Entry Level Oscilloscope Comparison:  Keysight, Rigol, Siglent

Introduction

In this blog, the features of three popular oscilloscopes marketed as being entry level will be reviewed and compared.

• Keysight DSOX1102G hereafter referred to as Keysight

• Rigol DS1054Z hereafter referred to as Rigol
• Siglent SDS 1102CML hereafter referred to as Siglent

tl;dr – See Conclusions at the bottom of this post.

The comparison will highlight features and ease of use and will not be a tutorial on how to use the scopes.  The manufacturer’s specifications will not be verified.  It is a somewhat subjective report of what it is like to live with these three scopes  over several days while performing basic and some more advanced analysis.

Focus will be on the Keysight since it is the newest of the scopes and I need to assess it.  The Keysight is also the most feature rich and will be used for base comparison to the Rigol and Siglent.  The training material for the Keysight will be followed and then a comparison to the Rigol and Siglent made.  The focus will be on suitability for students, hobbyists, and users needing an instrument costing in the $1000 and under range.  Alternatives to purchasing some of the features in the scope will also be discussed.

I own all three scopes and either purchased them outright or in the case of the Keysight DSOX1102G won the scope in a contest unrelated to Keysight.  I have had the Siglent SDS 1102CML the longest and up until now it has been my go to scope and daily user.  The Rigol DS1054Z was purchased preowned to get 4 channel capability.

All three are easy enough to learn although the Keysight has the best educational material for those just starting out and ultimately it was felt that it had the most intuitive controls.

To start, here is a summary table of the three oscilloscopes.  All three manufacturers have models with more and less features as well as higher and lower costs.  For example, the Rigol in the table has 4 channels and the Keysight and Siglent have 2 channels.  If a 4 channel oscilloscope is needed then the 4 channel models from Keysight and Siglent might be considered.  The table only covers those models personally at hand.  Siglent has been replaced the SDS 1102CML with a newer model but it is still sold new as this is written.

Feature Summary

Specification Discussion

As tested, all are 100 MHz scopes although EEVblog #978 (and the comment section below it) describes how the Keysight can be “hacked” to give 200 MHz performance with modified firmware (apparently the hardware modifications in the video are not needed). The Keysight used in this review is stock.  The 50 MHz Rigol can be “hacked”, or upgraded from the manufacturer, to give 100 MHz performance.  The Rigol in this review has been upgraded to 100 MHz.

These days, the low cost to get 100 MHz bandwidth makes it a worthwhile upgrade.

The Keysight has twice the sample rate of the other two scopes when in high resolution acquisition mode and also a higher waveform update rate.  A higher sample rate is required to get full bandwidth in a single acquisition (and the Keysight sample rate is adequate for a 200 MHz scope). Higher waveform update rate is important when attempting to capture random and infrequent anomalies.

Both the Rigol (24 Mpts) and the Siglent (2 Mpts) have more memory depth than the Keysight (1 Mpts). Keysight cites segmented memory as an offset for this but it seems surprising with memory so cheap that the Keysight only has 1 Mpts.  Segmented memory works by digitizing only “important events” and time tagging them. Keysight uses the illustration below taken from their “How to Select Your Next Oscilloscope” brochure to show the advantage.

Nonetheless, more memory would be welcome since events may not always be so accommodating.  Use of segmented memory is demonstrated below.

As noted earlier, the Rigol is a 4 channel scope while the other two are 2 channel.

The Rigol also has more triggering options.  This is typical of the Rigol – there are lots of options in the menus, some of them minor variations.  The Siglent is more basic.  It is worth looking through the User Manuals to see if options such as these might prove useful.

An area where the Siglent seems to offer an advantage is time base and vertical sensitivity.  The rise time seems to support the faster time base and the vertical sensitivity was useful in the recent Experimenting with Polymer Capacitors contest.  I am not sure if the lower vertical sensitivity is hardware or software based on the Siglent.  And while there is quite a bit of electrical noise at the lowest setting it has proved useful for my measurements in the past.

Mask testing is available on the Keysight and Rigol and is demonstrated below.

All three oscilloscopes have the same math functions.  FFT is demonstrated below.

