and its frequency multiplied by a phase-locked loop (PLL), leading to an extremely fine sampling resolution. Multiple PLL filters enable frequency multiplication settings of 16X, 32X, 64X, or 128X. At the 128X multiplier setting, data points are acquired approximately every 1 nm of carriage travel. The multiple PLL filters enable the user to balance the system parameters of resolution and sensitivity against the acquisition time and refresh rate.
A high-speed USB 2.0 link transfers the interferogram for the device under test at 6 MB/s with a ping-pong transfer scheme, enabling the streaming of very large data sets. Once the data is captured, the OSA software, which is highly optimized to take full advantage of modern multi-core processors, performs a number of calculations to analyze and condition the input waveform in order to obtain the highest possible resolution and signal-to-noise ratio (SNR) at the output of the Fast Fourier Transform (FFT).
A very low noise and low distortion detector amplifier with automatic gain control provides a large dynamic range, allows optimal use of the ADC, and ensures excellent signal-to-noise (SNR) for up to 10 mW of input power. For low-power signals, the system can typically detect less than 100 pW from narrowband sources. The balanced detection architecture enhances the SNR of the system by enabling the Thorlabs FT-OSA to use all of the light that enters the interferometer, while also rejecting common mode noise.
Click to Enlarge
A Typical Interferogram
The interferograms generated by the instrument vary from 0.5 million to 16 million data points depending on the resolution and sensitivity mode settings employed. The FT-OSA software analyzes the input data and intelligently selects the optimal FFT algorithm from our internal library.
Additional software performance is realized by utilizing an asynchronous, multi-threaded approach to collecting and handling interferogram data through the multitude of processing stages required to yield spectrum information. The software's multi-threaded architecture manages several operational tasks in parallel by actively adapting to the PC's capabilities, thus ensuring maximum processor bandwidth utilization. Each of our FT-OSA instruments ships complete with a laptop computer that has been carefully selected to ensure that both the data processing and user interface operate optimally.
When narrowband optical signals are analyzed, the FT-OSA automatically calculates the center wavelength of the input, which can be displayed in a window just below the main display that presents the overall spectrum. The central wavelength, λ, is calculated by counting interference fringes (periods in the interferogram) from both the input and reference lasers according to the following formula:
Here, mref is the number of fringes for the reference laser, mmeas is the number of fringes from the input laser, nref is the index of refraction of air at the reference laser wavelength (632.991 nm), and λref,vac is the vacuum wavelength of the reference laser. nmeas is the index of refraction of air at the wavelength λmeas,vac and is determined iteratively from λmeas,air (that is, the measured wavelength in air) using a modified version of the Edlén formula.
The resolution of the FT-OSA operating as a wavelength meter is substantially higher than the system when it operates as a broadband spectrometer because the system can resolve a fraction of a fringe up to the limit set by the phase-locked loop multiplier (see the Interferogram Data Acquisition section above). In practice, the resolution of the system is limited by the bandwidth and structure of the unknown input, noise in the detectors, drift in the reference HeNe, interferometer alignment, and other systematic errors. In Wavelength Meter mode, the system has been found to offer reliable results as low as ±0.1 pm in the visible spectrum and ±0.2 pm in the NIR/IR (see the Specs tab for details).
The software evaluates the spectrum of the unknown input in order to determine an appropriate display resolution. If the data is unreliable, as would be the case for a multiple peak spectrum, the software disables the Wavelength Meter mode so it does not provide misleading results.
The FT-OSA instruments incorporate a stabilized HeNe reference laser with a vacuum wavelength of 632.991 nm. The use of a stabilized HeNe ensures long-term wavelength accuracy as the dynamics of the stabilized HeNe are well-known and controlled. The instrument is factory-aligned so that the reference HeNe and unknown input beams experience the same optical path length change as the interferometer is scanned. The effect of any residual alignment error on wavelength measurements is less than 0.5 ppm; the input beam pointing accuracy is ensured by a high-precision ceramic receptacle and a robust interferometer cavity design. No optical fibers are used within the scanning interferometer. The wavelength of the reference HeNe in air is actively calculated for each measurement using the Edlén formula with temperature and pressure data collected by sensors internal to the instrument.
