|Product Performed to Expectations:||8|
|Specifications were sufficient to design with:||8|
|Demo Software was of good quality:||10|
|Demo was easy to use:||10|
|Support materials were available:||10|
|The price to performance ratio was good:||10|
|TotalScore:||56 / 60|
Keysight DSOX1102G Oscilloscope Road Test
First Impressions (May 1, 2017)
The very first impression I had of the DSOX1102G is that it is small. The 'scope I use most often is a Tektronix MDO4104-3, which is a large instrument, but I have become used to it and its size is my reference frame for what an oscilloscope looks and feels like. So, in comparison to the MDO4000 series, the Keysight 1000 X-series looks tiny. However, it turns out that a lot of capability can be packed into a tiny package at a tiny price.
Whereas my first impression of the DSOX1102G was visual, my second impression, upon opening the shipping box containing the DSOX1102G, was olfactory. That new oscilloscope smell ... IS STRONG, and not as status affirming as the new car smell, because alarmingly, it reminds me of the magic smoke that comes out of burning components. Perhaps I'm catching a wisp of sizzled flux? Or, more likely, trapped residue from a mildly carcinogenic industrial solvent. Not sure, but every time I walk into my lab the aroma reminds me that I have a new 'scope to review!
The DSOX1102G boots up to a ready to go state in 22 seconds flat. Much quicker than other instruments on my bench, except the Tektronix DMM4050, which turns on instantly, but it doesn't have a Windows CE operating system to initialize.
Because the 1000 X-series is so diminutive, the front panel controls are arranged in close proximity. This has not been a serious deterrent to use of the instrument so far, but I have had to move my line of sight a few times to read the knob and button labels that were occluded by closely positioned shafts. I think it wise that Keysight did not attempt to pack four analog channels into a chassis this size. There is room for no more than two sets of vertical controls on this front panel, and the screen in adequately sized to display two traces with a reasonable load of annotations and cursors without feeling cluttered. In a previous Road Test, I illustrated the consequence of trying to pack too much content into an undersized oscilloscope screen. See the photograph below of a very busy screen from a 4 channel Tektronix 2000 series oscilloscope. More is not always better.
Now a comment about the size of the knobs on the 1000 X-series. The smaller multiuse knob on the Keysight 1000 X-series is better for making selections from text lists than the larger multipurpose knob on the Tektronix MDO4000. The selection knob on the Keysight 1000 X-series seems to have superior responsiveness when compared to the response of the multipurpose knobs on the Tektronix 4000 series. The large knobs on the Tek have dead zones where nothing happens, which encourages faster rotation, leading to overshoot. The smaller knob on the Keysight is match tuned to the typical rotation action of a 50+ year old engineering technologist. Much less overshoot.
The push to select feature on most Keysight knobs is satisfying. Push to select allows confirmation of selection and removal of menu text from the screen without moving your hand away from the selection control. Contrast this with the Tektronix approach where confirmation is concurrent with selection. However, you still have to hit "menu off" to clear the menu text from the screen.
One of my initial curiosities about the DSOX1102G was how Keysight was able to pack so much functionality into such a low priced instrument. Other instrument companies are doing the same. Rigol, Tektronix, and Hantek, to name a few, offer feature filled entry level oscilloscopes at reasonable prices from just over $1000. to well below $1000. However, economic reality can not be avoided when manufacturing and supporting a sophisticated piece of test equipment, so after all of the design efficiencies were worked in, compromises had to be made
somewhere. I have found three compromises so far in the design of the DSOX1102G. Two are acceptable, one is kind of annoying.
First compromise: Serviceability
A couple of interesting tidbits I came across in the 1000 X-series service manual. First, from page 51, Replacing Assemblies, of the 1000 X-series Service Guide:
"The service policy for all 1000 X-Series oscilloscopes is unit replacement, so there are no instructions for replacing internal assemblies in this service guide." And from page 52, Replaceable Parts: "Because the service policy for 1000 X-Series oscilloscopes is unit replacement, no replaceable parts are available for the Keysight 1000 X-Series oscilloscopes."
