Hey everyone,
I'm hoping I can get some feedback on my thoughts and testing I've done, to help make a final decision. If I've missed something or gotten it wrong, please tell me! Similarly, if there's some other model I should definitely look at, please let me know!
Right before I started my new job at the university chemistry department in September (where I am the sole electronics tech in the workshop), the workshop got approved to buy a new oscilloscope because the existing LeCroy WaveRunner 8254-M's power supply was on the fritz. In between submitting the request for a replacement scope and me starting the job, someone in another department took a look at the WaveRunner and replaced a few caps, and since then, I haven't been able to reproduce the fault, so it seems fixed. Nonetheless, a new scope got approved, but not something quite as expensive as the 4ch, 2.5GHz, 40GS/s WaveRunner.
Because I only just started in the department, I don't really have a good feel for what types of signals I'll really be dealing with, nor what all the development activities I'll be facing. After talking to some of the professors whose research is the electronics-heavy stuff, I don't actually think the scope demands are
that extreme -- they use converters to translate extremely short pulses to longer ones, and it seems that the timing
between pulses, rather than their amplitude or duration, is the critical measurement. The one professor designs some of his own hardware -- things like ADCs, TDCs, or FPGAs, with MCUs running the show -- so I suspect some embedded development will come my way, too.
Additionally, my job involves a lot of repair and maintenance of all kinds of lab gear, from magnetic stirrers to mass spectrometers, including setups controlled by Labview or Python.
Based on that, the other scopes in the labs, and the WaveRunner being seemingly fixed, I came up with the following requirements/wish list:
- 4x 500MHz analog channels
- digital inputs for mixed-signal, for embedded development
- serial decode for embedded development
- fundamentally a reasonably well-equipped scope, options-wise, because of the uncertainty of the tasks to come
- arb generator not a requirement, since I have a separate one. Only potential advantage of an integrated one is Bode plotting.
I had the various reps bring by some candidate scopes (and in the case of R&S, bring by the MXO4, though there's no way they could fit that in my budget, and they know it!), so I tried out the following:
- R&S ScopeRider (ScopeMeter-like "handheld" scope with galvanically isolated inputs)
- R&S MXO4
- Keysight MSOX3054G
- LeCroy WaveSurfer 3104z and 4054HD
- Tek MSO24 and MDO34
I didn't request a demo unit of the R&S RTM3000, which would be a candidate, because that's what I used at my last position, so I'm already familiar with it (and can go upstairs and use it if need be).
The Tek rep was kind enough to loan me a
signal demo board, too. Its "rare anomaly" and "frequent anomaly" signals were very illustrative, consisting of repeating pulse trains (every 1.6us), with defective pulse trains (containing a spike and a runt) occurring approximately once every 800ms and 100ms, respectively. Every scope could easily trigger on either of those anomalies by using the pulse width or runt triggers. The differences were in even seeing that there
were any anomalies at all!
Here's what I've figured out so far about each candidate:
Common to all, except where noted:
5GS/s sample rate (4ch), 2.5GS/s (4ch) 4 analog channels with probe ID/power interface and switchable 50 ohm termination, 8-bit ADCs, 16 digital channels, ethernet and USB, touchscreen with mouse and keyboard support, "all the options" means digital inputs, serial decoders, analysis options, and arb generator. Sample memory given in NNN/nnn Mpts, meaning at 2 and 4 channels, resp.
Rohde & Schwarz:
RTM3000: Great scope in many ways (10 bit ADC, great FFT!), but it'd drive me nuts on a daily basis because its UI is irritatingly sluggish. Nice intensity grading. Bode plot option. 80/40Mpts
MXO4: Fantastic scope overall, with 12-bit ADC, insane waveform update rate, extraordinary FFT, big, sharp display, and a reasonably snappy UI. Nice intensity grading. But to get one of these, I'd have to sacrifice practically all options and go rather low on bandwidth.
Handled the rare and frequent anomaly tests with ease. Rotary encoder knobs are made completely of rubber, and one of them stripped its inside knurling already. 400/400Mpts. 5GS/s digital acquisition rate for all channels. Bode plot option.
ScopeRider (RTH1000): Snappy UI, nice waveform update rate, 10-bit ADC, battery-powered, with galvanically isolated CAT IV 300V inputs. 5GS/s for 1 channel, but drops to 1.25GS/s for 4 channels. No probe ID, smallest screen of all, and since it's in the "giant multimeter" form factor, not really optimized for bench use. So-so feature set (though excellent for a handheld scope). Lowest vertical range is 2mV/div, but maximum is 100V/div. No 50 ohm mode. Decent monochrome intensity grading. Decent FFT. 8 digital channels. External power supply. Handled the demo board's anomaly tests with ease. Feels really solidly built. Physical controls optimized for use with gloves. Optimized for real-time more than post-capture analysis. No arb generator option.
Common to all the R&S:
UI tends to require going through a lot of screens to find things.
Nice diagrams of things like triggers, but not super awesome online help texts.
Excellent web interface with high frame rate screen sharing and touch control.
Measurements slow down display updates, but not by too much.
Logic analyzer pods are 8-bit, so if you don't need 16 bits, you can just plug in one pod and not have those extra 8 leads flopping around.
Tek:
MSO24: Adorable form factor (other than the external power supply). UI is snappy most of the time but can get bogged down at times. Fairly light on features. Waveform update rate not impressive. Passable on the frequent anomaly test, took a while to capture the rare anomaly. Mediocre FFT. True multi-grid ability at full resolution. On-screen numeric keyboard refined.
MDO34: Nice form factor with great build quality and little touches like the feet that lock in place under the weight of the scope. Waveform update rate in normal mode not impressive. In FastAcq mode (where all features are disabled), decent on the frequent anomaly test, "meh" on the rare anomaly. Still fairly light on features. UI snappy most of the time. Nice intensity grading. Separate RF input for spectrum analyzer, FFT good. Would have to sacrifice bandwidth a bit to fit into budget. On-screen numeric keyboard a little rough around the edges.
Common to both:
Nice web interface with fast update rate on VNC connection.
UI can get sluggish when decoding serial.
Labels on voltage divisions, but not horizontal divisions.
Dual multipurpose rotary encoders (e.g. one per cursor, or one each for upper and lower thresholds).
Otherwise very light on hardware buttons, many core functions require double-tapping the screen.
Measurements slow down display updates.
Unclear what they're optimized for.
Setting sample rate and memory manually is cuckoo on these.
LeCroy:
WaveSurfer 3104z: In terms of sheer value for money, this is the winner, since they could fit it into the budget with ALL options and 1GHz bandwidth. 4GS/s. Mediocre waveform update rate. With persistence, the frequent anomaly signal took a few secs to capture, the rare anomaly quite a while. No labels on grid divisions (only an option to label the extrema). UI elements sometimes a bit small to hit with a finger, especially near the edge due to recessed display. UI often a bit sluggish, enough to annoy me.
WaveSurfer 4054HD: Superb value for money, with a lightly-used 500MHz demo unit fitting in the budget with all options. UI much snappier. 12-bit ADC, bigger display with no bezel, so on-screen targets easier to hit. Has labels on divisions. 25/12.5Mpts memory. Separate coarse/fine general-purpose knobs.
WaveRunner 8254-M (existing): Monster sample rate (40GS/s) and bandwidth (2.5GHz). Full Windows OS under the hood. UI is usually snappy. Big, sharp display. Tons of display options. Very mediocre waveform update rate: with infinite persistence, slow at capturing the frequent anomaly, almost entirely incapable of capturing the rare anomaly regardless of settings. 128Mpts memory. Unfortunately, this unit does not have the MSO option (which cannot be added after the fact) and has no serial decoder options installed.
Common to all LeCroy:
Identical UI across models: nice layout that appeals to me.
Practically nonexistent online help, other than on triggers.
Nice auto trigger level finder. '
Very much focused on signal analysis, not real-time signal viewing.
No peak or envelope acquisition modes.
So-so intensity grading.
Display persistence only in fixed steps.
Measurements slow down display updates.
Really nice serial decode display in the waveform grid.
Ability to save setups, waveform acquisitions, etc to disk and then import them to a higher model, like the WaveRunner, to use its more powerful analysis capabilities after the fact.
Poor acceleration control on rotary encoders (jumps to really fast at some point).
No web interface, just control via the comprehensive, but data-oriented, Windows-only desktop software (no live stream of the display). (I figured out after the demo units went back that there may be a web interface after all, accessible by changing the network settings on the scope. I don't feel stupid for not figuring this out before the demo units went back, since LeCroy's documentation is extraordinarily sparse, and even the LeCroy rep didn't know when I asked!!)
Irritatingly slow response to signing up for a web account, which is needed to download
anything from the company.
Keysight:
MSOX3054G: G models are the same as the T models, just with most of the options thrown in by default. Smaller display with bezel that makes it a bit hard to hit the top status bar. Second-highest waveform update rate after the MXO4, but with obviously higher display redraw (not refresh) rate. Hands-down, by a wide margin, the snappiest user interface of all, with so little lag that it usually feels like an analog scope. They say they do everything in hardware, and it shows. (The flip side is that it's a clearly "old" design in other ways.) Nice intensity grading (but 64-level monochrome only) that looks like an analog scope. Best context-sensitive online help: press and hold any button and a nice text pops up. UI is old-fashioned but easy to use, though a bit cramped on the small resolution of the display. Lots of physical controls, but also completely mouse-controllable. Measurements do not affect waveform update rate at all. Hardware serial decode. Zone trigger seems potentially really useful. Bode plot. Really lousy FFT. Web interface is complete, but screen sharing is not super fast (maybe 3fps). Desktop software is buggy. Probably the noisiest front-end. Just 2Mpts real-time memory (4Mpts single-shot). I'd really love it if the Keysight rep were able to finagle a MSOX4054G instead, simply for the larger display.
Bonus:
Rigol DS1054z: My dinky little scope at home worked approximately as well as the Tek MDO34 at making visible the two anomaly test signals. And its intensity grading is actually
nicer than the Keysight's, putting it right at the top alongside the MXO4 and MDO34. Impressive for being a bargain-basement model from 8 years ago. It also makes me curious as to how good higher-end Rigols are. Unfortunately, Rigol has not yet responded with an offer for a demo.
The upshot is, at this point, that unless R&S can come up with an insanely high discount on the MXO4, it's really down to the LeCroy 4000HD and the Keysight 3054G, with me leaning towards the Keysight. The LeCroy would have the advantage of the same user interface as, and high compatibility with, the WaveRunner, and it would bring with it hands-down the best set of options. But it's got the same strengths and weaknesses as the WaveRunner, in that it's an analytical scope. On the other hand, the Keysight would
complement the WaveRunner, in that the Keysight is more of a real-time scope that reacts instantly and makes it easier to discover the existence of anomalies. My thinking is that the Keysight would make it easier to identify what it is I want to look at, and that if I really need to analyze it in detail, I could then use the WaveRunner to do so. Similarly, the WaveRunner's FFT is pretty decent, so it could serve that purpose.
Am I missing anything?
Are there any "gotchas" with these models that I don't know about?
Am I giving too much weight to having digital inputs?
Would that be better spent on a separate USB logic analyzer like a Saleae or Analog Discovery?
Thanks for any insight or tips you may have.