For customers operating in the visible spectrum, the influence of relative humidity (RH) on the refractive index of air can affect the accuracy of the measurements. To compensate for this, the software allows the assumed RH value to be set manually. The effect of the humidity is negligible in the infrared.
The ability to measure low-level signals close to a peak is determined by the optical rejection ratio (ORR) of the instrument. It can be seen as the filter response of the OSA, and can be defined as the ratio between the power at a given distance from the peak and the power at the peak.
If the ORR is not higher than the optical signal-to-noise ratio of the source to be tested, the measurement will be limited by the OSA's response, rather than reflecting a true property of the tested source. The table to the right provides an example.
720p Resolution Setting
Full Screen Button
Software Tutorial Videos
To help customers learn about, use, and understand the Optical Spectrum Analyzer software, we have prepared several short narrated videos that describe the basic aspects of the software and the optimal settings for common types of measurements. Although the OSA model shown in the videos has been discontinued, the principles of operation have not changed.
Full Screen, 720p Resolution Recommended
Analyzing Pulsed Sources Using the OSA
Introduction and Summary of Results
In summary, for pulse rates over 30 kHz, standard mode can be used because the repetition rate is greater than the detectors' bandwidth. For broadband signals with low repetition rates, care must be taken to ensure that the "zero burst" of the interferogram coincides with one of the pulses. Also, when using a pulsed source "Automatic Gain" does not work properly, so the user must monitor the interferogram and manually set the gain so that a strong, but not saturated, signal is obtained.
Impact of a Pulsed Source on the Interferogram and Spectrum
Click to Enlarge
Figure 2: Stacked spectra for 55 pulse repetition rates between 100 Hz and 100 kHz for a 1550 nm DFB laser diode. The intensity is mapped in a logarithmic scale. OSA settings: High Resolution, High Sensitivity, No Apodization, 5 averages.
Mathematically, the resultant spectrum of a pulsed source can be described by a convolution between the spectrum of the light source and the spectrum corresponding to the pulses. As a result, the impact of these artifacts will vary with the pulse repetition rate and the modulation depth of the light source as well as the OPD sample rate (cm/s) of the OSA. The modulation depth of the light source determines the amplitude of the spectral ghosts; a weak modulation yields weak spectral ghosts while a modulation of 100% (on-off pulsation) yields the strongest spectral ghosts.
Figure 2 shows how the behavior of the spectral ghosts as a function of the pulse repetition rate for a narrowband source. In the figure, the spectra were measured for 55 pulse repetition rates between 100 Hz and 100 kHz for a 1550 nm DFB laser diode. We have offset the y-axis such that the true peak (the light gray horizontal line) has been centered at a relative frequency of 0 THz. The figure can be divided into three regions: fp ≤ 3 kHz, 3 kHz < fp ≤ 30 kHz and fp >30 kHz. For fp ≤ 3 kHz, the spectral ghosts are clearly observed symmetrically about the true peak within the resultant spectrum, and move farther and farther away from the true peak as the repetition rate increases. The second region starts above 3 kHz, when the first spectral ghosts have moved beyond the spectral range of the OSA. However, aliasing / folding create higher order spectral ghosts that appear within the spectral range of the OSA. In the third region, fp > 30 kHz, the resulting spectrum agrees very well with the CW spectrum because the repetition rate of the source has extended beyond the bandwidth limit of the detectors. As a result, the pulsed source appears like a CW source to the OSA electronics.
"Pulsed Mode" Operation
Click to Enlarge
Figure 3: Screenshot of the OSA software in Pulsed
Mode; the icon is indicated with a red circle.
Click to Enlarge
Figure 4: (Left) Measured spectra for a narrowband light source pulsed at 1 kHz with (from top to bottom) Low, Medium-Low, Medium-High, and High sensitivity settings (i.e. a decreasing OPD sample rate from top to bottom). (Right) Measured spectrum using the Pulsed Mode, i.e., a minimum hold combination of spectra similar to those shown in the bottom left plots.