So, one way to make an instrument cheaper is to not stock replaceable parts or pay for technicians to tear apart failed units. It is cheaper to replace a failed unit than to fix a failed unit. This saves Keysight, and you, the end user, money up front, but consider the consequences when your 'scope hits its 3rd (or 5th, with extended warranty) birthday. Anything goes wrong after that day and you either buy a new 'scope, or hunt for parts on the secondary market.
Second compromise: Quality of Integrated Bonus Features
It is great to have a digital multimeter, frequency counter, and waveform generator built right into your oscilloscope. Think of the space you will save on your bench. Think of the cost saving. Yes, but what sort of multimeter, frequency counter, and waveform generator can you pack into an oscilloscope that costs less than a good bench multimeter, less than a good frequency counter, and less than a good waveform generator? I did a quick A/B comparison of the square wave output from the DSOX1102G vs. the square wave output from a Keysight 33622A Waveform Generator. To be fair, the 33622A I used for comparison has a retail price of CAD $9400 and is a superb bench instrument where all of the cost went into making it generate high caliber waveforms. Nevertheless, as a point of comparison to a high caliber waveform generator, the DSOX1102G waveform generator has some limitations. The screen capture below shows 10 MHz square waves of identical amplitude generated by both instruments. The DSOX1102G is the magenta trace and the 33622A is the green trace.
Note that the DSOX1102G square wave has slower rise (16.80 ns) and fall times compared with the 33622A (3.466 ns). This won't always be a concern, especially in typical entry level applications, but it is a compromise to be aware of.
Some of the other compromises that seem to have gone into keeping the overall cost of the 'scope low include choices about which wave gen features to include. For example, the DSOX1102G waveform generator does not have an arbitrary waveform capability, but it does have a variable noise feature that allows you to sum in amplitude noise to any generated waveform. Both features are useful in my work, but if I had to give one up one of the two to keep an entry level 'scope affordable, I would have given up the arbitrary feature, just as Keysight did. I didn't start using arbitrary waveform generation until I was well into my career, so not having it on an entry level 'scope is probably okay.
Third compromise: Connectivity to a PC
Okay, this one kind of bugs me. Most of my other Keysight instruments (bench multimeters, frequency counter, waveform generator, SMU) are supported in Keysight's BenchVue application. My Tektronix oscilloscope has a handy but limited utility PC interface, as does my B&K Precision DC power supply. But, alas, the 1000 X-series is not supported in BenchVue, and so far, I have been having a difficult time getting my Windows 10 PC to talk with it over USB. I'm still working on that and hope to have communication set up soon. In the meantime, the most expedient method I could find to capture a waveform image, was to use a USB memory stick and the "Save to USB" feature. That is how I captured the image below.
This trace, by the way, is the result of holding the 'scope probe firmly in one hand, while simultaneously giving the probe tip a firm flick of a finger with the other hand. The resulting wobbly transient is probably the result of piezoelectric voltage generation caused by the impulse input of mechanical energy into ceramic caps in the probe. My inspiration for trying this came from watching Dave Jones EEVBlog #162. Fascinating stuff.
Now, to be clear, I think the Keysight DSOX1102G is an awesome entry level oscilloscope. I have been very pleased by several of the behaviors and features that clearly show that a lot of effort went into making this a great, low cost, high quality instrument. And I will discuss those features in subsequent updates to this review. This entry captured a few initial impressions, and they are good impressions, informed by decades of oscilloscope use and some sensible caveats.
The benefit of a custom ASIC (May 3, 2017)
Before diving in to a simple demonstration of how an entry level 'scope beats the pants off of a mid range 1 GHz 4 channel mixed domain oscilloscope, I'd like to present my unboxing video. I don't understand why people do unboxing videos. Why is this a thing? Do they expect to find something different in the box other than what the shipping documents clearly say will be in the box? At any rate, here is my version of an unboxing video.
Now that we have that out of the way, let's take a look at the difference between Application Specific Integrated Circuit (ASIC) or hardware waveform analysis and software waveform analysis. I set up a simple A/B comparison to illustrate one of the benefits Keysight 'scopes have over Tektronix 'scopes. I will be illustrating two examples that show the superiority of ASIC waveform analysis in this review. This first example is a very simple example showing how an automated peak-to-peak waveform measurement performs on a 1 Msa waveform record on the two brands. The second example (to be uploaded at a later date) will compare ASIC serial bus decode with software bus decode. The difference in performance is remarkable.