The Siglent has the lowest resolution display and the font is coarse.  However, it is large and quite legible.  The Rigol has a higher resolution screen but the screen is crowded and difficult to read compared to the Siglent.  It also shows glare and angle of view does not seem as good as the other two.  The Keysight has the nicest display of the three to my eye.  Screenshots from the Keysight and Rigol are much nicer than what comes off the Siglent.

Serial decoding is only offered on the Keysight and Rigol.  It is an option on the Keysight.  An alternate is to use one of the inexpensive 100 MHz logic analyzers available and signal analysis software like sigrok which is what I used when the Siglent was my only scope.

The wave generator and Bode plot are all unique to the Keysight and are demonstrated below.

All three oscilloscopes offer USB connectivity.  The Rigol adds has network connectivity while the Siglent adds RS232.

All three scopes have probes suitable for their matching scope.  The 200 MHz probes on the Keysight would again suggest the overall instrument might be good for 200 MHz with a firmware upgrade.  The Rigol probe cables are less flexible than the other two and the ground alligator clips stiffer.  I prefer the Siglent and Keysight probes for ease of use.

Subjectively, the Keysight fan is much noisier than the other two scopes with the Siglent being the quietest.  It is not overly annoying to me but is noticeable and runs constantly.

The costs are indicative only.  If interested in one of the scopes then spec it out and do an online search for the scope in your currency.  Costs can vary due to discounts, specials, etc.

The Keysight training material will be covered in more detail below.  The Keysight documentation is good with the Rigol somewhat behind it. The Siglent is a significant step below but it is a simple scope and not hard to figure out.

Setup and First Impressions

All three were set up in same climate controlled space and warmed up for the same amount of time.

All three were easy to set up.  The controls, menus, and placement differed enough to slow things down going from one scope to the other but all are manageable.  This is touched on further below.  All three are running the most recent firmware available from the manufacturer.  All scopes are using the probes provided when purchased and probes have been compensated per the manufacturer’s instructions.

Time to boot up:

• Keysight: 22 seconds
• Rigol: 19 seconds
• Siglent: 11 seconds

Feature Review

To review features it was decided to first take the Keysight Training Material (excellent by the way) and go through it with each scope lab by lab.  Additional features not covered in the labs such as the Mask, Voltmeter, and Bode Plot are also reviewed.

The reference or training signals available from the Keysight Help menu are used extensively in the review. The screenshot below shows some of the available waveforms.

The time to polish the display output for the screenshots on the oscilloscopes was not taken in all cases. Where there was difficulty getting a good display a comment is made however.

Measurements on Sine Waves

Lab 1 teaches manual measurement of a sine wave on the Keysight using the built-in training wave forms.  Since the other scopes do not have built-in wave generation, their probes measured the output from the Keysight.  Measurement on all of the scopes was equally easy for the Sine wave.  A screenshot was taken from each oscilloscope and posted below along with comments.

Keysight Screenshot

The Keysight screenshot is clean and easy to read as is the display.  After getting used to the placement of the horizontal controls which differs from the Rigol and Siglent I find it easy to use.

Rigol Screenshot

I find the Rigol screen a bit busy but easy to set up and use.

Siglent Screenshot

The display of the Siglent is the same physical size as the other two but has lower resolution and the vertical and horizontal cursors cannot be set and seen at the same time.  Only the horizontal cursors are set in the screenshot above.

Summary:  All are acceptable for entry level use but the Keysight shines and the Siglent lacks the ability to display horizontal and vertical cursors at the same time.  The control knob feel is best on the Keysight.

Basic Oscilloscope Triggering

Lab 2 goes over basic triggering using the sine wave generated in the training suite.

Summary:  All scopes have adequate basic triggering.

Triggering on Noisy Signals

Lab 3 uses the waveform generator on the Keysight to generate a noisy 1 kHz Sine wave.  The noise is sufficient to generate an occasional rising edge trigger outside of where expected.  There is built in hysteresis of approximately 0.5 divisions on the Keysight so the unwanted trigger only occurs when the noise exceeds that level.  Since the vertical resolution has been set to 500 mV per division, the noise in this instance is occasionally exceeding 250 mV.

Below the Keysight is shown with the default trigger setting and a noisy Sine wave.  The scope occasionally has unwanted triggering on the noise as it passes the trigger point.