Narrowband Light Source
Figure 5 shows the resultant spectra for the source in CW mode as well as four different pulse repetition rates between 100 Hz and 100 kHz. As the pulse rate increases, the spectral ghosts (as recorded in the high sensitivity mode) move further and further away from the true laser peak until nearly identical spectra are obtained at 100 kHz.
Broadband Light Source
In general, the spectral ghosts are less visible for the broadband peak compared to a narrowband peak. However, the noise floor is higher and the spectral ghosts are clearly seen for a repetition rate of 1 kHz and 13 kHz in Figure 6. Similar to the narrowband source, the spectral ghosts move farther and farther away from the true peak with increasing repetition rate. For a repetition rate of 100 kHz both the measurement using high sensitivity and pulsed mode agree well with the CW measurement. As seen, the shape of the peak is slightly different for the CW spectrum compared to the pulsed spectrum. This is not related to the behavior of the OSA but due to a true change in the peak during pulsed operation, e.g., a lower chip temperature.
It is extremely important to note that in general, one has to be careful when measuring broadband peaks at low repetition rates. Since most of the information in the interferogram is located about the zero burst, the peak can be completely missed if the zero burst coincides with no light falling on the detector as shown in Figure 7.
Click to Enlarge
Figure 8: (Top) Central portion of a captured interferogram from a broadband femtosecond laser. (Bottom) Measured spectrum captured using an OSA201 (red line) and a measured reference spectrum captured using a scanning grating-based OSA (blue line).
Femtosecond Pulsed Laser
Figure 8 shows the interferogram collected during acquisition, which does not contain any empty slots. This was expected as the 85 MHz repetition rate of the laser is well beyond the 40 kHz bandwidth of the OSA's detectors. Furthermore, the spectrum measured by the OSA agrees very well with the reference spectrum captured using a grating-based OSA that is scanned slowly enough to provide adequate signal for each wavelength measured.
Click to Enlarge
Hose Connections for Purging OSA Cavity
Gas Detection and Identification Using an Optical Spectrum Analyzer
As shown in the table to the right, many of Thorlabs' Optical Spectrum Analyzers (OSAs) offer detection extending into the mid-infrared (MIR) region of the spectrum, where many gaseous species characteristically absorb. Moreover, the software included with all OSA models supports files from the HITRAN database, a spectroscopic reference standard. These files can be fit to measured traces to identify unknown gases. With the ability to fit multiple analytes simultaneously and built-in hose connections (compatible with Thorlabs' Pure Air Circulator Unit) for purging the interferometer's cavity of trace gases, these OSAs are ideal for use in home-built gas detection setups.
Click to Enlarge
A gas detection setup using the OSA203C. A multipass cell is constructed around the sample chamber () in order to provide high detection sensitivity for the gaseous species sealed inside.
Assigning Peaks in an Unknown Spectrum
The user may optionally allow the software to shift the reference spectrum in wavelength in order to account for measurement effects related to the sample environment. In the case of gas mixtures (i.e., fits performed using more than one reference spectrum), the software scales the intensity of each reference as needed to reproduce the measured spectrum. As shown in the figure below to the right, the output of the fit operation is a graph comparing the measured spectrum, each scaled (and possibly also shifted) reference spectrum, and the sum of the scaled reference spectra.
Click to Enlarge
In the Reference Fit Setup tab, checkboxes are used to indicate which gaseous species to consider in the fit. The absorption lines can be either "fixed" or "free"; the latter allows the software to shift the reference spectrum in wavelength. The measurement conditions for the HITRAN references are also displayed.
Click to Enlarge
In the Reference Fit Result tab, the fitted spectrum is displayed simultaneously with the measured spectrum. The fitted spectrum is the sum of the scaled reference spectra included in the fit. The scaled spectrum for each individual gaseous species is also shown.