As an interesting aside, you could buy 17 DSOX1102Gs for the price of the MDO4000 series 'scope I'm using for comparison purposes.
For the first illustration I sent a 1 KHz, 1 Vrms sine wave into both 'scopes through a BNC T connector. The inputs on both 'scopes are DC coupled because I will be adding DC offsets to the sine waves and I want to see the shift in waveform position as the DC offset is changed. To maintain a good trigger as the DC offset shifts, the trigger on both scopes was set to AC coupling. On both 'scopes an automatic peak-to-peak measurement was selected and on the MDO4000, a 1 M sample record length was selected to match the 1 MSa memory in the DSOX1102G. The MDO4000 series has a much deeper sample memory capable of being set at up to 20 M samples. But to keep the comparison fair, the MDO4000 was throttled back to 1 MSa.
In the video clip below you will see a side by side and individual comparison of how the two oscilloscope perform when a DC bias is added to the sine wave. Both scopes perform beautifully tracking the signal as the DC bias first goes to -2 VDC, then to +2 VDC, and finally back to 0 VDC. On both displays watch the dashed lines that show where the scope has determined the peaks to be on the sine wave. See how the DSOX1102G responds to a sudden shift in DC offset vs. how the MDO4104-3 responds. The Keysight ASIC computes the P-P markers and values in much less time than the Tektronix software. You can see this especially well in the final portion of the video where the MDO4104-3 is shown alone.
How important is this difference in approach between the two vendors? Well, if you are not in a hurry, there is no real difference. They both get the job done and done well, its just that it takes longer on the Tektronix. In this first illustration the difference is only a few seconds. In the next illustration I will be using both scopes to decode lengthy I2C serial data streams. I think we will see a more significant lag in performance from the software decode used in the Tektronix.
Just before getting to the comparison of segmented vs non-segmented memory and ASIC vs software bus decode, I'd like to illustrate how to assemble a 1X/10X scope probe right out of the package and show two methods for performing a compensation check. First, a quick video on probe assembly:
Nothing too difficult there, but all important nonetheless. Once a probe is assembled, it needs to be checked for proper compensation. The compensation check procedure is not executed very often, especially if a probe is not moved from one scope to the next or from one channel to another on the same oscilloscope. In many cases a probe kind of takes up residence on one particular channel of one oscilloscope, thus the value of putting on the colored channel identification rings as shown in the assembly video. It is important to make sure your probes are properly compensated, otherwise they will introduce distortion to the signals you are measuring. The first video below shows the traditional method for compensating a probe using the non-metallic tool provided with the Keysight probes that came with the DSOX1102G.
The next video illustrates the probe check feature built into the 1000 X-series oscilloscopes.
Now, although the assisted probe compensation feature is cool, it is not essential. I think of it as being similar to finding an extra cup holder in your new car - nice to have, but not the deciding factor in making the purchase. Probe compensation is not a frequently performed operation in most use cases, so this feature is not crucial. It is useful, however, as a teaching tool for those users that are new to oscilloscope operation. The on screen text provides all the detail necessary to walk a novice through the whole procedure and the software includes a check to make sure the compensation was done correctly.
The DVM/Frequency Counter and challenging triggering (May 15, 2017)
The DVM and frequency counter functionality in the DSOX1102G are nice extras. Because they are secondary to the main application of the oscilloscope they do not have the same range of capability or precision you would find in dedicated DVM or counter instruments, but both the DVM and frequency counter built into the DSOX1102G perform well. In the video below I first set up the scope to 200 mV/division with the ground reference at the very bottom of the screen. This establishes a 1.60 V range on the screen (8 div x 200 mV/div = 1.60 V). When the DVM feature is selected from the Analyze menu (oddly not from the Measure menu), a 3-digit voltage is displayed along with a bar graph scaled to the same range available on the scope display. In this case the bar graph auto scales to 0.00 on the left end and 1.60 VDC on the right end. In addition to the digital display of DC voltage, a small blue triangle is placed on the bar graph to give a visual representation of the current DC voltage. The bar graph has a history feature that provides an indication of recent changes in DC voltage. A white history bar appears with a time constant of a second or so that follows the blue triangle as the input voltage changes. This behavior is illustrated in the video below.