There are two ways to reject noise on the Keysight:

• Noise rejection which raises the trigger hysteresis to 1 division

• HF rejection which rejects frequencies above 50 kHz in the analog trigger circuitry

Since the Sine wave is 1 kHz and the noise is less than 1 division then either rejection method will remove the noise from the trigger.  In the screenshot below the trigger is set for both noise rejection and HF rejection but either is sufficient to remove the unwanted trigger.

It is possible to smooth the waveform and remove noise by setting the acquisition to 8x averaging which allows for more accurate manual measurement as shown in the Keysight screenshot below.

The Rigol occasionally triggered on noise the same as the Keysight.  Noise reject and averaging worked as expected and similar to the Keysight. It does not have a high frequency reject for trigger.  A screenshot of the Rigol with trigger noise rejection and 8x averaging is shown below.

The Siglent did not trigger on the noise with vertical set to 500 mV.  It started to trigger on the noise when vertical resolution was at 200 mV and not all of the wave was visible in the viewing area.  It rejected the noise most of the time, but not all of the time when Noise Reject was turned on. It does not have a high frequency reject for trigger.  The cursors look fine on the Siglent display but intensity could have been turned up for the screenshot.

Summary:  Removing noise was possible on all of the scopes.  All scopes have a noise rejection mode.  The Keysight also has the ability to do fixed 50 kHz frequency rejection in the trigger circuitry (but is not suitable at frequencies above 50 kHz).  The Siglent has a larger default hysteresis window (greater than 0.5 divisions).

Saving and Recalling Information

Lab 4 demonstrates how to save screenshots, scope settings, and waveform data as well as recalling it.

The Siglent only saves in BMP format for screenshots.  The Keysight and Rigol can also store PNG and the Rigol also stores JPEG.  The Keysight can name files which might be useful but entry is tedious with buttons and knobs.  The Rigol and Siglent have dedicated screenshot buttons.  The Keysight has a save to USB button that will save to whatever mode it is in (screenshot, scope setting, etc.).

Summary: All of the oscilloscopes have acceptable save and recall capability.

Probe Compensation and Probe Loading

Lab 5 covers how to compensate probes and has a useful discussion on the impact of probe loading. The Keysight N2140A 200 MHz passive probes provided with the instrument differ from the Rigol and the Siglent in that trimmer capacitor is on the BNC connection instead of on the probe but this is no consequence.

Summary: All of the oscilloscope probes have acceptable probe compensation.

WaveGen Waveform Generator

Lab 6 covers Waveform selection and setting Frequency, Amplitude, and Offset which is available only on the Keysight.  The waveforms available are shown in the screenshot below where a pulse has been selected.

Duty cycle can be set for Square waves and pulse width for Pulse. Noise can be added.  AM, FM, and FSK modulation can be added.  Output load can be set to 50 Ohms or High-Z.  The example below shows a 200 kHz triangle wave 3 V peak to peak with 0.5 V offset.

The available ranges follow.

Frequency:

• Sine: 0.1 Hz to 20 M
• Square wave / pulse: 0.1 Hz to 10 MHz with duty cycle 20 to 80%
• Ramp / triangle: 0.1 Hz to 200 kHz

DC offset:

• Square, Pulse, Ramp:  ± [10 V – ½ amplitude] into Hi-Z  ± [5 V – ½ amplitude] into 50 Ω
• Sine:  ± [8 V – ½ amplitude] into Hi-Z  ± [4.5 V – ½ amplitude] into 50 Ω

Amplitude:

• Square, Pulse, Ramp:  2 mVpp to 20 Vpp into Hi-Z (offset ≤ ±0.4 V)
• Square, Pulse, Ramp:  1 mVpp to 10 Vpp into 50 Ω (offset ≤ ±0.4 V)
• Square, Pulse, Ramp:  50 mVpp to 20 Vpp into Hi-Z (offset > ±0.4 V)
• Square, Pulse, Ramp:  25 mVpp to 10 Vpp into 50 Ω (offset > ±0.4 V)
• Sine:   2 mVpp to 12 Vpp into Hi-Z (offset ≤ ± 0.4 V)
• Sine:   1 mVpp to 9 Vpp into 50 Ω (offset ≤ ± 0.4 V)
• Sine:  50 mVpp to 12 Vpp into Hi-Z (offset > ± 0.4 V)
• Sine:  25 mVpp to 9 Vpp into 50 Ω (offset > ± 0.4 V)

Summary: Although not extensively tested the Keysight waveform generator worked as advertised over the range and settings examined and is quite useable.  The function generator is unique to the Keysight and not present on the Rigol or Siglent.  It elevates the scope beyond basic capability.

