RoadTest: Tektronix RSA306 USB Spectrum Analyzer
Author: Gough Lui
Evaluation Type: Independent Products
Did you receive all parts the manufacturer stated would be included in the package?: True
What other parts do you consider comparable to this product?: Hobby software-defined-radios (SDR) such as RTL-SDR, Nuand BladeRF x40, Winradio G31DDC
What were the biggest problems encountered?: Some software options are placed in slightly counter-intuitive positions, and automatic acquisition and analysis settings are sometimes not sufficient to get multiple displays functioning simultaneously.
RoadTest Review: Tektronix RSA306 Real-Time Spectrum Analyzer
I would like to thank Tektronix and element14 for being so generous in providing the kit and choosing me as the RoadTester for this product. As usual, this is going to be a thorough and impartial review.
I hope you all enjoy this review, which I have put in a lot of time in creating. This is a highly sophisticated package, which means that there is lots of ground to cover. I've tried to present it in the most logical progression, although in reality, some of the experiments and tests were not done in the order presented.
My initial plan was to review it over the new-year break, but that was not possible because of the shipping schedule (items received 2nd February 2015). I was somewhat under-prepared with this one, as I am more an RF-enthusiast in the short-wave listener, ham and satellite sense with access and knowledge about many different signals on the air, and had less experience with spectrum analyzers. I lost some time in ordering and waiting for adapters and antennas to arrive. Furthermore, continuing commitments with my PhD research in writing papers meant I wasn't quite able to perform as many experiments as I would have liked.
That being said, it has been a busy two-months with the unit, and it's time to turn in a verdict. Please feel free to leave a comment, leave a rating, give a like, or visit me at my personal blog at http://goughlui.com
In the past decade, the area of radio communications has been forever changed by the reasonably priced availability of high-rate analog-to-digital converters and digital-to-analog converters, along with a wealth of digital signal processing power enabled by the CPUs in modern computers. This ushered in the era of the "software defined radio" which allows us to emulate some of the the physical, analog components of a radio, such as filters, mixers and detectors, in the digital domain.
A software defined radio receiver generally minimises the analog domain to the signal conditioning required to isolate the bandwidth of interest and provide it in a format suitable for encoding by an analog to digital converter -commonly used components include, bandpass filters, impedance matching transformers and variable gain amplifiers. The high-rate digitized data is then processed by a computer to perform the "demodulation" itself.
Such a design, while initially quite difficult to implement due to high DSP computation requirements, proves to have several advantages over analog radios and front-ends, by allowing for the monitoring of a wider bandwidth in real time, enabling the implementation of ever-higher-bandwidth modulation modes with sophisticated forward-error correction and equalization features, as well as freeing us from the bulk, imperfections and drift experienced by analog components.
When I first started with software defined radio back in 2010, I had invested a hefty sum (as a student) into a Winradio G31DDC 0-50Mhz USB 2.0 receiver. This unit offered continuous monitoring of just 2Mhz of spectrum at 16-bit resolution in its "digital IF" over the USB 2.0 link, and it required the best dual-core CPUs available at the time to achieve this. Regardless of the initial frustrations, I still use this unit today as a short-wave listener, utility monitor and ham alike, and it's relatively "easily" handled by modern equipment. This SDR was nothing short of a revelation - I could examine a whole 2Mhz band of channels at once, in real-time, and I could record the whole band and play it back at any time - re-tuning into any "narrowband" channel within this range. I didn't have to sit behind an old fashioned radio's VFO, tuned into a quiet channel, wondering if there was some action elsewhere across the band.
The recent popularization of the RTL-SDR, based around the RTL2832U USB DVB-T TV Tuner chipset, shows just how useful the SDR concept is, even with only 8-bits resolution and about 2.4Mhz of spectrum available. Real-time ADSB aircraft monitoring, formerly very difficult due to the bandwidth requirements of 1-2Mhz, were now in reach of the average person. More importantly, this one was cheap, and accessible to many, with frequency ranges of about 70-1200Mhz (more or less depending on your front-end).
All of this action has resulted in a wide range of SDR-based devices, most of which are aimed at the hobbyist, and do not offer a guaranteed flat-pass-band, lack of signal images or response linearity. Naturally, such a design would be very applicable for test equipment as well, which are needed to keep up with the new wide-band modes which are becoming more common, but impose much more stringent requirements on the implementation. This has ushered in a new age of "real-time software defined" spectrum analyzers.
Traditional spectrum analyzers can be thought of as very similar to a traditional analog radio, which is swept over a range of frequencies of interest, producing a plot of the detected signal power at a given frequency. Such an approach has many disadvantages, mainly due to the accuracy of such an analog front-end detector when it comes to sweep response at high sweep rates, and the fact that you will miss or distort short transient burst signals because you're not really looking at the whole spectrum all the time. Furthermore, with very old spectrum analyzers with limited memory capabilities, your ability to analyze such captured signals may be limited to a set of basic measurements offered by the unit itself (e.g. markers) and any screen-captures you might be able to get. With more and more demanding digitally-based RF applications in radar and wireless communications, such issues can make proper measurements very difficult, frustrating or impossible to make.
Another consideration is the size and bulk of traditional bench-top units which make them less suitable for field work and interference hunting/direction finding. Early attempts at small, hand-held units tended to feature much poorer specifications, especially when it comes to bandwidth and screen resolution because of the need to squeeze numerous analog components into small cases and compromises in the design making them more difficult to use effectively.
The Tektronix RSA306 tries to bring the advantages of SDR-like technologies into a "software defined instrument". It leverages the SignalVu-PC application along with its measurement options, which has been used with their high end oscilloscopes (MDO/MSO) to do spectrum analysis, but offers it at a much more compelling price of US$3490 from Newark/element14. Instead of providing you a "whole" oscilloscope, you get a 3.5" hard-disk sized external unit, which connects to your computer via USB 3.0 and uses your computer to do all the analysis and display.
The RSA306 unit itself has a frequency range of 9khz to 6.2Ghz, from +20dBm down to -160dBm. It utilizes a 14-bit 112MS/s analog to digital converter to perform its sampling, an FPGA to perform data manipulations, a Cypress Semiconductor bridge chip to move the data over USB 3.0 and provides a 40Mhz real-time spectrum display. The RF connection is made via a 50-ohm N-type connector, with two SMA connectors for trigger and external 10Mhz frequency reference input. Connection to a computer is by a USB 3.0 connection, which also supplies power to the unit. Teardowns of the unit have been performed by KF5OBS and is further discussed elsewhere, where disagreements on the interpretation of the internal design still persist.
With such a high-specification device, pushing in excess of 224MiB/s of data to your computer for analysis, the system requirements are suitably high. It is recommended that you use a system equipped with an Intel Core i7 4th-generation CPU, SSD storage for streaming recording (>300MB/s write), 8Gb RAM, USB 3.0 and Windows 7 or above, 64-bit edition. The unit will function with machines with slower CPUs but analysis may be more limited and it may affect the probability of intercept, and you will lose the streaming recording if you use a slower storage device. However, 8Gb RAM, USB 3.0, and 64-bit Windows is NOT negotiable, so if you don't already have a PC of sufficient specifications, you should factor it into your purchasing decision as well.
The unit and software was somewhat mysterious to me at first, as I had many questions which took some digging to answer. Tektronix currently has a range of very useful documentation for the RSA306 available by searching RSA306 in their manuals site and elsewhere on the site. I recommend users and prospective owners to read it all (as I have done), as some information isn't quite where you might expect it - but it's there!
The most important documents (and a summary of their contents) are:
- RSA306 USB Real Time Spectrum Analyzer Datasheet - List of most basic specifications and options available for the RSA306.
- Spectrum Analyzer Installation and Safety Instructions Manual - Regulatory and safety information, list of available accessories, getting started with installation of software and hardware connections.
- Spectrum Analyzer Application Programming Interface (API) Programmer Manual - Main API reference for direct access to the RSA306 through C, C++ and Python. Also documents the recording .r3a, .r3h, .r3f file formats.
- USB Spectrum Analyzer Self-Guided Demo Training Manual - Mainly for use with the RTSA Demo Board, but does introduce you to the SignalVu-PC software. Some demo files are bundled with SignalVu-PC which can be used even without hardware.
- USB Spectrum Analyzer Specifications and Performance Verification Technical Reference - Detailed and complete technical specifications, as well as procedures used to verify the performance meets specifications. It's great for avoiding potential disagreements over specifications, but it is not a calibration document though.
- Spectrum Analyzer Declassification and Safety Instructions Manual - Information on how to sanitize and remove memory devices from the RSA306, which amounts to pretty much physically destroying the product.
- SignalVu-PC Vector Signal Analysis Software Datasheet - Detailed information on the measurement capabilities of SignalVu-PC base edition and options.
- SignalVu-PC Vector Signal Analysis Software Programmers Manual - Information for Ethernet-based SCPI remote-control of the spectrum analyzer through TekVISA.
- SignalVu-PC Read This First - Details hardware requirements, installation and activation procedures for the SignalVu-PC software suite.
- SignalVu-PC Vector Signal Analysis Software Online Help - Complete printable reference to how to use and configure the software and its options, although written mainly for the RSA5100A, so not all options may be available for the RSA306.
- eGuide to RF Signals - A beginners guide which shows examples (with example audio, files, video) of the types of on-air signals users may experience, interactive and requires internet connection. More useful for educational purposes, especially for those who haven't laid eyes on an SDR or spectrogram before.
Do note that the documents for the SignalVu (i.e. running on the signal analyzers) and SignalVu-PC (i.e. running on the PC) are slightly different, so do make sure to download the right version of document!
I suppose one disappointment with the documentation was a lack of block diagram explaining the architecture of the device - i.e. what reference levels engaged which gain amplifiers, what protection and filters are provided on the front-end, what internal IF/mixers are used, etc. From the datasheet and verification reference, it is claimed that a 28 +/- 0.5Mhz digital IF is used, and that reference level steps of 20dBm, 0dBm, -13dBm and -30dBm exercise all gain amplifier conditions. Some of the other details can be deduced from the teardown video, and performance verification document above to some extent, but without great levels of certainty. This may be because the design is commercially sensitive, but it would make understanding the limitations and best applications of the device a little easier.
The RSA306 comes in a relatively small (by Tektronix standards) cardboard box.
As usual, the items are presented neatly, and the packaging involves thick padding and adequate clearance to ensure no damage in transit.
Calibration certificates, regulatory, safety registration information and the installation media were included. Moving along with the times, the software is provided on a USB memory device rather than an optical disc (which is good for those without an optical disc drive).
It seems that this package may have already been opened prior to my receiving it, as the pouch where the software would have been expected to be placed was empty, but it comes with a sensible advisory to check online for the latest version of SignalVu-PC.
The unit is constructed very robustly, with its metal body shield encased in a thick rubber sleeve which is glued into place. The sleeve does pick up some dirt, but it also does an excellent job of protecting the unit against sharp impact and letting it stand "on end" to save bench space.
There is no flex in the unit at all. There is a rubber cap for protecting the N-connector as well, although the place where it joins onto the body seems to have peeled a bit.
Also provided with the unit is an L-com branded USB 3.0 cable of about 2m length. The connectors on the end seem to be high quality and fit easily and securely without requiring excessive force.
Tektronix have been generous enough to "throw in" a Lenovo ThinkPad Edge E431 to make sure we have something decent to hook the RSA306 up with. Seeing as the star of the review is the RSA306, I won't be focusing that much, photo-wise, on the laptop.
This particular unit would have been the best specification available at the end of 2013, when it was manufactured.
The packaging was already disturbed, but the laptop was clean. It seemed that someone went to the effort to test the units out and pre-install the software, and label the laptop. The power-on hours count on the SSD was a very low 13 hours.
There was a need to wrangle with the laptop to update all of its software, as it was provided in a fresh, un-upgraded state - in the end, it was running the latest version of Windows 8.1 Professional after wrangling with some driver difficulties.
Seeing as this instrument runs predominantly on a computer, I felt that explaining some of the experiments with just screenshots didn't really do the unit justice and could be a little difficult for readers to follow along, especially if they haven't already been exposed to the SignalVu-PC software already. As a result, I ended up buying an HDMI capture card just so I can make this 20 minute video showing some of the experiments and possibilities behind the RSA306.
The review will make more sense if you watch it - I promise ;).
Tektronix were generous enough to provide a laptop along with the RSA306 to ensure that us RoadTesters can experience the best of what the RSA306 has to offer. This was a very nice gesture, of course, but lets examine their choice of laptop.
The laptop comes with the following specifications:
- 14" 1366x768 Touch-Screen
- Intel Core i7-3632QM with Intel HD4000 Integrated Graphics
- Samsung OEM 128Gb 2.5" SATA SSD
- 8Gb DDR3L 1600Mhz RAM
- DVD+/-RW drive
- 2x USB 3.0 + 1x USB 2.0
- HDMI and VGA Output, Gigabit Ethernet, Intel Centrino Wireless N 2230 (2x2 2.4Ghz with Bluetooth 4.0), 720p Webcam with Array Mic, Fingerprint Sensor, Card Reader
- Windows 8 Professional 64-bit edition
On the whole, this is a relatively high-specification laptop for when it was built (as the warranty was about to expire within a week of receiving the unit), but is probably not the best choice today when it comes to pairing up a laptop with the RSA306. Users should probably opt for a more recent model sporting at least an Intel Core i7-4700MQ/HQ to ensure they meet the 4th-generation Core i7 requirement, and also check that their SSD is actually capable of 300+MB/s writes (i.e. choose or install a large capacity "pro" series SSD rather than a value-oriented model). When I tested the included SSD, it was not capable of meeting the requirements.
Fortunately for them, I cared enough about the laptop that I spent some of my own money to replace the 120Gb Samsung OEM SSD with a 480Gb Sandisk Extreme, and upgrade the RAM to 16Gb to make it move just a little faster. For security and functionality reasons, a lot of time was spent updating Windows to Windows 8.1 (i.e. the latest edition with all patches applied), which involved some difficulty with certain drivers, and the loss of USB-over-IP Wi-Di functionality. After all of the tinkering, I would estimate that the i7-3632QM used in the laptop doesn't represent that much of a limitation to the analysis power after all and seems sufficient for the average use case.
The RSA306 is supplied with a USB key containing all the relevant documentation and software, which is a nice touch given that more and more computers are starting to eschew optical drives. However, because the software itself is upgraded very often, the version that is provided is more than likely out of date, and so downloading a new version is highly recommended. The Tektronix SignalVu-PC software is available for download online, each release weighing in at about 900Mb.
Portions of this review were done with SignalVu-PC versions 3.4.0245, 3.4.0253, 3.5.0119 and 3.5.0134. The software was run on the supplied Lenovo ThinkPad Edge E431 running Windows 8.1 64-bit, as well as an Intel i7-4770k-based system and an older AMD Phenom-II x6 1090T based system, both running Windows 7 SP1 64-bit.
The software arrives as a self-extracting ZIP file, containing all the necessary files to use the API as well as install the SignalVu-PC package. Installation is straightforward, by following prompts from an install wizard, which will also install other supporting software including TekVISA. On a fresh machine, this is easily accomplished, followed by a reboot to get all the required services running. Two new icons on the desktop are created - one for Tektronix SignalVu-PC, and another for Tektronix RSAMap.
A feature of the software, which is great in my opinion, is that you can download, install and test out the software before you've purchased any hardware. The base edition of SignalVu-PC offers the basic signal analysis features, costs nothing and comes with sample files which you can practice analyzing and begin to get comfortable with the interface. Best of all, is that you can continue to use the software later on to analyze captures you have made with real hardware - so in the case of a shared lab, the time with the instrument may be primarily spent capturing data, with analysis happening "off-line" later, somewhere else, on another computer.
Another nice feature is the option to enable 30-day evaluations of all the available option modules. The modules generally retail from about US$990 and upwards (from what I can find), so it's good to know what is possible and whether it fits your needs before spending more money. The list of options include:
- Option SVA AM/FM/PM/Direct audio analysis
- Option SVT Settling Time (frequency and phase) measurement
- Option SVM General purpose modulation analysis
- Option SVP Advanced Signal Analysis (including pulse measurements)
- Option SVO Flexible OFDM Analysis
- Option SV23 WLAN 802.11a/b/g/j/p measurement application
- Option SV24 WLAN 802.11n measurement application (requires option SV23)
- Option SV25 WLAN 802.11ac measurement application (requires option SV24). Limited to 40 MHz bandwidth on RSA306
- Option SV26 APCO P25 measurement application
- Option SV27 Bluetooth Basic LE Tx measurement
- Option MAP Mapping and signal strength
- Option CON SignalVu-PC live link to the MDO4000B series mixed-domain oscilloscopes
- Option SIGNALVU-PC-SVE SV2C Live Link to MDO4000B and WLAN 802.11a/b/g/j/p/n/ac measurements (includes options CON, SV23, SV24 and SV25)
The evaluations can be enabled without an internet connection, whereas the final purchased keys need to be activated via the internet either directly on the machine which is to be running the software, or via another machine connected to the internet through the activation tool.
It will be evident to readers that the RSA306 is not bundled with any accessories except for the USB 3.0 cable. There are no antennas nor adapters provided. I think this is a sensible move by Tektronix, as most people who would be purchasing the RSA306 are probably experienced RF engineers with a set application in mind, and this often means direct cable connection through attenuators, very specific antenna arrays (e.g. log periodic dipole, loops) or near field probes which are often quite costly. Bundling in any sort of antenna is not likely to be able to satisfy the intended end user, and would probably just end up as a waste. Adapters would probably be nice, although, their omission is entirely understandable from my point of view.
For those who need absolutely the best, it should be noted that Tektronix do have a list of accessory parts including antennas covering certain ranges and adapters for use with the RSA306, which are a separate purchase. Accessories, including a bag, carry cases, are also available.
Unfortunately, being the "old-hat" sort of person that still primarily uses BNC connectors despite their frequency-range shortcomings and the more fragile SMA connectors for most things, I didn't have any stock of N-connector adapters so I had to order some in. With a limited budget, I opted for the less-pricey option of "China" made generic adapters, which is hardly optimal and likely of sub-par, but sufficient quality for use. These took a month to arrive, but it was definitely necessary to allow me to have sufficient access to different signals to give the unit a thorough test. Further to that, I decided to purchase a telescopic dipole antenna as well for very basic field usage through a wide range of frequencies (based on contracting/extending the antenna different amounts).
Starting the software is as simple as double-clicking the appropriate shortcut, where you are then greeted by a splash screen, followed by the main user interface.
From there, you can begin your analysis tasks by reconfiguring displays, and loading stored files, or by connecting to the RSA306.
The interface is driven mainly by menu, with a shortcut bar as well. The menus are File, View, Run, Replay, Markers, Setup, Presets, Tools, Live Link, Window and Help. Some of the key features are explored in the following section.
Other than that, the graph-area is also useful for navigation, by a right-click pop-up menu which allows you to perform panning and zooming.
The first time you wish to connect to an RSA306, a search for instruments needs to be performed which takes about 30-seconds.
Once the instrument is found, you can connect to it, which takes another 20-seconds or so.
While spectrum measurements begin to show as the instrument is placed into a run state, however, you're not quite ready to use it to make accurate measurements just yet. Tektronix recommends leaving the instrument for a warm-up period of 30 minutes before performing the alignment run.
The alignment option itself needs to be used fairly frequently to ensure the readings from the RSA306 are within their specifications. Certain changes to reference level and centre frequency seem to trigger the "not aligned" status message. The alignment process is somewhat frustrating because of the need to manually remove any RF input before the alignment is run. Given how frequent the alignment process is needed, I would prefer if the alignment status was shown as a coloured square on the toolbar (as that would make it very easy to see at a glance if we're all good), and that the alignment shortcut be placed as a quick-button on the tool-bar itself.
There also seems to be an issue if the alignment includes the spike at DC/0Hz on the spectrum display, causing it to fail, so just be aware of that.
The base displays are grouped into three categories - General Signal Viewing, Analog Modulation and RF Measurements. The displays are as shown in the screenshots below, and any combination of displays can be engaged to simultaneously analyze a signal, provided you have enough screen real-estate and CPU grunt to handle it.
Regular presets are available which allow you to reset the displays and settings with one click. The blue "preset" button on the toolbar invokes the main preset which brings everything back to a basic spectrum view. Other pre-defined presets are provided, however, the user may also define their own presets. This can be done by saving the current setup, when you have configured it to your liking, into your C:\SignalVu-PC Files\User Presets folder, which is not as straightforward as it could be.
Application presets are also provided which address certain measurement applications, although due to the limited analysis blocks available, the presets are quite sparse.
The same also goes for the standards presets which are provided to help users perform pre-compliance tasks for P25, Wireless LAN and Bluetooth. More blocks will be added once you have licensed the appropriate displays.
The setup menu allows you to configure options, some of which are independent for each display, and others global. Selecting Settings configures the settings independent for the display . Clicking on the display updates the ribbon across the bottom, where the settings changes are made. These settings are generally grouped into several tabs and need to be explored for specific measurement needs.
The rest of the setup menu generally corresponds to global settings - e.g. acquire ribbon if you need to do streaming recording (more on this, later on) or change the acquisition time, or the analysis ribbon to change the analysis time. Longer analysis or acquisition times let you explore lengthy signals, but will make the software less responsive due to needing to analyze more data.
It seems that early versions of the software didn't have an external trigger option despite the hardware having such an input. It seems like the feature has now been added, although I didn't have a chance to test it. In earlier versions of the software, the trigger based on RF input power was not functional at all, leading to continuous running. The latest version seems to fix this, although, it does retain a rather inflexible trigger bandwidth of 40Mhz which is fixed. Early versions of the software did not feature the Bluetooth options either.
The replay menu allows you to configure the "replay" system. The spectrum analyzer can be considered as capturing a certain number of samples as defined in the acquisition settings, and then analyzing those samples alone while another group of samples is gathered. The results of the analysis and the actual raw data are stored in a buffer during a continuous run, allowing you to stop and go back to look for a signal you might have spotted. The replay menu allows you to choose what you're replaying (i.e. samples into the analysis chain, or the results from the analysis chain) and go through them step by step, continuously in a loop, etc. This can get quite tedious, so the replay options are also available in a separate replay toolbar, as well as a quick button on the main toolbar.
The View option allows you to enable or disable certain views and features. For touch-screen users, the keypad option allows for easier data entry into the frequency and reference level fields.
Markers, as expected, allows you to add/remove or define your markers and can invoke the markers toolbar. Markers are provided for reading out signal levels, which is useful, although limited to four markers and a reference marker. The marker toolbars allow them to be configured (i.e. delta or absolute mode, read-out units, attachment to what trace on the plot) and a readout table provided. I really like the automatic peak detection snapping, which makes accurate readings easily achieved. One frustration, however, is using the marker toolbar, where the "All Markers Off" option is right next to "Delete Marker" which can lead to inadvertent clearing of all markers with no confirmation when you just wanted to remove one marker.
Tools contains license management utilities and mask test features, which allows you to activate 30-day evaluations. Window allows you to rearrange and lock the arrangement of your displays. Help allows you to get help from the on-line help and find information about your instrument and software options.
As I don't already have experience with SignalVu/SignalVu-PC in any form, coming to the software as a new user presents a mild frustration, as some of the settings I had expected were not placed in obvious locations and required some digging around. The learning curve from just playing with the software is fairly steep, but there is sufficient documentation to get around this. It is probably must more productive to read the entire manual first, before using the software as that completely changed my experience from mild frustration to one of feeling comfortable with the tool.
Some of the options are probably not as easily accessed and there is a need to jump around through tabs and different ribbons to get at what you need (e.g. Save and Export in Options, Record in Acquire), so a few more toolbars, maybe even user-customizable ones, might be welcome. I would also like to see a labelled x-axis to make the frequencies easier to read-out without counting grid squares. It would be nice to be able to configure the grid and axes entirely as well - there's no reason to be limited by the old legacy fixed-layout CRO style screens!
One thing to note is that everyone's experience with the unit might differ slightly, especially when used to examine "on air" signals. The problem is the amount of RFI that might be present around the unit - as it relies on a host PC, if you have a "noisy" PC, this can make examining weak on-air signals difficult.
My initial attempts to pull on-air signals was somewhat frustrated by the presence of occasional "spurs" in the read-out. By pure accident, when the power adapter fell off the table onto the ground, the signal level of the spurs changed. I realized that quite a few of them were no fault of the analyzer unit itself and was instead coming from the power adapter that runs the laptop.
I wonder if some of the other spurs could be attributed to the switching converters inside the laptop providing USB bus power, for example. It seems that there are also issues with USB 3.0 causing noise across the spectrum as well, so proper shielded cables and connectors can make a difference here too.
As a result, I think it's important to understand that some of the results could be affected by your environment, and your choice of computing hardware. Unfortunately, it is not easy to understand what the RF noise performance of different hardware combinations are ahead of time.
A figure of merit in comparing spectrum analyzers is the displayed average noise level (DANL). This represents the "noise floor" of the analyzer with no signal input, normalized to dBm/Hz, with lower values being generally preferred.
The procedure detailed in the Specifications and Performance Verification Technical Note was followed, with the input terminated by a 50-ohm terminator, although step 4h, RBW was left at auto which resulted in 1khz, rather than 100Hz as stated in the document (which is possibly a typo). A reference level of -90dBm was chosen to make spurs as easy to spot as possible. A plot of the measured value versus the specifications is shown below. At all specified points, the analyzer exceeded its specifications when spurs were disregarded, as the procedure states.
A distinct difference is seen at a frequency of 20Mhz and 22Mhz due to the internal signal path. The low-frequency path is selected for centre frequency settings of below 22Mhz, whereas the RF path is chosen for 22Mhz and above.
It's important to remember that DANL is just one of the figures of merit - the phase noise, dynamic range, presence and level of spurs, and linearity of response are all important as well. Unfortunately, measuring those parameters are not as straightforward especially without even more expensive precision signal generators.
The same 50-ohm terminated configuration as used for DANL was used in looking for spurs. The datasheet claims the residual spurious response is <-85dBm with a reference level of -50dBm, with exceptions for <-78dBm for harmonics of 112Mhz in 1680-2668Mhz, 4750, 4905-4965Mhz.
Rather than manually leafing through all of the frequencies, I decided to follow the same methodology as above, but increase the span to the maximum, fix the RBW at 1khz, and leave it to plot the average of 100-sweeps. On the whole, it seems that they are able to keep to their specifications quite well, at least at the "zoomed out" state we're looking at, with trace points set to the maximum of 64001.
I suppose it was an unfortunate need in the design of the unit to have some spurs show up in the 2.4Ghz ISM band, which I suspect will be a popular band to analyze with the unit.
One of the touted features of the RSA306 is the ability to do live audio demodulation in SignalVu-PC. Indeed, this is the case, and it is very easy to use. Selecting the Audio option brings up the panel, which allows you to start/stop the demodulator, configure the demodulation mode for the current analysis frequency, adjust the volume and save the audio as a 32khz .WAV file
This is available within the base version of SignalVu-PC and covers the following modes:
- AM 8Khz
- FM 8Khz
- FM 13Khz
- FM 75Khz
- FM 200Khz
This covers the majority of analog modes which most users can expect to see on the air, and is a great bonus to allow for live identification of signals, however, radio amateurs will note that no SSB modes (USB, LSB or ISB) are implemented which seems to be an omission. Furthermore, no variable bandwidth decoding other than the selections provided above, is available. Stereo FM, or decoding of RDS, is not provided either, although it is important to remember that this product isn't intended to be used as an (expensive) radio.
The audio quality seems sufficient for purpose, and the demodulation features generally worked well in my testing in parallel with various displays. When too many analysis features are engaged, the audio can get choppy and exhibit discontinuities, which is expected behaviour.
Overall, my favourite display has to be the DPX, so much that I don't use the spectrum or spectrogram displays much at all. It has very good utility wherever you might consider using a spectrogram or spectrum display. The density-based display allows you to gauge what time-varying signals are doing, and even get a rough understanding of channel utilization for discontinuous-transmission modes. It's nice that this is included even with the base edition of SignalVu-PC and it really gives you a bit of a "CRO" style feel to a digital instrument and saves you from the headache of trying to work out what a very "jumpy" trace on the spectrum display actually means. The split-view is especially useful in combining the density display and a waterfall-style display, and was used in chasing different signals across the bands.
I suppose one of the things most people who get a spectrum analyzer do is have lunch ... or watch someone else have lunch. In my case, it was the latter, but the chicken curry sure smelt nice ... As expected, the microwave that cooked it, did a good job of spewing out radiation all over the upper half of the 2.4Ghz band, chirping with an irregular "dominant" frequency.
The time-varying nature of the chirping was best illustrated by the DPX spectrogram, whereas the spectrum showed us the relative power levels of nearby 2.4Ghz wireless 802.11n APs. It's quite interesting to see that some "more desperate" cards actually did put some data-bursts in-between chirps.
Unfortunately, the 40Mhz is not enough to cover the whole ISM band, so instead, it has to be covered in a "swept" mode. Using 120Mhz, it's possible to squeeze in the whole ISM band (and a little more), which allows us to see Bluetooth frequency hopping activity, 802.11g/n from a heavily loaded AP and 802.11b-rate beacons from a lightly loaded AP.
The time varying nature of the hopping can be seen with a Bluetooth device initiating a scan, in the DPXogram.
Tracking down 2.4Ghz interference is no problem in general, although a small spur is evident within the ISM band range of the RSA306.
The mobile phone service bands in Australia are a bit of a mess, with all the technology rollouts and the three different carriers buying spectrum from a competitive auction. As a result, bands and technologies are not always standardized and result in haphazard heterogeneous band utilization amongst the carriers.
Recently, the introduction of LTE, better known as 4G radio services, has resulted in the refarming of spectrum from the 1800Mhz band (formerly used for dual-band 2G services), as well as new spectrum in the 700Mhz, 2300Mhz, 2600Mhz band to provide services. As the details are a bit thin on whether signals are available from the other bands (as some of the bands are still considered in "trial" phase), it was easier to try directly measuring the downlink spectrum to determine the presence or absence of signals at a given location.
Signals found included a 10Mhz Optus and 20Mhz Telstra carrier in the 700Mhz band, currently termed 4GX by Telstra due to its superior building penetration compared to the traditional 1800Mhz band arrangement.
A surprise was found in the 850Mhz band, which was not expected to house any LTE signals because of its use for UMTS by Telstra, but a 5Mhz wide LTE carrier was also caught at 877.5Mhz. A quick look up at the ACMA Radcom's Database seems to show that it is likely operated by the third carrier - Vodafone.
The 1800Mhz band seems to show three LTE carriers (two 15Mhz, and one 20Mhz) "crowding up" the band and forcing the older dual-band 2G GSM services off the air. The band is also now shared with GSM-R parcels. This doesn't seem to match some information circulated online which suggests 10Mhz LTE carriers for all.
The 2300Mhz band of ~100Mhz width, exclusively operated by Optus was supposed to house a new TDD-LTE rather than the traditional FDD-LTE, intended for use by high-speed data-only services. Checking the band with the spectrum analyzer seems to show two parallel 20Mhz LTE carriers, with the rest of the spectrum left "unoccupied", possibly for uplink purposes.
No signals on the 2600Mhz band were found.
Recently, our local government train operator Sydney Trains had begun phasing out the old analog trunked Metronet network for voice communications with rolling stock. The replacement system was GSM-R - a specialized version of GSM intended for railways usage containing features such as group calling and automatic train protection. These features were needed to meet the Waterfall Enquiry's recommendations in avoiding future train accidents.
I have been an interested hobbyist following the progression of the GSM-R radio network implementation. I have watched the bases go up, similar to mobile phone towers, and plotted locations based on planning data. It was of interest to see if the bases were already active, or were just installed but left switched off.
In order to do this, it was important to exercise the field work capabilities of the unit. GSM-R signals can only be received within the railway corridor, as the network was designed for security, safety and reliability, and ensuring that the signals would not be jammed or otherwise received outside the corridor was a requirement. In order to measure the signals, I had to measure the spectrum while on a train ... not something you see people doing everyday.
I was able to do so, and have a recording of it, by having my RSA306 and dipole antenna in my bag, along with my HDMI capture card. I sat the bag next to me while my laptop was on my lap. By setting up the DPX spectrum view with peak hold, I can tell whether the frequency allocations were used or not. Indeed, it was discovered some of the bases were indeed alive.
I had tried this experiment previously with an RTL-SDR with mixed results. Sure, it's not a test instrument, but it is an SDR which many users are now embracing due to low cost, sometimes without considering its disadvantages.
I had previously concluded the bases were not online, based on not being able to see the signal. However, since 1800Mhz is so far outside the specification of the front-ends used by TV tuners, I couldn't be certain of the sensitivity of the receiver and the noise floor. Further to this, there was a potential for misidentification due to mistuning or signal aliases due to poor filtering, and the narrow ~2.4Mhz bandwidth did not allow for imaging the complete span simultaneously.
I'm very satisfied that by using the RSA306, that I was able to eliminate all these concerns.
Living in Sydney, I am in a location which is covered by numerous weather radar stations. The weather radar stations operated by the Bureau of Meteorology typically broadcast on S-band (~2.8Ghz) at high power. Short pulse trains are sent from a dish with narrow beamwidth, which rotates to scan the sky and receive echoes from clouds and precipitation.
This was one of my first experiments, as I was eager to receive the signal. I never personally had any equipment that could display the necessary bandwidth at this frequency, so I rigged up just a short random length of wire as an antenna. My previous narrow-band receiver was relatively deaf and too narrowband to ensure the detection of the radars which could be within a 200Mhz span.
In order to capture the infrequent pulses which can occur in bursts once every 6 to 10 minutes, I decided to set up one of the traces as peak holding, so I can leave the analyzer running and come back to a series of peaks representing the transmissions from the radar stations.
In all, I have to say it was a success, as adding markers in peak hold allowed me to positively identify each of the weather radar stations by looking up the frequency details at the ACMA's Radcomms Database.
I later narrowed the bandwidth and repeated the experiment with a much more suitable antenna, which confirms the detections as positive.
By splitting the DPX display, you can also see the time varying nature of the pulses - while not resolving each pulse in terms of its pulse repetition frequency, it is capable of showing trains or groups of pulses emitted by the radar.
In conclusion, I have to say that I'm quite happy with the reliability and performance of the unit, as it has effortlessly uncovered the presence of signals which I have been otherwise looking for. The serious instrument-nature of this product overcame many of the uncertainties with sensitivity, interference with "hobby" SDRs and gave me confidence that the signals I were seeing were on the air. Additionally, the SignalVu-PC software performed well out of the box and only required limited adjustments to perform measurements, whereas some hobby-SDRs require more effort in developing a flowgraph or repurposing an existing utility for measurement/spectrum viewing purposes.
The SignalVu-PC application comes with a wealth of options which can be enabled to perform more advanced signals analysis, which is of interest to designers of 802.11 wireless devices, Bluetooth and Bluetooth Smart/Low-Energy devices, radar devices, as well as those involved with digital signals (P25, *PSK/*QAM/*APSK) and analog signals (audio analysis). These options provide additional measurement displays, which are also added into the presets. Activation of the trial can be done through Tools -> Manage Licenses ... and appears to be tied to the computer which the SignalVu-PC application is installed. Activation of a trial does not require internet connection.
Of course, this provides an ideal opportunity for me, as a RoadTester, to get a glimpse into what is capable when all the options are unlocked. I would be lying if I said I wasn't smiling when staring at the wealth of displays now available.
This also refreshes our lists of Standards and Applications presets as well.
Now that we have many more displays, it is important to remember that each display itself has its own needs when it comes to the amount of captured data and analysis time available, and it is very well possible that some combinations of display will not work together because they expect different bandwidths, analysis time, or samples. Further to this, you can run into this issue when you start choosing your own combinations of displays with the acquisition and analysis settings left to the defaults of auto. It pays to read the error messages and perform some tweaks under the settings to see if you can get them to play well.
Some of these settings can enable analysis periods and times which are different to the "global" analysis time values under acquisition and analysis tabs. In fact, the acquisition tab sets the data length and sample rate which is gathered from the RSA306, and the analysis tab sets the "general" analysis time (amount of data analyzed) which can itself be overridden in each display due to its own analysis time/scaling settings.
The displays themselves can throw up errors which are sometimes cryptic - for example "IQ analysis error", which doesn't exactly give you a good idea why the analysis failed - was it because of incorrect acquisition/analysis settings, a lack of valid signal or a software fault. Help is offered, but it was not particularly useful for the most cryptic errors - it would be nice if a more descriptive error message was provided. Others are more straightforward, e.g. "Data acquired during ADC overrange".
On the whole, it seems that the processing load goes upwards rather quickly when multiple displays are enabled, and it is quite possible that your computer will struggle to keep up. Those working on smaller screens (resolution wise) may find multiple screens difficult to work with because of the tiled-nature of the screen and the loss of space due to the toolbars surrounding each analysis window - a large resolution screen may be a big asset for those looking to use multiple displays simultaneously.
Direct-to-Home television broadcast satellites are more than they appear. In many places, not only do they carry television to homes, they also carry contribution feeds, internet, and private data networking communications which use more specialized equipment.
In fact, the satellites can throw up a myriad of mysterious signals. One of them is the satellite beacon, used to send telemetry from the spacecraft to the ground station for monitoring of operational health parameters. Often, the information on these signals are not published, although Optus provides the frequencies. They can serve the purpose of power-compensation for rain-fade in a control loop as well.
Although it would be a stretch to demodulate the data, it would be nice to see the telemetry beacon itself and understand a little bit about how it might be modulated. As a result, I grabbed the IF frequency from the LNB from the loop-out port on my TBS6925 (which powered the LNB) and plugged that into the RSA306. As this was a receive only situation, I ignored the slight impedance mismatch (75 ohms into 50 ohms) as that would just normally result in incorrect power readings.
By doing some quick calculations, I determined the beacon's frequency in the IF, and tuned to it. I discovered a wide FM-style signal, which seems to carry two FSK signals encoded on sub-carriers. Nothing particularly "sophisticated", but it's robustness they're aiming for. The sub-carriers were identified in the Audio Analysis view. You can see it in the video as well. In fact, I did check the beacons of several satellites and it seems to have the same sort of basic idea.
This also gave me a chance to exercise the general purpose digital modulation options to draw constellation displays for a high-rate DVB-S carrier. By configuring the filters, symbol rate and type of constellation, an acceptable plot was very quickly displayed. Other parameters are also available in Signal Quality, and so are plots of EVM versus time, phase, magnitude, etc. depending on your display configuration. The stats on this signal aren't too bad, considering this is at the end of the receive chain after all the noise has "gotten in".
Of course, some of the signals on the satellites are not DVB-S/DVB-S2 style signals, which means that a satellite tuner card cannot lock onto them. However, it is possible for us to determine that there is signal energy there (http://goughlui.com/2013/10/07/review-tbs-tbs6925-pci-e-professional-satellite-tuner-card/). The DPX spectrum and peak-hold traces helps us out there, by allowing us to see what's going on over time, and collecting the "carriers".
There are different schemes, representing vendors of different hardware and proprietary standards. Adding markers allows us to gauge the spacing between the carriers.
Seeing as the RSA306 has many options geared towards Wi-Fi analysis, I had to test it out. Trying it with on-air signals was not the easiest way to begin, as signals are buried in noise and many of the measurements are geared for standards compliance tests aimed at devices. I suppose that is appropriate, as a design aid for new internet-of-things designers.
As a result, I decided to get a pristine signal, by hooking up a USB wireless card (brand unspecified, maximum power +18dBm by datasheet) directly to the RF port of the RSA306 (a pad/attenuator is probably best advised if you have one). The USB wireless card was driven by hostapd on a Linux virtual machine to act as an access point, using a 15ms beacon interval, thus putting out periodic beacon frames at 1Mbit/s which can be analyzed.
The constellation plot allows us to see the signal quality visually.
A wireless client was joined to the network, and data transfer (a large ping back and forth) was sent, which allowed us to observe 802.11g modulation modes of QPSK, 16QAM and 64QAM.
Other displays of phase, magnitude error can be engaged.
Wireless LAN power graphs and flatness also help in troubleshooting power-amplifier behaviour.
A summary of the data is provided as well in the WLAN Summary display which makes important signal parameters visible at a glance.
For those who want to get into the nitty-gritty, a symbol table feature allows you to see what each subcarrier is sending at any given time-step.
Of course, the regular DPX spectrum can easily resolve each of the sub-carriers - you can count all of them!
In all, the wireless LAN options provide a wealth of data about the operation of a wireless NIC/device and allow you to troubleshoot and verify design parameters. Mask testing is also available, but not pictured, as it seems to require 40Mhz acquisition bandwidth whereas the other WLAN displays require 20Mhz bandwidths.
While testing, I came across a Bluetooth 4.0 USB dongle, and I decided to plug it in and have a check of what it looks like while it is scanning, which is when I came across the fact that the dongle itself is not very well designed and instead spews significant energy outside the ISM band. Inadvertently, I had tracked down a device that could easily impinge upon the 2300Mhz LTE 4G carriers, and is seen to do so on the graph.
While I was at it, I was intending to start analyzing some on-air P25 as well when sitting at my desk at the university when I saw this very unusual signal.
By moving the antenna around, I was able to localize it to a particular monitor on my desk, and it only occurred when the monitor was displaying information, suggesting it was some noise leaking from the digital domain. Unfortunately, it mostly wiped-out the 403-430Mhz band where the P25 happens, so I had to move away from the office where there are nearly a hundred of these "noisy" gremlins to somewhere "quieter".
Also on the air locally are APCO P25 signals, used by the NSW Government Radio Network. This network currently runs Phase 1 C4FM signals in the 403-430Mhz band. I managed to tune one in, and show the eye diagram and constellation which distinctly shows 4-levels at each symbol interval despite the noise. The constellation also shows four "clusters" of dots indicating the four levels. Unfortunately, the majority of the other measurements are geared towards transmitter testing and didn't make that much sense looking at a weak on-air signal.
I was also able to tune into a POCSAG pager signal and get a look into the approximate frequency shift. A more accurate measurement could be made in the Spectrum view, where the RBW can be set to smaller values (e.g. 100Hz).
Of course, I also managed to check out local NDB's, AM and shortwave stations by hooking up my loop antenna to the RSA306, and the aviation ATIS amongst other signals, but I think that would unnecessarily lengthen the review, so I'll skip those for now. The frequency range does, however, cover most of the signals you are ever likely to encounter.
Avid Raspberry Pi users may be aware of a software called PiFm (http://www.icrobotics.co.uk/wiki/index.php/Turning_the_Raspberry_Pi_Into_an_FM_Transmitter), which manipulates the spread-spectrum clock generation hardware on the Raspberry Pi to perform the task of wide FM modulation. The modulation happens entirely in software, digitally, taking in a .wav file and turning it into an encoded clock/radio signal which appears at GPIO4. By attaching a short length of wire, it is possible to get a high quality, stereo, pre-emphasized FM signal to be broadcast across several rooms with no additional hardware requirement. This is something quite unique and not easily achieved at low cost.
Of course, being the home of Raspberry Pi, I'm sure users have experimented with this from time to time. I was curious to see what its performance was really like - listening to it on a radio proved to be satisfactory, but is it really a good thing to use all the time?
The answer could be found by hooking GPIO4 and ground directly to the RSA306. A pad or attenuator is strongly advised, although as the output of the Raspberry Pi is 3.3v peak to peak into 50 ohms, that corresponds to an RF level "around" 14-15dBm which is "sort of safe". Otherwise you can just look at the signal from the analyzer pulled off the air with an antenna.
I set the transmission frequency to 103.6Mhz, a free channel here in Sydney (as required by ACMA Low-Interference Potential Devices licensing conditions), and blasted out one of the latest tunes you'd hear on the air (sorry about that).
It was found that the deviation was a bit narrow, at ~25khz on wide FM which is below the ~100-125khz normally found, which is why the signal sounds a little quiet. The RSA306 doesn't seem to have any de-emphasis on the reception side, so the audio sounds a bit tinny.
The true nature of the beast is revealed when looking at a wider grab of the spectrum and it was what I expected. When dealing with "square" clock signals, harmonics are inevitably generated which contain a good portion of the signal energy. In this case, the harmonics are about +/-18Mhz with them being 14dB and 22dB down, which is not very far. Other sidebands are 42dB down, which seem to be due to the PWM modulation frequency. The interference spews outside of the FM band entirely, and the upper image sits very close to the aviation emergency frequency. That's not very nice.
When stopped, it can put out some buzzing which shows up quite clearly and is a wide pattern of spurs spaced about 6Mhz apart. Needless to say, I'm not going to continue putting this into the air - I'm a good RF citizen after all.
Lets also look at broadcast FM stations. As most broadcast FM stations don't send out test tones or "dead air", it's not exactly possible to analyze the quality of the signal. One figure that can be checked is the bandwidth of the station itself, which can be directly seen on the DPX spectrum (roughly, given the 3khz auto-RBW).
However, the Audio Spectrum display actually proves useful, especially when setting the bandwidths to a wide level to display the subcarriers in FM broadcasts which can be used to carry secondary audio services, RDS text and song titles, and other services.
Of course, you can also listen to the audio in mono with the demodulation panel, although it seems like de-emphasis is not implemented so the audio will sound slightly tinny.
I had two different models of UHF CB radio, so I decided to charge them up and test them out. It was possible to visualize the signal as well as demodulate it using the FM 13Khz option, which I used to do the introduction to the video and confirm that the radios worked. I did make use of the Channel Power and ACPR display to see how the two CB radios compared in terms of their adjacent channel emissions, which on the whole, were very similar.
Interesting glitches, upon key-up where bursts of carrier hopping about the band are also visible in DPX, although I didn't capture a screenshot of it. That might be due to the power-amplifier powering up while the PLL synthesizer hasn't stably locked yet. It was also possible to check the CTCSS sub-audible tone encoder deviation, as well as the voice-deviation, which measured 730Hz (almost spot on, 750Hz expected) and 3.8khz (a little narrow, expecting 5khz) respectively.
I also used the Audio Spectrum display and transmitted a 1khz tone. The display easily identified the signal, and the harmonic distortion/non-harmonic distortion frequencies and their levels. How neat!
The RSA306 is capable of streaming data to the PC for recording. With the recording tab in SignalVu-PC, you can capture the data in formatted (i.e. .r3f) or unformatted/raw (i.e. .r3a/.r3h) format. Initially, I was rather confused, as the captured files were not usable in SignalVu-PC (there was no open option for them), but reading the API manual made it clear that these were framed/un-framed 16-bit samples without any phase or magnitude corrections. The unformatted option was particularly attractive, as it produces a stream of 16-bit values, which does not require the receiving application to know anything about the framing or metadata formats described in the API manual. However, it is important to note that making measurements based on the streamed raw data may not give you the accurate results you expect.
One of my favourite utilities for examining software-defined radio data is GNU Radio. This is an open source framework based around a collection of blocks (many pre-built, although you can build your own as well) which you can use by constructing flow-graphs to build complex demodulators and analysis tools. I was very pleased to report that I could use the data in a flow-graph by using a file-source set to 'short' data type and a short-to-float conversion block. The plotted spectrogram clearly shows the broadcast FM stations which were captured in the .r3a file.
Note: I had inadvertently keyed in 114M as the sample rate rather than 112M, and was playing with the scale factor value - 8192 (for 14-bits) or 32768 (for all 16-bits) is the best value to use as far as I can tell. Smaller values for scale factor cause the apparent signal amplitude to rise.
SignalVu-PC's File Length option serves to split the file at the selected interval, with a certain number of these files captured based on the "max saved files per run" option. Because of the way the instrument is constructed, there is also no way to continuously capture data at a lower sampling rate than the 112MS/s that the ADC runs at, as no digital filtering/down-conversion seems to be implemented - with this large amount of data, it becomes difficult to store and perform real-time processing using third party tools, and it may only make sense to make short captures for analysis.
It does, however, show us the possibilities where up to 40-56Mhz of real spectrum is available in the analysis (although with roll-off towards the edges) allowing for the whole LF, MW, HF bands to be captured simultaneously - every single AM and shortwave station, NDB beacon, all at the same time, or alternatively, every single FM-station. It boggles the mind!
It would be valuable, if a source block for GNU Radio (i.e. a driver) for the RSA306 was provided for Linux, or some basic capture tools, which would further enhance the instrument's value as a data source, as the APIs presently only cover the Windows environment.
Corrected baseband data with a bandwidth of 40/2^n Mhz and sample rate of 56/2^n MS/s can also be saved from SignalVu-PC, but this is limited to an acquisition time of 1s which is sufficient not useful for long term signal monitoring. This can be exported from SignalVu-PC as a Matlab Level 5 document or CSV which is not directly usable with GNU Radio, but can be easier to handle in other Windows-based tools.
Unfortunately, we come to the part of the review where I must acknowledge some of the features which would be of interest to the community, but I haven't been able to explore. This could be because they are not measurements which I have the tools to use properly, or due to time and health constraints - but I will mention them here for completeness.
Pulse-measurement options are available and especially useful for radar and pulsed RF applications, although I don't have any within reach at this time. Mask-based testing is provided for ensuring compliance with standards requirements (pass or fail, capture on violation).
For those who are doing EMC compliance testing, the RSA306 has a real treat with its multi-zone spur search display which sweeps the relevant frequency ranges and has a list of pre-set limit values ready for loading. External gain/loss tables can be used, and the units of measurement can be changed to suit your probes and application.
A fully documented API interface which interacts with the RSA306 for making custom applications is available, which allows for accessing the RSA306 from C, C++, and Python programming environments. However, due to the high data rates involved, I can expect this to be somewhat challenging to achieve real-time performance. A MATLAB and Simulink interface is also available. TekVISA also supports remote instrument control through SCPI commands over Ethernet, although I have not been able to try any of this.
One option, that is not just a display within SignalVu-PC is the RSA-Map option. This enables a second application which takes data from SignalVu-PC and allows you to produce a direction-finding style map including signal levels, a manually chosen azimuth angle, and integrates GPS support (if you have such a unit connected). Maps in several formats (including images) are accepted. This is particularly useful when doing interference hunting, or plotting signal strength in key locations as part of a site survey. Unfortunately, because I haven't been too mobile due to ankle problems and I don't have a GPS unit handy, I haven't been able to test this out properly.
For those in the education sector, which is more restricted when it comes to budgets, there is a separate version of SignalVu-PC available known as SignalVu-PCEDU with all options enabled but will watermark all results with "Education Version" (or something similar). This particular version is sold at a significant discount for those who qualify, and pairing it up with an RSA306 really brings some serious analysis power for researchers and instructors in "tight" financial spots.
I had wanted to also use the RSA306 to help me in determining the sample rate limits and check the transmission characteristics of a Nuand BladeRF x40 with XB 200 transverter, but as I haven't gotten around to mastering the transmission configuration flow, I haven't been able to do this yet.
I would have to say that working with the Tektronix RSA306 and SignalVu-PC has been generally a positive experience. While the software itself is a little difficult to master at first, and sometimes feels like it's not intuitive, there is a wealth of helpful documentation available which will help you along. The software receives regular updates from Tektronix on a monthly basis or thereabouts, and seems to be stable with no crashing experienced during my time using it.
The features available, especially with all of the options enabled, is quite extensive and of practical use. They really showcase the possibilities which come about due to the software-defined nature of the instrument, and the ability to save and analyze data later (or with other tools) adds flexibility beyond that of the one unit and one software installation.
During usage, it was important to fully understand where the options are placed in the UI to have the best control over the analysis. Specifically, it wasn't clear initially that the Settings tab controls settings which are specific to which display is active at any given time.
I do really like the DPX spectrum display, but when you set the RBW to less than 1khz, it throws up an error of "RBW set to 1khz due to span" - regardless of my reduction in span, I cannot get it to go below 1khz even with maximising the acquisition samples, even though the regular spectrum display is happy to do so. I chalked this one up to a limitation with the DPX display mode, but the error seems to lead the user to believe that it should be possible to get to less than 1khz RBW with DPX.
Chasing signals around the bands is quite easily done with the centre-frequency pan option, which allows you to click and drag. However, depending on your acquisition settings, the re-tuning of the frequency can take up to several seconds as SignalVu-PC finishes the current acquisition and analysis before moving on. This can be a little frustrating when chasing some transient on-air signals.
Another thing of note is that any changes to zoom level, or inadvertent pan, also seem to clear any peak-hold traces which I didn't expect, and can lead to the loss of some collected data if you're not careful.
While 40Mhz might seem like a decent amount of spectrum, and it definitely is when dealing with narrowband signals, it is not sufficient for real-time coverage of the whole 2.4Ghz ISM band (for example) or analyzing 80Mhz/160Mhz 802.11ac or wider modes. It is possible to use the spectrum analyzer like a "swept" unit and have it "hop" between frequencies taking 40Mhz wide "snapshots".
The spurious response and dynamic range of the RSA306 seems to be a point of contention amongst potential purchasers. On the whole, to expect high-end real-time benchtop-analyzer grade performance from a lower-cost portable unit like this could be considered unfair. However, even with the limitations, it is very suitable for many applications, especially if some care is used in setting the reference level and in interpreting the results. I didn't feel that it was limiting throughout my experiments with the instrument.
The SignalVu-PC options, while powerful, are definitely worth evaluating before you purchase. Some options, such as the "flexible" OFDM one, seem to be very much geared towards 802.11 and 802.16 usage, and do not offer pre-set settings which are useful for other sorts of signals. I had intended to also look at DVB-T and DAB signals with the instrument, but I didn't have enough time to get around it.
Power consumption was something that I couldn't directly measure, although the fact that the device draws power directly from USB can be considered both a blessing and a curse. On the one hand, it obviates the need for any power bricks or battery packs, it also means that the host machine's battery runs down noticeably faster during field usage. This is especially apparent given the high computational loading the software places on the host. As a result, I was seeing about a halving of run-time on the Lenovo E431 (from four hours doing word processing writing this review to about two hours doing live-link analysis). Another possibility is for RF noise to be introduced into the receiver by a noisy USB power supply - although it does seem that they've done a good job of conditioning the power from the PC.
The USB connection also carries with it a little annoyance when it comes to the reliability of the connection. While it is evident that Tektronix has opted for quality connectors and cable, I did experience two events where the connection to the RSA306 was lost during analysis due to the cable being moved slightly within its connector. This can make it a little frustrating if you're trying to carry a laptop and walk around with the RSA306 in a sling-bag, although that is pretty difficult to juggle. The RSA306 also consumes a large amount of bandwidth on the USB 3.0 bus, so it's not advisable to try streaming captures to another USB 3.0 device, or even using it with other USB 3.0 devices if at all possible to ensure there is no impact on performance.
With the increase in MIMO signals which need phase-coherent capture for proper analysis, I am not sure if this unit is a good choice for such tasks. While there is an external frequency reference and external trigger, I'm not sure if a USB 3.0 bus could handle three or four of these units, and I'm not sure if the units themselves will capture phase coherently especially when the latency of USB 3.0 is factored in. Those experimenting with MIMO might have to look at higher end multi-channel analyzers instead.
It might sound that I'm focusing on the negative, but these small niggles are a part of every product and are often the sort of thing that users discover after using a product for a while. Of course, none of these are particularly deal-breakers on their own, and on the whole, Tektronix has delivered a product which is pretty much on the edge of what is technologically feasible given our present computing and connectivity resources at a very reasonable price. In fact, it's on the edge of what is capable so much to the point that the streaming record data files it produces are a bit too big for comfort, and the higher-bandwidth oscilloscope-based signal analysis seems to be based on "short" high bandwidth captures sent for analysis and is not "streaming" as this device does.
In my opinion, Tektronix have delivered a rather unique software-defined instrument, arguably the first in its category. The RSA306 offers a wide frequency range, with a suitably wide power-range, in a form factor which is highly portable, at a price which is compelling and good value-for-money. It is ruggedly built, compact in size, easy to set-up and does not require bulky external power supplies.
The RSA306 leverages the SignalVu-PC software for analysis, which is the same as the tools available on/for use with more expensive MSO/MDO/RSA series devices, which can make future upgrades much more compelling should you purchase options or invest time in developing workflows with the software. The software is stable, frequently-updated, and powerful and features an array of extensive options which can be purchased for your application-specific needs.
Options can be freely evaluated for a 30-day period to determine their suitability for your needs. The base edition of SignalVu-PC provides sufficient useful measurements to get started with basic measurements. The capture ability allows for instrument time to be spent acquiring signals with analysis performed later - or for permanent capture of signals for records.
Because of this, the RSA306 will be attractive for many applications, including 802.11 wireless device development, Bluetooth and Bluetooth Smart/Low-Energy device compliance testing, EMC testing, interference hunting, coverage tests, transmitter tests and radar characterization. Furthermore, to make the unit even more attractive to educators, an education edition of their software is available as well at a very good price.
The unit itself offers bleeding-edge capabilities in terms of real-time USB 3.0 streaming capture and analysis with a 112MS/s 14-bit ADC throwing out 224MiB/s of data and thus requires a beefy computer to ensure maximum performance. Because of the software-defined nature of the instrument, in future, further options may be offered by Tektronix to expand the capabilities of the unit. If not, you are still free to develop your own applications with the provided APIs, or skip that entirely and analyze the raw/formatted/corrected data from SignalVu-PC.
Because of the size, price and design, it is not without some compromises. To expect identical performance to larger, more expensive bench-top units is unfair. That being said, in my experience, the compromises in the linearity, spurious responses and dynamic range to some extent were not really relevant to my usage and with careful, proper use of the instrument, it was more capable than required for most applications and its limitations were not apparent. I have a feeling that, as specifications are easily compared, some users have a habit of comparing them without regard for what they actually require.
Hobby-grade SDRs will continue to rise in popularity, and be available at even more lucrative price-points with more sophisticated specifications, but it must be remembered that the requirements for instruments in terms of their performance is much more stringent. Hobbyist-grade SDRs generally have no/limited guaranteed specifications. Specifically, IF images, spurious response, resolution/dynamic-range and linearity all can be quite problematic as many of these SDRs repurpose application-specific front-ends and ADCs and operate them at frequencies which they were not specifically designed. No calibration is provided in general, and they may not be built robustly. Likewise, measurement and analysis options with those are limited, and can involve time spent in producing flowgraphs or applications to perform the measurements you need. It's important to remember that the Tektronix RSA306 is more of a refined "package", ready to go out-of-the-box. In light of this, the RSA306's price is actually quite attractive.
It is a very likeable unit, even for those who may already own an RSA benchtop analyzer, as it is very convenient for field work. However, those who really need multiple-channel analysis, better linearity and spurious response or wider frequency ranges will probably need to spend significantly more to get what they need.
This was a rather long review, sorry about that! You can definitely see that I had a lot of fun with the device - I hope you did get something out of it though. That being said, I will continue to use the RSA306 to hunt down and analyze more signals. When the time permits, I may post up additional content on my blog (here at element14 and at http://goughlui.com). However, I do have another RoadTest to get on with and more PhD research tasks, so we'll see when that might be.
Sorry for the slight fuzziness in some of the screen captures - they were actually frame-grabs from the HDMI capture card I used to capture footage and audio of my experiments so I didn't have to think about interrupting the experiment to take a screenshot.
Once again, thanks to Tektronix and element14 for their generosity in providing the unit for review. Now ... I wonder how one qualifies for SignalVu-PCEDU ... *ponders*.
Hi Gough Lui,
Excellent review! (as expected ; ) - I've only started absorbing some of it, but it looks great.
I noticed you tried pifm to show how bad it was (as expected!). I was surprised to see some…
Hi Gough, I read your review and watched the video. It was a beautiful presentation and I enjoyed it even though 90% was above my understanding level. The amount of time you have spent and the detail you…
As usual your review is very detailed but covers what we don't see in the documentation and information.
Well done, and hopefully you get the software upgrade you deserve....
Very well done!
I've been looking closely at purchasing for my home workbench. Insightful and enough pictures/gifs to give a good idea of what to expect visually, along with a very well presented analysis.
Thank you for a well done road test.
Just in case someone was wondering why I couldn't get around to testing out RSAmap, one of the things I really needed was more run-time to be able to make sure I could get enough measurements on-foot, and a GPS unit as well. Time was ticking on the 30-day window, and it is pretty much run out now.
To extend the run-time, I tried to purchase a clone laptop battery from a local seller, which used a courier I didn't like (despite the listing saying that they didn't), who claimed the package was delivered (when it wasn't), who took longer than expected to deliver, which shipped me a battery that doesn't work due to Lenovo's draconian DRM (they also have wireless-LAN and wireless-WAN whitelisting too, so you can't upgrade those cards either). For some humour, here's the article: Fails: eBay Laptop Battery Seller & Fastway Couriers
I think it's right to say that I'm as frustrated as you probably are, but needless to say, I tried my best. I really did.
Thanks for the comment - always nice to hear from you shabaz!
Gee, I don't know what guys are thinking hacking around with pifm - I think they really got far in implementing proper stereo and pre-emphasis on the transmit side, and improving the resolution by a "dithering" process using the PWM - my first "stereo" transmitter kit using the BA1404 sounded nothing like the quality that pifm puts out, but whenever dealing with pure digital, square waves are the enemy. The deviation itself seems to be limited by the steps offered, and it's not like it's going to put out anything but FM. It would take a lot of filter-work to get rid of all the harmonics, and by then, maybe it's just better to use a quality high-rate ADC and some FPGA magic and go the whole hog SDR. I suppose it's a shame that so few actually understand the limitations when everything seems to "just work".
As I did say, the unit itself was very likeable, but it is also on the edge of what is computationally feasible and reasonable for today. Users with older PCs who may have installed PCI-E to USB 3.0 cards might actually suffer degraded performance or problems outright - PCI-E x1 only offers 250MB/s to the host and would struggle (overheads included) to keep up. Other SDR users have noted that not all USB 3.0 chipsets perform equally, and while Intel seems to be most consistent, I did try it on an older Renesas/NEC without any obvious issues. I wouldn't be able to comment on users of Fresco Logic, VIA or Asmedia chipsets, however.
Another thing I would like is if they packed a bit more power into the FPGA to offer on-board digital filtering and decimation, so you can actually capture streaming "raw-ish" data at lower rates, even if it's limited to powers-of-two. The old Winradio receiver does this, as it samples 50Mhz and only puts out a 2Mhz "tunable" window via direct digital-down-conversion because USB 2.0 links couldn't cope with much more back in 2010. If they did provide this, then that could make long-records much more feasible. That being said, it seems that they have done a lot of work optimizing their API and SignalVu-PC software to be able to handle the torrent of data coming from the RSA306 which is not easy, as GNU Radio on a modern i7-4770k desktop fails to do a spectrum plot of the captured 112MS/s file at real-time (close, but no dice).
That's my thoughts anyhow, I'm sure in several years that this won't be much of an issue - maybe we can get 100Mhz real-time spectrum when USB 3.1 becomes more solid and CPUs become even better.
Not that I understand much of it, but very thorough and detailed review, great work!
Two thumbs up!