While setting up this illustration, I noticed the Keysight DVM tool was registering a frequency of about 30 kHz, on a 1.23 VDC output from a DC power supply. Odd. The scope had triggered on a periodic. low amplitude burst of noise that was riding on the DC from the supply. I don't know if this noise burst came from the supply, or (more likely) was being coupled into the long leads connected to the supply from something else running on the bench, but the impressive thing was the Keysight scope triggered on this low amplitude noise signal with great ease. You can see the noise as displayed on the Keysight scope in the video. Look at the background noise around the periodic burst. You can see that the Keysight scope is updating waveforms at at a really good rate.
I was curious to see if the Tektronix MDO4104-3 on my bench could also trigger on this low amplitude noise burst. After some careful configuring, I was able to see the exact same periodic noise. To see it using the Tek scope I had to turn on the Fast Acquisition feature. Thus limits the record length to 1000 samples, which allows for much a much higher waveform update rate. You may be able to see in the background that even on Fast Acquisition mode, the Tek scope does not update as fast as the Keysight scope. However, the Tek scope has a useful Waveform Palette feature that colors traces according to their frequency of occurrence. In the video you will see that periodic parts of the signal (the noise burst) which occur most often, are colored red, while the random very high frequency noise in the background is colored blue. Bottom line, both scopes triggered on a low amplitude periodic noise burst that had some how crept into my DC power supply output. The Keysight scope triggered very easily on this signal with just one control adjustment (trigger level). To achieve similar results on the Tek scope I had to switch to Fast Acquisition mode, or I wouldn't have seen the burst at all.
Why it can be important to let instruments warm up before use (May 24, 2017)
Many operating manuals recommend that instruments be allowed to warm up for 30 minutes prior to use. In the days of vacuum tube (valve) based equipment this was necessary to let the heaters inside each tube reach operating temperature resulting in stable operation. Even with 100% digital instruments running complex operating systems like the Keysight DSOX1102G, it is wise to give the instrument some time to warm up before making any critical measurements. It still takes time for solid state devices to reach a stable operating temperature. Most everyone knows that CPUs in personal computers can run much hotter than ambient and require forced air cooling, and massive heat sinks, to keep them within safe operating temperature zones. Nearly every piece of test equipment on my bench has a cooling fan designed to remove excess heat from the electronics inside the instrument.
Some instruments have thermally controlled ovens to keep reference oscillators at specific temperatures to obtain maximum stability. For example, the Keysight 53230A Universal Frequency Counter can be configured with an optional Ultra High Stability OCXO (Oven Controlled Crystal Oscillator) Timebase. The 53230A on my bench has this option. The photo below shows the oven temperature after about 30 minutes of warm up time.
After taking the photo above, I let the counter run for another hour then checked the oven temperature again. The oven had reached 37.8 degrees C. As you can see it takes quite some time to reach a steady state oven temperature.
I data logged the output frequency of the DSOX1102G Wave Gen on the 53230A counter for 30 minutes following a cold start. The Wave Gen was set up to produce a 500.0 kHz sine wave with an amplitude of 5.00 Vp-p. The 53230A counter had been running for several hours and had reached a stable oven temperature. The resulting log file is shown below.
As you can see, the output frequency decreased over time until it stabilized at 499,997.673 Hz after about 30 minutes of warm up. Now, to put this in perspective, the power-on output frequency was 499,998.563 Hz, so we are talking about a drift of 0.89 Hz over 30 minutes, which for all but the most demanding applications is trivial.However, the more general point is that test equipment needs time to stabilize after power-up. In this case it has been shown that the Wave Gen in the DSOX1102G powers up very close to the set frequency and shows very little drift over time, however, after about 30 minutes the drift has dissipated and the output remains very stable from then on.