Triggering on a Digital Burst

Lab 7 covers the use of Holdoff to trigger on a digital burst.  A burst can cause the trigger to occur on any rising edge without Holdoff as shown in the screenshot below from the Keysight.

A single shot capture shows that the burst interval is around 840 uS and time between bursts is about 1000 uS.  The Holdoff interval is set to be somewhere between the width of the burst and the time interval between bursts.  In the case of the lab Holdoff is set to 920 uS causes triggering on the first rising edge as shown on the Keysight screenshot below.  Horizontal settings are unchanged.

All of the scopes have Holdoff triggering.  Setting Holdoff on the Keysight is coarse but quick.  It jumps from 60 nS to 10 uS and then increments in 10 uS intervals and is velocity dependent.  Using it took some getting used to.

The Rigol has very fine adjustment with the knob that is not velocity dependent. However, pushing the knob brings up a numeric entry pad which is nice.  The Rigol is triggering properly in the screenshot below.

The Siglent has fine adjustment but is not particularly velocity dependent so adjustment takes a longer time.  The Siglent triggers properly but ends up with a very coarse display.

It is relatively easy to set Holdoff back to the minimum level on all the scopes.

Summary: All three oscilloscopes performed acceptably while triggering a digital burst after adjusting Holdoff.

Triggering and Analyzing Infrequent Events

Lab 8 has a 50 kHz clock signal with an infrequent and intermittent glitch.  By turning intensity up, and setting the persistence to a high value it is possible to see the glitch as shown on the faint trace from the Keysight in the screenshot below.

It is not nearly so clear what is going on when viewed on the Rigol as seen in the following screenshot. Is it a lower update rate on the Rigol causing the waveform to not capture correctly?

I could not get the Siglent to properly show the glitch, even with infinite persistence.

Pulse width triggering mode can be used to isolate the glitch on the Keysight.  From the screenshot above it can be seen that at a 50% trigger level the glitch has a pulse width less than 1.6 uS.  Since the glitch occurs less often than the default Auto triggering rate it is necessary to change to Normal trigger mode.  Changing the trigger type to Pulse Width and less than 1.6 uS captures the glitch cleanly as shown below.

While both the Rigol and Siglent have Pulse Width trigger types I could not get either to properly isolate the glitch.

Summary: The Rigol could not cleanly isolate the glitch but it was possible to at least tell that something was wrong.  I could not get the Siglent to recognize the glitch at all. Only the Keysight was able to quickly show the problem and cleanly isolate the glitch.  This bears further comment.  Just because a scope is say 100 MHz and has a feature such as Pulse Width for trigger types it does not mean it can necessarily perform as well as another scope with the same specifications in all instances.

Capturing a Single-shot Event

Lab 9 describes a single shot event and the scope settings to capture it.  The training signal is a single shot pulse with ringing. 

Here it is captured without problem on the Keysight.

The Rigol and Siglent also captured it without problems.

Summary: All three scopes are easy to set up and use in single shot mode and captured the training signal without a problem.

Automatic Parametric Measurement of Digital Waveforms

Lab 10 demonstrates the many automatic parametric measurements that can be done on a digital oscilloscope.  A repetitive pulse with ringing is set up on the training signals.  Up to four parameters can be displayed at the bottom of the scope along with cursors indicating the last one entered.  The following screenshot shows top, base, rise, and fall at the bottom of the screen for the Keysight with cursors framing the fall time.

A quick capture of all the parameters can be done with a “snapshot” that tends to overlay the signal since it is placed in the middle of the display.

It would be nice if Keysight moved the summary table off of the center of the display.

