Agilent N9322C Spectrum Analyzer & TI CC11XLDK-868 - Review

Table of contents

RoadTest: Agilent N9322C Spectrum Analyzer & TI CC11XLDK-868

Author: Instructorman

Creation date:

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?: Tektronix MDO4104-3 mixed domain oscilloscope with 3 GHz spectrum analyzer

What were the biggest problems encountered?: There were some firmware issues that caused occasional lock up on parts of the instrument. Keysight was very responsive and helpful in solving these problems.

Detailed Review:

A User Experience Perspective on the Keysight N9322C and TI TRXEB


Engineers at Keysight have packed many useful measurement capabilities and many powerful analytical features into the N9322C Basic Spectrum Analyzer.  Keysight refers to the N9322C as a “basic” spectrum analyzer, and perhaps it is “basic” when compared to the über-sophisticated top end X-series or Keysight’s Real Time Spectrum Analyzers.  However, the list of what it cannot do is shorter than the list of what it can do, so I will not be referring to the N9322C as a “basic” spectrum analyzer in this review.

Of the three Road Testers selected by Agilent to test this awesome instrument, I probably have the least RF experience.  However, I have 30 years of polytechnic teaching, course development, and applied research experience.  I have used several generations of test equipment produced by HP, then Agilent, and now Keysight.  I have taught instrument use to over 3000 students and used a range of test equipment in many applied research projects.  Cosmin and Shabaz have been posting excellent blogs about the many capabilities of the N9322C and RF concepts in general in the Element14 Test & Measurement group. I heartily recommend reading each of their blogs; they are informative, well written, and entertaining.

My review will concentrate on my experience as a user of the Keysight N9322C and the TI TRxEB kit.  I will discuss use of the N9322C as a teaching tool, as an applied research tool, and as a hobbyist tool.  My experience so far has included extended interaction with Keysight technical support and I am delighted to be able to share that experience in this review.  I will also compare the N9322C to the spectrum analyzer in the Tektronix MDO4104-3 mixed domain oscilloscope.


N9322C user experience- as a teaching tool

Imagine having a full lab complement of N9322C spectrum analyzers to teach RF principles to electronics technology students.  Modern, reliable, accurate, full featured, networkable, and professional grade – wow, an instructors dream come true.  That is until you turn on the N9322C and hear the fan.  My, that is a loud fan – and it does not change RPM with temperature. Nope, it just roars away at the same speed and volume all the time.  Now imagine spending two hours in a laboratory with two dozen of these things humming away in unison. Yikes!   Kidding aside, the fan on this instrument is noticeably louder than the fan on any other instrument on my test bench.  A room full of these operating at the same time would pose a practical problem to verbal communication and occupant weariness in a teaching environment.


Let us put aside the noise from the fan for a moment and consider other aspects of the N9322C in a teaching context.  When considering test equipment for a teaching laboratory there are many factors to consider including, of course, the capital investment required, the durability of the instrument, popularity of the brand in the real world, ease of use, upgradeability, and after sale support (in and out of warranty).  Here follow my findings on each of these factors.


Capital investment

As a capital investment, spectrum analyzers are always a troublesome bit of kit.  Good ones are very expensive and cheap ones are usually not very good.   As tallied in the table below, a base N9322C beefed up with the options I would like to see in a teaching environment prices out at nearly $27,000.


Item or option

Canadian price on Keysight web page

Base N9322C

$12,353

AMA – AM/FM modulation analysis

$717

BB1 – Baseband input

$2688

ASK/FSK modulation analysis

$920

MNT –spectrum monitor w/record & playback

$1792

P07 - 7 GHz preamp

$1003

RM7 – Reflection measurement

$2688

SCN – Channel scanner

$1792

TG7 – tracking generator

$2688

Total

$26641

 

Even with educational discounts, it will not take more than a few N9322C spectrum analyzers to decimate a typical schools capital budget.  Keep in mind; you get what you pay for and you do get Keysight engineering quality and support services.  So maybe you convince your department chair to buy one a year for a few years then rotate students through a few labs on RF spectrum principles.  That could work.


Durability

In terms of physical durability and build quality, I have no concerns.  Like all Agilent instruments I have worked with, the N9322C is rock solid and sturdy.  The only front panel element that might suffer under continued student use is the display screen.  Students poke pens, pencils or fingers into display screens, perhaps in the absent-minded hope that tactile connection with a waveform will reveal its meaning.  A mentioned elsewhere, by Cosmin or Shabaz I think, the screen on the N9322C is quite reflective, unlike the duller satin finishes found on many modern instrument screens. This is not really a problem, but I suspect fingerprints will be more noticeable on such a reflective surface.


The buttons on the N9322C are larger than the buttons on the bench top series of Keysight instruments (like the 34461A digital multimeter or the 33622A waveform generator).  Buttons on the N9322C have an ergonomic curved surface, which I feel gives a better visual presentation of text and offers the user a better activation sensation.  See the photo below to see what I mean.

I also prefer the plastic used in the rotary knob on this instrument compared with the material used on other Keysight rotary knobs (33622A for example)  The relative level of use of a Keysight (Agilent) instrument can be estimated by glancing at its rotary knob.  Skin oils seem to adhere to the soft rubbery plastic used on many Agilent knobs and they quickly take on a grimy dull brownish green patina with frequent use. The harder knurled rotary knob on the N9322C stays cleaner longer.


Popularity

There is no reason to worry about the popularity of Agilent, now Keysight, instruments in the real working world into which we release our graduates.  Grads will probably encounter Agilent test equipment at some point in their career journeys.  Ease of use, on the other hand, is harder to dismiss as a non-issue.


Ease of Use

As mentioned previously, the many measurement and analysis capabilities available in the N9322C make calling it “basic” feel awkward.  Great capability can lead to great operating complexity. To illustrate what I mean, let me walk you through one particular measurement scenario on the N9322C.  The N9322C, with an antenna and appropriate options installed, is capable of capturing, measuring and decoding the data in FSK data packets.  As part of a hands on laboratory experience in using RF test gear, I would like to have students set up the N9322C to perform FSK decode.  Before asking students to perform a measurement, I do a step-by-step walk through myself to see what configuration steps are required to achieve the learning outcome.  Here is what I found:


I used TI’s SmartRF Studio 7 to configure a TRXEB with a CC110LDK radio module to transmit GFSK packets continuously with the following parameters:

Parameter

Value

Carrier Frequency

  1. 867.999939 MHz

Data Rate

  1. 9.9926 Kbaud

Deviation

  1. 19.042969 kHz

Packet Size

11 bytes (length<1>, count<2>,payload<8>)

Packet Interval

500 ms

TX power

0 dBm

Packet data payload

01 23 45 67 89 AB CD EF

Sync Word Length

No preamble/sync symbols

 

Setting these parameters up in TI’s SmartRF Studio 7 was a breeze, by the way.


With the TRxEB broadcasting a known payload of GFSK packets every 500 ms, I set about configuring the N9322C to take advantage of its cool FSK decode capability to see if I could get it to deconstruct radiated RF energy and show me what was in those invisible packets.  I am happy to say the N9322C is perfectly capable of achieving the decoding task as described.  However, the configuration steps are numerous.  With the included adjustable antenna attached to the RF input, and the N9322C in power up spectrum analyzer mode, I pressed Auto Tune.  The instrument quickly narrowed from a full sweep of its 7 GHz range down to a group of the strongest local FM radio stations as seen below.

Sixteen keystrokes later, I was looking at the spectrum of the GFSK bursts as shown below.  I had to make small adjustments to the carrier frequency to find the center as measured by the N9322C.

After an additional 36 keystrokes, I arrived at the decoded display of hex data in the packet, shown below.

From left to right the decoded hex values represent the packet length (0A16 = 1010), followed by a sequence count (happens to be 25D816 = 968810), followed by the payload bytes I configured in TI’s SmartRF Studio (01 23 45 67 89 AB CD EF). 


The number of keystrokes needed to set up this measurement might be reduced through careful optimization, but this exercise nevertheless requires a lot of careful configuration and instrument familiarity to carry out successfully.  This, I believe, is still a useful teaching exercise, and is worth doing, but it would be preceded by weeks of simpler instrument familiarity exercises.


More exploration of FSK decode on both the N9322C and the MDO4104-3 can be found in my blog post here.


Counting keystrokes: The MDO4104-3 v. the N9322C

Occupied bandwidth can be measured by the Keysight N9322C and by the Tektronix MDO4104-3.  Here is a comparison of the number of set-up steps required by each instrument and screen captures of the resulting measurements. To set this measurement up, an antenna was attached to each instruments RF input.  SmartRF Studio was used to configure the TRxEB to broadcast continuous random GFSK packets.

Both instruments were set up to a common starting point where they were acquiring a spectrum across their complete frequency range (0 to 7 GHz for the N9322C and 0 to 3 GHz for the MDO4104-3).  From the common starting point, here are the steps I took to get each instrument to measure occupied bandwidth.


Tektronix MDO4104-3

Keysight N9322C

Freq/Span

Center Freq

8,6,7,.,9,8,3,3,3,9,Mhz

8,6,7,.,9,8,3,3,3,9,Mhz

Span

Span

1,0,0,KHz

1,1,0,kHz (different because Tek adds 10% to span)

BW, Auto

BW,RBW,1,0,0,Hz

Measure, Select Measurement

Measure

Occupied Bandwidth, Menu (to clear screen)

Occupied Bandwidth

 

 

 

 

 






Setting up the Tektronix took 23 keystrokes.  The result is shown below.

Setting up the Keysight took 25 keystrokes because the RBW had to be adjusted to match the RBW on the Tektronix.  The result is shown below.

Both instruments certainly get the job done.  The difference in keystrokes is trivial.  What the electronic screen captures above do not show is that the screen on the MDO4104-3 is 10.5" diagonally, whereas the screen on the Keysight is 6.75” diagonally. The bigger screen on the Tektronix can reduce clutter and allows for larger fonts.


Screen clutter

Some of the layout decisions made by the GUI team at Keysight in charge of the N9322C are questionable in my opinion.  Take, for example, the choice to display the entry of a new center frequency as a text overlay on top of the grid for the spectrum display.  This awkward positioning makes reading the new center frequency very difficult.  See the photo below.

Switching off the graticule does not improve readability much.  Moving the trace by adjusting the Ref Level will finally reveal the Center Frequency text, but why would you want to make all those adjustments just so you can see what you are entering for a new value? 

The short video clip below shows the user experience view of changing span from 100 kHz to 200 kHz. When Span is pressed, a box on the screen turns light blue to highlight the current span setting.  As soon as the first key of the new span setting is pressed, the soft menu choices on the screen turn to GHz, MHz, kHz, and Hz. The values the user enters appear in the lower left corner of the screen, on top of the graticule, on top of the vertical scale markings and quite probably on top of a jagged, rapidly updating spectrum trace.  With a little redesign of the GUI it would be possible to show the new value being entered in the highlighted box near the soft menu keys and list the units choices below on some of the unused soft keys.

 

Upgradeability

There are plenty of upgrade options on the N9322C.  It can be purchased as a bare bones spectrum analyzer, then, as budgets permit, optional features can be enabled with the purchase of licenses.  A nice model, but, as mentioned earlier, the N9322C really becomes useful I think only when most of the options are enabled.


After sale service

On this aspect, I have first hand experience, and I can happily report the experience has been excellent so far.  Within a day or so of receiving the N9322C I began to encounter occasional lock ups where the instrument stopped updating the screen.  The front panel controls continued to function, but it appeared as though the memory buffers were not being refreshed, causing the display to freeze on the last swept dataset.  After some expected navigation of the Keysight website, I was able to issue a support request. The next day I received e-mail from a Keysight Application Engineer with a promise ot look into the issue.  The following day I was asked to provide more detail, including capturing and uploading a state file from the instrument.  Once I provided the requested information, the application engineer was able to reproduce a similar lockup on his reference instrument. 

Four days after confirming the lock up was instrument related and not user related (hey, even the most experienced user can make mistakes), I was provided with a beta firmware upgrade file. I loaded the beta firmware and repeated the steps that caused lock ups previously.  With improved firmware the N9322C behaved much better, but it still locks up from time to time.  My interaction continues with Keysight to solve this problem.

As a bonus, the beta firmware added some extra features to the FSK decode menu.  The beta firmware also marks the transition from the Agilent brand to the Keysight brand (see below). I suspect the beta firmware was in development anyway, to at the very least, complete the brand transition, and I happened to come along with a problem that happened to be on the fix list. Nevertheless, the response time, level, and quality of service I’m experiencing in this interaction with Keysight is refreshing.  Great customer service builds brand loyalty.