Illustration of segmented and non-segmented memory (June 15, 2017)
Many digital oscilloscopes, like the Tektronix MDO4000 series, do not offer memory segmentation. When a trigger event occurs, a small amount of the waveform before the trigger point is captured, along with the trigger event and everything following the trigger event until acquisition memory is filled. In many cases this approach works well. For example, if you are examining periodic waveforms that are continuous in nature, non-segmented memory works well.
There are many real life situations where non-periodic signals, or signals that are intermittent, or signals that occur sporadically with long intervals of uninteresting nothingness between bursts need to be examined. During a recent research project our team had to capture and count very low amplitude signals (essentially just a little higher in amplitude than the background noise) coming out of a prototype in-line particle detector. At very low particle concentrations, several seconds could go by between particle detection events. In situations like this, segmented memory is a great feature to have in your 'scope. It was important for our team to capture as many detection events as possible in order to characterize the particle interaction signatures. A 'scope with non-segmented memory would capture a particle detection event and then fill up the rest of acquisition memory with noise, or maybe it would catch another detection event, but there is no assurance another event would occur before memory filled. A 'scope with segmented memory on the other hand, captures a particle interaction event, then stops acquisition until the next occurrence of the trigger event, whenever that happens to occur. It continues to do this until the specified number of acquisition events occur.
'Scopes without segmented memory can compensate somewhat by offering huge amounts of acquisition memory. Deep acquisition memory captures much more post trigger activity, increasing the probability of capturing additional events of interest in the record. In our case we were using a non-segmented memory Tektronix MDO4000 series with 20 MSa of acquisition memory. We made it work, but we wished we had a segmented memory oscilloscope.
To illustrate the differences between non-segmented and segmented memory I used the DSOX1102G and the MDO4104-3 to capture 10 separate Real Time Clock I2C interactions between an RTC chip and a micro-controller. One second of inactivity passes between each interaction. By using the segmented memory capability in the Keysight DSOX1102G, only the interactions of interest are captured. The lengthy gaps of idle bus signals in between are not captured.
The video clip below shows how the DSOX1102G captures 10 individual real time clock transactions, each 1 second apart, quite nicely using segmented memory, even though the X-1000 series has a relatively small 1 mega sample acquisition memory.
The next video clip shows how the Tektronix MDO4104-3 without segmented memory, but with a beefy 20 mega sample acquisition memory, is still capable of capturing the 10 I2C interactions. The difference is, the Tektronix 'scope also captures the bus idle time in between interactions in full detail. There are cases where it might be helpful to know what happened during the bus idle intervals, and a 'scope with deep memory can help with those situations.
The Keysight user interface experience (June 21, 2017)
Front panel knobs and switches are the primary manual interface that allow users to adjust the configuration of an oscilloscope. Of course, there is a trend toward increased use of capacitive touch screen interfaces, such as on the Rohde & Swartz RTB2004 currently up for Road test evaluation, and on higher end Keysight scopes. Some Keysight scopes even have touch screen AND voice control. However, for the majority of models, the primary manual control interface is the the array of knobs and buttons on the oscilloscopes front panel. Using the knobs on the Keysight X-1000 series provides a different experience than when using the knobs on the Tektronix 4000 series. The difference is primarily related to Keysight's generous provision of "push to select" on ALL nine of the front panel knobs. The Tektronix MDO4000 series 'scope on my bench has fifteen knobs. Only one of those fifteen knobs, the trigger level control, has a "push to select" capability. Obviously, the Tektronix user interface is designed to operate without the use of "push to select" knobs, and for the most part, everything you can do with the Keysight knobs, you can also do with the Tektronix knobs, albeit with a few more steps. The one feature I could not duplicate on the Tektronix MDO4000 'scope was fine step adjustment of the time/division setting.
I find the "push to select" feature on the Keysight knobs very intuitive. The "push to select" feature is also a bit of a time saver. Examples of what can be done with the "push to select" feature on the X-1000 'scope are shown in the video below.
For comparison, the video ends with a example of how to obtain fine control of the vertical V/div setting on the Tektronix MDO4104-3.
Still to come: