Evaluation Type: Test Equipment
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?: Although not a directly comparable oscilloscope, I had a Tektronix MDO4000 series available, so I used it for comparison purposes when running tests on the Keysight MDO3034T
What were the biggest problems encountered?: The capacitive touch screen didn't behave as well as I had hoped. There were several issues with awkwardly places screen icons and it took me a while to learn how to get swipe gestures to work well. Certain system operations were unfamiliar (zone triggering, for example) but were easy to learn with usage. As problems go, there were really no big problems at all.
How I approached this Road Test
There are several conditions that candidate applicants have to accept before being selected to perform a Road Test. One of those conditions is confirming ability to complete the Road Test within 60 days of receipt of the product. With 60 days to investigate the myriad capabilities of the MSOX3034T I managed to evaluate many core features, but I ran out of time to explore every feature packed into this powerful and sophisticated tool. This thing is like a tool box with hidden drawers packed with unexpected treasures. It will take many months of use to fully appreciate and evaluate all this instrument has to offer. However, during the 60 days available for the review I put the MSOX3034T to use in several measurement challenges and have gathered my findings here.
Some of the measurement capabilities in the MSOX3034T require specialized differential voltage and/or current probes that I do not own. Other capabilities are specific to applications that are outside my current sphere of activity, but it is reassuring to know, for example, that if I ever had a reason to do some CAN bus or NFC testing, the MSOX3034T handles both protocols (and many more).
My approach for this review was to report on my experience using the MSOX3034T as a bench tool working on my current measurement challenges. Where possible I tried to push the 'scope to the limits of its capability to help me find its boundaries. Also, where possible, I compared operation and performance of the MSOX3034T against my regular bench oscilloscope, a Tektronix MDO4104-3. Any tool, for example a hacksaw, a French press, or a spectrum analyzer, can be evaluated as an aid to interaction with the physical world. The efficiency with which a hacksaw cuts can be evaluated. The filter system on a French press can be assessed. The user interface on a spectrum analyzer can be critiqued. I am interested in evaluating how well a vendor's test equipment allows the user to interact with electronic systems. Ideal test equipment, in my opinion, behaves something like a concierge. Whereas a hotel concierge helps a guest book a restaurant reservation or find tickets to a play, the oscilloscope as concierge helps the user extract knowledge from electronic systems. If a user wants to know the frequency and amplitude of a certain signal, the oscilloscope, as concierge, knows how to obtain that information for the user. There is always some back and forth between guest and concierge to clarify and bound specific requests, but in the end, the concierge knows how to get things done. An ideal oscilloscope should be able to get things done with only a little back and forth with the user. Much of my energy in assembling this review was devoted to gathering details about the worthiness of the Keysight MSOX3034T as a signal measurement concierge.
As mentioned, I put the MSOX3034T to use in real world measurement use cases. Several of the use cases came from the work I am doing to upgrade an off-grid solar PV electrical system in a garden shed. In addition to upgrading the PV system I am designing landscape lighting and an automated watering system. I am also installing a weather station and security camera. Many of these upgrades and designs generate measurement use cases that require an oscilloscope. I was able to detail my experience using the MSOX3034T in seven use cases before the review submission deadline cut short my fun. Six of the use cases have been posted as blogs. Links are provided below. I recommend exploring the blog posts as they dive into several use cases in some detail exposing features and limitations of both oscilloscopes. The seventh use case is embedded in the body of this review along with general musing about the user interface and instrument behavior.
Basic facts (and some of my opinions) about the Keysight MSOX3034T
The defining specifications for this instrument situates it as a mid-range bench instrument well suited to applications in system engineering, instrument education, fault troubleshooting, and system test. It is equipped with four 350 MHz analog channels, 16 digital channels, segmented acquisition memory, and a USB interface. The Road Test model came with a LAN/VGA adapter. There are many optional features that can be enabled after purchase including various protocol decoders and standard test procedures. Additional functionality like wave generation and bandwidth extension can be enabled through purchase of license codes. The Road Test instrument I evaluated for this review seems to have all, or nearly all, of the optional features enabled.
A feature on the MSOX3034T that sets it apart from other comparable 3000 series oscilloscopes is coded in the T at the end of the model number. This 'scope has a capacitive touch screen! Ideally, a capacitive touch screen enhances user interaction with an oscilloscope by removing operational distractions. A well executed touch screen interface smoothly translates user intent into desired instrument behavior. Alas, the touch screen on the MSOX3034T comes up a wee bit short of the ideal. To be sure, the touch screen is a worthwhile and welcome enhancement, and I wish my Tektronix MDO4000 'scope had one, but using the MDOX3034T touch screen occasionally confused me, frustrated me, or plain baffled me (why can't I edit the size of a zone trigger window, for example?). Examples of touch screen oddities are offered a little later in the review.
InfiniiVision series oscilloscopes are handsome instruments with good esthetic appeal. The 3000T series instrument is light, very light. Most of the meager mass that it has is concentrated on the right side. Attaching the digital signal harness to the connector under the screen, or connecting a USB keyboard or BNC connector to the front panel usually results in a backward pivot of the 'scope on the bench. Other than power supplies it seems many modern test instruments share this characteristic of low mass. Not a significant problem, but the addition of high friction feet would be appreciated.
Observations about the touch screen user interface
A touch screen makes the MSOX3034T easier to use. Given the option to effect an action on the 'scope with a physical button or with a touch screen icon, I most often go for the touch screen icon. Once I learned the correct touch pressure it became first nature to touch icons on the screen to accomplish set up and measurement tasks. In most cases use of Keysight's touch screen features was intuitive and reminded me of ubiquitous touch screen behavior found in tablets and smart phones. There are plenty of examples where the touch screen interface on the MSOX3034T has been implemented flawlessly. In these cases operational barriers and distractions have been removed, allowing the user to concentrate on navigating a path to better signal understanding. A good touch interface is like having your fingers in the signal dough, so to speak, poking, flipping, and stretching signals with ease. Vertical, horizontal, and trigger settings on the MSOX3034T can be easily and quickly adjusted by touching the top of the screen. A ribbon menu pops up offering intuitive and rapid access to essential channel, time base, and trigger set up, as shown below.
On the left edge of this ribbon is an icon that, when touched, brings down a menu of essentially every oscilloscope feature, as shown below.
It is actually fun to use these touch features to set up the 'scope. The touch response is quick and accurate and the behavior of the system is intuitive and familiar. Intuitive familiarity builds user confidence which, in turn, encourages freedom and creative flow. This is one reason why the touch interface can be fun to use. You can, of course, activate and manipulate all instrument functions using traditional physical buttons. I find I am touching the screen more and pushing buttons less as I become more familiar and comfortable with screen based signal interaction.
There are exceptions, however, where I find it more expedient, and less frustrating, to use physical buttons. My choice to use physical buttons is usually motivated by awkward placement or sizing of screen icons. In some cases touch screen icons have been situated far into the corners, mere millimeters away from the screen bezel. Positioning icons in the corners of the screen, especially small icons, makes them difficult to activate. If small icons are collocated with other icons, the probability of activating the wrong icon is significant. For example, the screen draw/screen move toggle icon is situated in the extreme upper left corner of the screen, as shown below.
Often when attempting to touch this icon I end up activating the icon below it. Diagonally opposite this icon is the back icon for the bottom menu ribbon. It is pretty small. I have to turn my finger on edge to avoid activating the much larger menu button beside it.
Something else I find odd about the MSOX3034T touch interface is that, in spite of being capacitive, it is not set up to support pinch/zoom gestures. If you want to use touch to zoom in on a waveform, you draw a box, with a single finger, around a portion of a waveform, then, from a pop up menu, you can choose a "Horizontal Waveform Zoom" or a "Waveform Zoom". The difference between the two options is in which axes are adjusted to achieve the zoomed view. A Horizontal Waveform Zoom adjusts the time base to fill the screen with the boxed portion of the drawn window while preserving the vertical scale settings. A Waveform Zoom adjusts the time base and vertical sensitivity to enlarge both the horizontal and vertical size of the waveform. In my evaluation of touch zooming I encountered several cases where the zoomed result did not exactly match the boxed boundaries I drew on the screen. The video clip below demonstrates the use of the touch screen to zoom in on a single pulse in a pulse waveform. The clip shows a case where the zoomed result does not match the box drawn on the screen.
I learned early on to take care to draw boxes carefully because they can not be edited after your finger is lifted from the screen. If any aspect of a box is misplaced your only option is to delete it and start over. This user constraint probably made UI coding easier while still providing touch based support for waveform zooming, but my preference is to have pinch/zoom gesture support.
A use case illustrating the benefit of segmented memory
I would like to demonstrate an oscilloscope use case that highlights benefits of segmented memory. Not all 'scopes have segmented memory. The Keysight MSOX3034T has segmented memory, the Tektronix MDO4104-3 does not.. This use case comes from experiments I am doing to develop interesting landscape lighting features. I extracted a string of miniature LEDs from a low cost Christmas ornament. The original ornament had a simple drive circuit that blinked the LEDs at various rates. The string is assembled with the LEDs in alternating polarity, so half the LEDs illuminate with one polarity while the other half illuminate with the opposite polarity. You have likely seen these strings in many low cost solar powered lighting products. I wanted to try to make the LEDs behave like embers fanned by a gentle breeze. Think of a flickering effect that changes slowly and randomly in brightness and duration. To do this I used a Keysight 33622A Waveform Generator and combined the two channels into a single output. The first channel generates a PWM signal modulated by very low frequency noise. The second channel generates very low bandwidth noise that is amplitude modulated. Detailed set up parameters for the two channels can be read from the HTML5 web control images below.
The net effect of combining these signals is a random slow changing signal that varies in pulse width and amplitude. In my opinion, the effect kind of does look like embers fanned by a gentle breeze as shown in the short video clip below.
Note that the rapid flickering seen in the video is an artifact of the camera. This flickering is not apparent to a human observer. The measurement challenge is to quantify drive signal parameters. The drive signal is changing continuously but slowly over time in a pseudo-random fashion. I chose to find out how often the noise modulation caused the nominal 10 ms pulse width to decrease below 5 ms and to quantify pulse width behavior when sub 5 ms pulses are generated. My expectation is that the pseudo random noise modulation would drive the signal below 5 ms randomly over time and that the number of sub 5 ms pulses generated would be different each time. For this study I will not worry about the modulation in amplitude provided by channel 2.
To get the oscilloscope to extract information about signal behavior that would verify or falsify my expectation I needed to capture the drive signal whenever pulses with durations less than 5 ms occurred and to do this many times, even if the instances occur seconds, or minutes apart. Only a 'scope with segmented memory can do this effectively. I had no problem setting up the Tektronix MDO4104-3 to capture a single occurrence of sub 5 ms pulses and measure every sub 5 ms pulse that followed the first one, but I also wanted to know how often sub 5 ms bursts occurred. A single capture will not do that. The Keysight MSOX3034T has segmented capture memory and is capable of capturing a series of trigger events even if they are days apart.
I set up segmented acquisition under the Acquire menu. For demonstration purposes I chose to have the 'scope capture 5 instances of sub 5 ms pulse behavior. The trigger condition was a positive pulse width under 5 ms. Trigger mode was set to normal. The 'scope was further set up to search the segmented acquisitions looking for all pulses less than 5 ms in duration. This search would mark all sub 5 ms pulses in each acquisition record and allow me to get duration measurements and a count of sub 5 ms pulses within each instance of sub 5 ms behavior. When Run was pressed, the 'scope began looking for a pulse with a duration under 5 ms. It took 14.7598 s for that to happen. After acquiring the first record, the 'scope began looking again. Four more instances were recorded to segmented memory, the last trigger occurring 53.8399 s after the first trigger. The screen capture below shows the 4th instance of sub 5 ms pulse behavior.
From this image and the analysis work done by the 'scope I see that in this instance there were 10 sub 5 ms pulses with the first pulse having a duration of 4.3057 ms. The instance occurred 29.0794 s following the first captured instance. Analysis of the other captured segments shows that each one had a different number of sub 5 ms pulses (5, 4, 3, 10, and 6). The duration between instances was also variable (14.76s, 1.56 s, 12.76 s, and 24.76 s). These findings support the expectations outlined above. Once properly configured, the Keysight MSOX3034T took care of acquiring and analyzing the data. Segmented memory can be put to use as a nice time saving feature.
I did observe some trade offs and questionable behaviors in this use case experiment. The trade off occurs between sample rate and number of acquisitions dialed into the 'scope. The main time base setting for this experiment is 40 ms/div. With 5 segments captured at that time base the sample rate is 833 kSa/s in high resolution mode or 1.67 MSa/s in normal acquisition mode. If the number of segments is bumped up to 1000 (the max allowed), then sample rate plummets to 8.13 kSa/s in normal acquisition mode or 4.07 kSa/s in high resolution mode. I set up a 1000 segment capture in high res mode and let it run for 10 segments. The low sample rate was evident in the captured trace, but this experiment did reveal some random modulation of pulse amplitude over time provided by channel 2 on the waveform generator. So the experiment provided useful insight. What I question, however, is the believability of one set of measurements shown on the screen capture below, and I question the labeling of those measurements.
The screen capture shows the 5th segment and in addition to a change in pulse amplitude it marks 4 pulses with a duration less than 5 ms. Here is the issue with the measurements. Look at the Events window on the right. It is labelled Pulse Width, which is misleading because although yes, the events listed are the time stamps of when pulse widths were measured below 5 ms, the table does not list the measured pulse widths. Instead it lists the time within the acquisition record that a sub 5 ms pulse was detected relative to the start of the record. That odd choice of label aside, look now at the list of measured time stamps in the Events window. Are we meant to believe that the Keysight MSOX3034T can place the first < 5 ms pulse at precisely 77.98195262 ms? That measurement is provided with 8 decimal places of precision down to the 10 picosecond realm. And this was accomplished at a sample rate of 4.07 kSa/s or one sample every 245.7 us. Very impressive. Or maybe not. My guess is that maybe a variable, say 32 bits wide, holding a value derived from the sample clock was fired off to the screen through a binary to decimal conversion routine without going through any sort of precision limiter. Maybe I'm wrong. Maybe the clock does have 10 ps precision. But why then is pulse width only displayed to three decimal places? Something doesn't make sense here.
Just for comparison, the screen capture below shows how the Tektronix MDO4104-3 non-segmented memory oscilloscope presents pulse measurement data gather in a similar experiment. In this case the single acquisition with 1 M points was sampled at 2.5 MSa/s or one sample every 400 ns. In this capture ten sub 5 ms pulses were found and automatically measured. The search table tabulates all 10 pulse width measurements and their relative position within the record to 1 or 2 decimal places. That level of precision seems more reasonable. The trade off here is that all of the 'scopes memory resources were used up in one acquisition.
I captured the same instance of sub 5 ms pulses on the MSOX3034T. The two instruments were in complete agreement about the relative position of each pulse and their durations, once rounding was taken into consideration. None of my observations in this use case point to flaws or limitations in the MSOX3034T. The 'scope performs measurement tasks very well. What I have observed and reported here are quirks of behavior that can be addressed with firmware revisions, if Keysight decides to do so.
Accessing and controlling the MSOX3034T through the built-in web interface
e14 member michaelkellett asked that I look into the web based instrument control features available through the LAN connection on the MSOX3034T. I am happy to do so. It turns out I learned a few things about the MSOX3034T, and other instruments on my bench, by taking the time to investigate the web interface. Thanks for the suggestion Michael.
The optional LAN/VGA module provided with the Road Test 'scope supports connection of the instrument to an Ethernet LAN and to an external VGA monitor. I find both of these connections valuable in my work. When I was instructing, I often made use of VGA outputs on oscilloscopes to project instrument screens in class. Before VGA outputs became available I used to point a video camera at the screen on the 'scope and project that image for students to see. Definitely not ideal. The LAN connection is of course very useful for capturing screen images for use in reports and Road Tests. Of potential benefit also is the ability to remotely control an instrument over a LAN connection. Many modern test tools support instrument control using the LXI interface and HTML5 or SCPI commands.
I connected an Ethernet cable between the rear LAN port on the MSOX3034T and an 8-port switch on my bench that is in turn connected to the Ethernet port on my laptop. The MSOX3034T is configured for automatic LAN set up, so when it powers up, an IP address is assigned to it. The IP address is needed in order to open up a web browser connection on the laptop. To find the IP address the user opens the Utility menu either via touch or physical button then reads the assigned IP address off the I/O overview page, as shown below.