Greater flexibility is available on the Rigol for individual parameters.  There is also the ability to do a quick capture of all signals and the signal can be placed such that the information does not obstruct it. However, I find the information somewhat harder to read due to size and layout.

Of course the Siglent has its own way of doing things.

Summary: All of the scopes do automatic parametric measurement.  It would be nicer if the Keysight did not place the full summary directly in the middle of the screen.  I prefer the layout of the Keysight information in general though that is subjective.

Using Zoom Timebase to Perform Gated Measurements

Lab 11 sets up a digital burst with infrequent glitch on the training output.  The unwanted trigger from the glitch is then removed with Holdoff.

By setting measurement to +width the width of the first pulse is displayed.  To measure the other pulses, the magnifying glass button is used to enter the “Zoom Timebase” mode and set the zoom timebase can be modified.  The first pulse then shows up magnified with measurements in the lower window and the full pulse burst and zoom window visible above it.   The window can be moved with the horizontal position delay.  For example, the 4^th pulse can be selected and displayed like shown below.

In the screenshot below the Rigol is doing something similar and tracking the 4^th pulse with the cursors after pushing the horizontal control into “Scale” mode.

The Siglent enters zoom mode after pushing the Horizontal Control labelled “Push-Zoom”.  Using track in the cursor setup gives the following. Apologies for pushing the screenshot button before the USB FlashDrive message cleared – my fault and not the scopes.

Summary: All three scopes can perform Zoom or gated measurements but the Keysight seems easier to set up and the display is cleaner.

FFT Analysis

Lab 12 demonstrates using FFT to find the frequency of a glitch occurring in a “clock with infrequent glitch”.  The timebase is set  at 2 ms/div to capture many cycles which improves the precision of the FFT math function.  The FFT key on the front panel is pushed and frequency domain pops up over the time domain. The cursors on the Keysight can then be manually placed over the fundamental and 3^rd harmonic to measure them as shown below.

It was difficult to get a good FFT display on the Rigol with a timebase of 1 to 2 ms/div so it is set at 100 us/div.  The measurement is again on the fundamental and 3^rd harmonic.

The Siglent was able to measure at 1 ms/div but did not display as much information as the Keysight. Below it is measuring the fundamental and 3^rd harmonic, same as the others.  Unfortunately the information is superimposed right on top of the fundamental and adjustment is not easy.

Summary: All three scopes can do FFT analysis.  The Keysight is easier to set up, the display is cleaner, and it seems to be capable of more precision.

Using Peak Detect to Overcome Under-sampling

Lab 13 uses a Sine wave with a fast glitch to illustrate peak detect.  Without peak detect on a relatively low frequency signal the intermittent glitch might be missed or not all points on the glitch seen.  With peak detect the highest value is retained and visible.

All of the oscilloscopes have peak detect which worked without issue.  Below is the Keysight with the glitch visible just before the peak of the Sine wave.  Intensity could have been set higher to make it more visible but it was OK on the display.

Here is the Rigol.

And the Siglent.

Summary:  Peak detection worked well on all three oscilloscopes.

Using Segmented Memory to Capture More Waveforms

Lab 14 demonstrates how segmented memory can capture bursts separated on a long time scale accurately. The lab starts out by capturing a “RF burst” at 10 us/div time scale.

All of the scopes can capture this accurately.  The time scale is increased to 50 ms/div and a capture is made with the Run/Stop Button.  This time the bursts are captured over a much longer time period.

Degradation is expected when zooming in on one of the bursts and that is what occurs.  Here is the above 50 ms/div capture with horizontal time zoomed in to 5 us/div.

The detail is lost.

Segmented memory captures bursts and stores them as time stamped segments.  Here is the 21^st burst from a long string using segmented memory.

All the detail is back.

However the Keysight only has 1 Mpts of memory depth.  The Siglent has 2 Mpts which is not that much better.  But the Rigol has 24 Mpts.  For this test at least can it compete?  Turns out it can.  Here is the Rigol after a capture at 50 ms/div and then zoomed in to 5 us/div.  It doesn’t look that bad.

Look at that tiny little window at top center to see how much of the memory is being displayed.  A contrived test could be developed with bursts further apart but that does not change the fact that the Keysight has low memory depth.

Summary:  Segmented memory works on the Keysight but it does not completely make up for the low memory depth.  The Siglent is also lacking and does not have segmented memory.  Memory depth on the Keysight and Siglent are acceptable for basic use however.  The Rigol has much better memory depth and the need for segmented memory is reduced.

Masking

The Keysight has a sophisticated masking feature which detects excursions outside a preset mask and keeps statistics.  The feature was not explored deeply here but instead a simple example was set up by generating a Sine wave with the WaveGen and placing a mask around it.  Noise was then added to the Sine wave sufficient to occasionally cause an excursion as shown below.

Persistence is turned on to infinite when doing the mask test.  In this example 3.34451 M tests were run with 421 failures for a failure rate of 0.0126%.  The places where excursions into the mask occurred can be seen as little red marks at the top and bottom of the trace.

The Rigol has a mask feature with adjustable horizontal and vertical excursion around an input signal as demonstrated below.  The input signal and mask settings differ from what is demonstrated on the Keysight.

Summary: Masks are nice for checking timing errors, certain glitches, and other excursions.  More complicated masks can be created on the Keysight than what is demonstrated here.

Serial Protocol Decoding

The Keysight has optional software for decoding serial.  The triggering is hardware based rather than software based post processing.  Several of the training signals were observed but these features were not examined closely.  The Rigol also has serial decoding.

Summary: This feature was not tested as use is not planned. A separate logic analyzer connected to a PC with software such as sigroc is not expensive and has greater memory depth and a much larger display for visualization.

Voltmeter and Frequency Counter

The Keysight has an integrated 3-digit voltmeter and 5-digit frequency counter inside the oscilloscope.  They operate through the probes but measurement is decoupled from the oscilloscope triggering.  AC RMS, DC, and DC RMS can be measured. Here it is measuring a 1 V peak to peak 1 kHz Sine wave generated by the Keysight wave generator.

After warming up a Tenma 72-1020 bench multimeter capable of measuring true AC RMS it reported 0.3531 V and 1.000 kHz on the same signal.

Summary: The voltmeter and frequency counter worked as expected and may prove useful in circumstances where higher precision and accuracy is not required.

Bode Plot

To test the Frequency Response Analysis (FRA) features of the Keysight a RLC Band Pass Filter was built on a breadboard with the values shown on the following schematic.

There are a number of free tools on the internet for doing frequency analysis.  A tool by Okawa generrates the following for the RLC filter on the breadboard when frequency limits are set to 10 kHz and 100 kHz.

The center pass frequency is calculated to be 28.6 kHz.

It is very easy to do Frequency Response Analysis on the Keysight.  Output from the WaveGen is connected to the input of the DUT along with the Channel 1 probe.  Channel 2 is connected to the output of the DUT.

The Analysis menu contains settings for the minimum and maximum frequencies to be tested along with the number of points per decade.  It would be nice if the minimum and maximum settings were not so coarse (they are in decades starting at 10 Hz and going to 10 MHz.  Why not 20 MHz?).  As well, the maximum number of points per decade is 50 which is also somewhat coarse.

Once set up the analysis is easily run and gives gain and phase as a function of frequency.

The markers can be moved to any recorded frequency point along the curves.  At approximately 0dB the frequency is 28.8 kHz measured which compares well to 28.6 kHz calculated above.  Other values of gain and phase at various frequencies are also close to the calculated values.

Summary: This is a unique feature for the Keysight compared to the Rigol and Siglent which elevates it beyond basic capabilities.  It would be nice if the settings were not so coarse and the full 20 MHz of the WaveGen was available.

Conclusion

This review applies solely to the oscilloscopes in the report.  In other words, do not apply the observations to other models from the same manufacturer without researching further.  And under no circumstances should these observations be taken to apply to a manufacturer in general.  Some of the observations are subjective but they reflect my experience with the instrument.

All of the oscilloscopes in this review perform basic functions well and are useable for most student and electronics enthusiast activities.  They are also capable of basic 100 MHz duties that might be undertaken in a working environment.

The following table summarizes the observations in a very coarse way.  The review itself contains much more information and insight.

Comparison Summary

N/A means feature is not available

Comments and corrections are always welcome.

Correction: 15 July 2019 added Rigol mask feature and corrected Summary table