N9322C user experience- as an applied research tool

Many characteristics make the N9322C useful in certain applied research environments.  If the work does not require the analytical power of a full vector signal analyzer and the signals under consideration are no greater in frequency than 7 GHz, then the N9322C has much to offer.  For embedded controller prototype development that incorporates RF modules operating on common protocols like Bluetooth, WiFi, and GSM cellular radio, the N9322C provides a good tool kit of RF measurement and analysis capabilities. 

The next two screen captures below show the spectrum of an SMS text message exchange occurring between my cell phone and a prototype instrument I developed that incorporates a commercial-off-the-shelf (COTS) GSM modem with custom microcontroller hardware and sensors. The top image is from the N9322C, the bottom image is from the MDO4104-3.  The prototype was placed equidistant between the antennas on each instrument in an attempt to balance the signal strength impressed on each antenna. The reference levels and attenuation setting on the N9322C caused the apparent difference in signal amplitude.

I ran this test several times in order to get an SMS exchange that occurred within the spectrum I had the instruments observing.  On this particular exchange, the cell phone and the prototype used the same channel to send data.  In other runs, I saw the cell phone send on one channel and the prototype respond on the adjacent channel.

The image above illustrates a few of the helpful features of the N9322C. There are two traces captured here; the yellow trace is a live update of the swept portion of the spectrum, the green trace is set to display Max Hold. Because SMS transmissions are brief, they appear and disappear quickly on a swept trace display. Enabling a second trace that retains the maximum signal signature permits a more leisurely analysis of fleeting signals.  Two markers, labeled 1 and 1R on the Max Hold trace, roughly show the corners of the SMS spectrum. The text in the PC image captured by Keysight’s HSA and BSA software is garbled in the table below the spectrum traces:  what shows up as ΔM1 on the instrument screen appears as i÷M1 in the PC image.  The markers in the image above are easy to see, but in other cases, they can be buried in the signal trace. Contrast the markers on the N9322C with the cursor approach used on the MDO4104-3 below.  I prefer the cursor approach because of its clarity.

AVI movie making

A nice feature available in the Keysight HSA and BSA software is the ability to make AVI movies of the N9322C screen at up to 15 frames per second.  The HSA and BSA PC software captured the clip below.  The video clip shows an SMS message containing a query command transmitting from my cell phone through the cellular network to the prototype followed a short while later by the prototype responding to the query with a response SMS on a different channel.  Note that the close proximity of the cell phone and prototype to the instrument antenna resulted in a final IF stage overload (warning message can be seen at bottom of the screen).




N9322C user experience- as a hobbyist tool

To conclude my review of the N9322C, a few words about the N9322C as a hobbyist tool.  Well the obvious first observation is that not many hobbyists can justify the cost of this instrument, even in its basic configuration.  However, if you have the money, you should consider other competitive instruments before make a purchase decision.  I hope to have shown that the Tektronix MDO4104-3 provides much of the functionality found in the N9322C.  Both instruments can measure channel power, adjacent channel power ratio, and occupied bandwidth.  Both offer spectrogram displays.  The MDO4104 lacks a tracking generator, does not provide modulation decode, and it has a less feature rich PC application. However, the MDO4104-3 does provide integrated frequency domain and time domain signal analysis.  The screen on the Tektronix instrument is larger and less cluttered.  Of course, there are many other competitors in the market and it would be wise to compare specs, reviews, costs, and support from each vendor before making an expensive investment in a spectrum analyzer. 

The choice will depend on which capabilities are more important to you based on the type of hobby work you do.  A spectrum analyzer is a great tool to increase understanding of electronic signals as it offers a distinct perspective of signal behavior compared with the perspective offered by oscilloscopes working in the time domain.

Both Tektronix and Keysight offer extensive support literature on their respective websites. Keysight has a more extensive selection of white papers, tutorials and application notes specific to spectrum analyzers and RF testing.  Explore the spectrum analyzer resources from Tektronix here, and the resources from Keysight here.


TI TRXEB user experience

Included with the Keysight N9322C road test was a Texas Instruments CC11XL development kit.  I was eager to use the C11XL development kit as I have been curious about working with RF modules as add-ons to embedded systems. The CC11XL was very easy to set up and use, and in combination with SmartRF Studio, it provided a nice suite of useful development features.

In a matter of minutes after opening the box, I had the two TRxEB boards communicating.  I was able to quickly assess the range of communication by walking away from the transmitter while holding the receiver as it ran the built in packet error rate test.