Back on the laptop a web browser is opened (I used Chrome in this case) and the IP address is entered into the URL box. Hit enter and a few seconds later the web interface home page appears.
From here the user can get an image of the screen (this is how I obtained screen images for this review), save and recall set up and measurement files, check on configured options and calibration status, and control the instrument remotely. Let's take a closer look at remote instrument control options.
There are two options: issue and receive Standard Commands for Programmable Instruments (SCPI) commands, or use a VNC to create a virtual front touch panel. Keysight provides several resources to help learn how to program with SCPI. A brief demonstration will suffice for now.
In this example the *OPT? query was sent to and executed by the MSOX3034T. This SCPI command requests a list of options installed on the instrument. The instrument responded with the text following READ: in the response history window. As you may imagine, controlling an instrument this way is tedious. A much better option is the HTML5 remote front panel. This option brings up an image that duplicates the oscilloscope screen. The computer cursor, controlled by the user through mouse movements, acts like a virtual finger. By positioning the cursor over any touchable portion of the duplicate image the user can activate any and all 'scope functions with a click of the mouse. There is a brief delay as the command is sent to the 'scope, executed, and a screen refresh sent back to the computer. Some virtual front panel implementations are alarmingly slow. For example, the remote front panel for the Keysight E36313A power supply. Screen updates on the PC following activation of virtual buttons can take 4 seconds on my system. The MSOX3034T remote front panel is not that laggy. The video below gives you a sense of the delay I experienced in my system. The screen on the left is driven by the VGA output from the module on the back of the 'scope. The screen on the right is from my laptop where the web interface is running. Simulated touch commands are created with mouse clicks on the right screen, executed on the left screen then updated finally on the right screen. In the video I use the remote touch interface to change the vertical scale on the displayed waveform. I apologize that the remote screen is small, but the Keysight web interface does not allow it to go full screen.
Since the remote front panel is implemented using VNC technology, I thought I would give UltraVNC viewer a try. Yup, it works. UltraVNC Viewer has the advantage of operating in full screen mode. The video clip below shows Ultra VNC running on the right monitor with the MSOX3034T VGA output driving the left monitor. Again, I use the remote touch screen to manipulate the vertical scale on the displayed waveform. The delay in response is noticeable, but acceptable I think in most situations.
Touch screens on oscilloscopes are a great idea. Many vendors now offer oscilloscopes and other instruments with touch interfaces. If you are in a position to purchase an upgrade, I recommend considering a 'scope with a touch screen. Even though the touch interface on the MSOX3034T frustrated me from time to time, there is enough good and useful about touch interfaces to still recommend them. I don't think there is a single touch interface that I haven't had issues with because each has fallen short of its full potential. Check out this fun article in Nautilus magazine which confirms that my frustration with touch screens is shared.
In terms of segmented vs. unsegmented memory, consider the benefits and limitations of each in relation to the use cases you may encounter. I have had need for both over the years. There are two cases I remember well where a huge unsegmented memory was required. In both cases my team was working with scientists that wanted heaps of sample data from transient waveforms to support their research. In those cases 20 M sample acquisitions on single event waveforms were gathered - over and over again. In other cases where the system presents periodic waveforms with a whole lot of nothing happening between events, segmented memory is the better choice.
In spite of the impressive feature set packed into the MSOX3034T I sometimes found it difficult to operate this very fine tool. This is wholly attributable to a gap in my skill set which I blame on decades of indoctrination under the Tektronix method. Apparently using Tektronix oscilloscopes for decades makes one think like a Tektronix engineer. I am therefore grateful that Keysight and element14 gave me the opportunity to spend quality time with an outstanding representative of Keysight measurement technology.
If you didn't take time to explore the blogs at the start of the review, please have a look now. Many other features of the MSOX3034T are covered in the blogs.
All the best,
Great review Mark.
Thanks a lot.
I've been reading the blogs as they came out but I'll go through them all again now your done.
I agree about how one gets used to a particular scope and company, I've used HP/Agilent/Keysight and then LeCroy for a very long time.
It must be more than 20 years since I've used a Tek scope !