The battery life was not spectacular on the boards I received.  After only a few hours of use, the AA batteries provided with the kit were exhausted. This could be related to the quality of the batteries, or it could be related to the power consumption of the boards. I have not investigated this enough to answer the question.

My experience with the CC11XL development kit and SmartRF Studio has been enjoyable and I will continue to work with these tools to develop a specific application introduced in a blog I posted here.


Thank you

To conclude I want to draw attention to and give thanks to the many people and businesses that make the Element14 Road Test program highly successful and amazingly worthwhile.  Dr. Defeo at Element14 does a great job of managing the continuous stream of desirable and useful road test products and provides just enough feedback to calm the anxious nerves of road testers waiting for selection decisions and shipping notices.

Without the generous and liberal support of vendors, Road Test would be a theoretical exercise.  I continue to be amazed by the quantity, value, and variety of products offered on Road Test.  Sincere and deep thanks go to Keysight for providing the opportunity to road test the N9322C with a rich set of included options.  I would also like to acknowledge the generous support of the Road Test program provided by Texas Instruments.  There have been many TI development and evaluation kits offered on Road Test this year.  I think everyone would be happy to see TI continue this level of support.

Thanks, gratitude, and respect to my fellow Element14 community members who take the time to delve deep into the minutiae of every product offered for testing, and then take the time to compose and construct detailed reviews with photographs, videos and well-written text.  It takes a lot of time to put together a comprehensive review and I think we all feel obliged to do so considering the great fortune of being selected to receive such cool gear for free!

Anonymous
  • Thanks Cosmin.  Keysight provided me with a beta MCU firmware update to version A.05.11.  The current version on the Keysight web site is A.04.47.

     

    The beta version I was given to evaluate is more stable than A.04.47, and it has some nice new features, like hex decode of FSK packets.

    I was asked to not distribute the beta firmware.  If you add me as a contact, I can provide you the name of the application engineer at Keysight that I worked with; perhaps you can contact him and request a copy of the beta firmware.

     

    Mark

  • Very nice and informative review Mark; I enjoyed reading it.  You mentioned about firmware as cause of locking, so have you updated the firmware to a different version than the one that came with the instrument?  What version have you updated it to?

    Best Wishes,

    Cosmin

  • There are three photographs of the N9322C instrument screen in my review:

    1. The close up of the decoded hex contents from the FSK packets (because the beta firmware Keysight gave me is one step ahead of the PC software and provides hex decode but the PC software only allows for binary decode)
    2. The close up of the screen clutter where new center frequencies are entered.
    3. The close up of the logo differences on the escutcheon and screen with the beta firmware update to Keysight brand.

     

    All other images from the N9322C and the MDO4104-3 were obtianed via vendor supplied PC software using a USB data path.

     

    I like the performance and features available on the N9322C, but honestly, the GUI is disappointing.  The layout seems haphazard and it feels like more than one designer was working on it as there are inconsitencies in where information appears and how information is entered.  The screen, I think, is large enough to handle the job (barely), it just needs to be managed better.

     

    Aquisition rate on the N9322C is generally better than the MDO4104 - but it is direclty related to span and RBW settings, so a wide span with a super narrow RBW will slow things down significantly.  After posting my review I took some time to observe the two instruments as they simultaneously analyzed slices of the cell phone spectrum and WiFi traffic.  Neither instrument seemed to intercept all the traffic.  One would catch a burst that was missed by the other for example, but with peak capture turned on they would both develop comparable envelopes over time.

     

    I think I will post the results of an experiment I did on the WiFi spectrum with the two analyzers in spectrogram mode.  The spectrograms are quite different even though the two instruments were running simultaneously in the same space.  The N9322C seemed to be much better at detecting frequency hops.  The MDO4104 seemed to smear all the hops in a burst into one wide spectrum.

  • Nice read.

     

    Thanks for the commentary on the interface. I was suspecting it was sub-par based on the other reviews. Expensive test equipment can be a delight or a frustration depending on the user interface.

     

    Are the images in your review mostly from the web interface? I had noticed the layout of the N9322C display seemed quite unpolished, which is partly a function of the lower screen size and resolution, but perhaps also a sign of immature firmware.

     

    Also, how does the acquisition rate of the N9322C compare with the MDO4014?