R&S MXO3/04/05 - Reviews, Bugs & Wanted Features/Improvements Thread

[Daniel Bogdanoff]

I just took another look in the box to see what else was in there....

In addition to the protective cover for the front, there’s a pouch containing BNC cables and 8 probes—nice.
However, I find the build quality of these probes to be rather poor, especially the accessories, which are rather flimsy.
So there’s room for improvement, as Magnova demonstrates with its high-quality probes from Testec.
I kept looking in the box, but couldn't find anything:
The probes for the digital input aren't included—that's disappointing for an official demo unit, especially since this option is enabled....Too bad.

On the probes topic - my initial impression of the passive probes that came with the scope is that they felt pretty standard to me? I wasn't impressed in either direction and it feels like they do as well as I'd expect. I'll have to try out the Testec ones - do you have a model of them you prefer?

The digital probes are included with the -B1 option that turns on teh MXO, but I was told they aren't always included in the demo kits. One thing to be aware of with the digital probe connections is that they come out the upper right side, so you have to plan your bench space around that if you are a digital channel user on the MXO3.  On the MXO4 and MXO5 they come out the front, but there wasn't space there with the MXO3. Overall it's a pretty compact 8 channel scope so it typically won't be a problem for folks.

[Martin72]

Hi Daniel,

This is the one:

https://www.batronix.com/shop/measurement/probes/Testec-TT-HF-612RA.html

Apart from the alligator clip, which is a bit too big, the probe is top-notch in terms of build quality, and it comes with plenty of accessories as well.

[Daniel Bogdanoff]

Thanks! I'll check it out

[Bernd_2]

I’m also very satisfied with the Testec TT-HF 612RA.  It actually holds its own even when compared to the much more expensive ZP-11, which comes bundled with the MXO5.

Conveniently, some of the Testec accessories also fit the R&S ZP-11.

Here are a few measurements comparing the Testec with the R&S using the LB-1322 pulser. As expected, the ZP-11 performs better than the TT-HF in terms of overshoot.

https://www.eevblog.com/forum/testgear/rs-to-launch-mxo5/msg5915990/#msg5915990

Best Regards
Bernd

[Xiphos]

A few things: 

1. I would be interested in MXO-3 and how it compares to speed of its bigger brothers.  I didn’t notice a big difference the few times I have played a little bit with one.

2. I am also curious on the performance of the function generator on the MXO-3. I have an MXO-5 and it is a shockingly good function gen with low harmonics, etc. The one big downside is there is no external input for modulation.  It also sometimes wanders a little bit (even when I am on distribution lab reference) relative to other standalone function generators. I don’t know why.  But, it is one of the best 100MHz class function gens around.  Kinda surprising since it is just an add-on to the scope board.


Would others be interested in a group “test” of some of the simpler functions of scopes (and other test equipment).  On this forum, users of so many different models and companies are represented I wonder if it would make sense to systematically choose a few key parameters that are testable with simple equipment that most people would already have and they can post results?  That way we could do comparisons of a bunch of scopes from the R&S, Keysight, Magnova, LeCroy, Siglent, etc.

I can see in older posts similar things have been done with noise floor etc. but perhaps there is a way to make it more systematic? 

As I said, I have an R&S MXO-5 and I love it.   It is so fast and the UI is great—things that don’t necessarily come across in tests so we would need to have a subjective comment area that that covers those areas. If there are enough people interested I would even be willing to “donate” a few simple devices to test scopes (Voltaire pulse tester, maybe a function generator for square waves, etc.) that could be mailed around for others to use.

Or is this an area that won’t get traction among enough users?

[Martin72]

Hi,

Unfortunately, I won't have the MXO3 here much longer—maybe another two weeks.
Here's what I'd like to check out during that time:
- FFT function in general
- X-Y mode (“Music” test)
- Decoder functions
- Test of the internal AWG
- Waveform update rate
- Performance when all 8 channels are active
- Trigger tests
And whatever else comes to mind or I can manage in this short time.
I have the SDG6052X, SSG3032X-IQE, Batronix demo board, Siglent demo board, and various other resources at my disposal.

I have an R&S MXO-5 and I love it

  

[tooki]

@MASCH the MXO4 and MXO5 do have a more powerful processor, so if there's a level of lag that you are unhappy with I'd definitely look into getting your hands on a 4 or 5.

As someone who has an MXO4 at work, the thought of this software running on an even slower processor is kinda horrifying, given that IMHO the MXO4's GUI already skates on the knife's edge of tolerability in regard to UI responsiveness. (If I'd been product manager of it, I would have insisted on much better UI performance.)

Does the MXO5 use a faster application processor than the MXO4?

[core]

From what I've seen in teardowns, the application processor is the same for the MXO3 and MXO4. The MXO5, however, has a much more powerful processor.

Even so, the processor used in the MXO3 and MXO4 is significantly more powerful than what the SDS3000X HD has, for example.

For this reason, it seems strange that there is lag, considering how smooth the Siglent UI generally is.
It appears that the software is still far from being optimized to match the hardware capabilities.

Edit : On YT teardown from The Signal Path for MXO5 and MXO3 and teardown from Dave for MXO4 we can see that Zync Ultrascale+ XCZU7EG is used for MXO3 and MXO4, but a much large and different one for MXO5. In contrast, Siglent SDS3000X HD is using XCZU2CG if I remember correct..

[Martin72]

Zync Ultrascale+...Siglent scopes tend to be on the conservative side when it comes to such things.
The SDS7000 is, of course, an exception, but also the SDS5000X HD has more power under the hood (a total of 8 cores).

As I said, based on my experience with the MXO3 so far, it doesn't seem to be lagging much—though my tolerance level might be a bit higher than that of others...
I can well imagine that you need more power for 8 active channels than for 4.
The MXO5 is likely equipped accordingly.
If the MXO3—which is also available as an 8-channel model—is any different, it’s simply because it’s “smaller.”
After all, the appeal of larger models has to be maintained.

[2N3055]

If I may make a note. I am not lucky enough to be given one for testing to see for myself  , but from what I see on videos, it has that ANNOYING short delay and then stupid animation when opening and closing windows/menus. If only GUI designers would give that up. Originally Apple did that because it was showing where window was folded to icon. So you know. Here it serves no purpose except delay and stealing precious CPU cycles that on CPU limited devices (scopes are ALWAYS CPU limited) could be used for something useful.  Like scanning encoders at higher rate.

[Martin72]

I just noticed:

I just took a video of what I experience with the unit

No wonder I didn't notice anything; I use the encoders at normal speed, so to speak.

[skander36]

I just took a video of what I experience with the unit.

...

MXO4 do the same. It is not about the encoder reading, but about the number of operation/data processing during adjusting the timebase and displaying the waveforms. Yes you can say that at this point these scopes are slower than others.
Anyway this kind of operation is not recommended or desired.

[KungFuJosh]

I just took a video of what I experience with the unit.

I saw in your video 4us/div and 14bit HD mode, and made this video with the Magnova for comparison with the R&S scope:




Note that the Magnova has fine and coarse knob response options.

Anyway this kind of operation is not recommended or desired.

Why not? Some people prefer knobs.

Thanks,
Josh

[skander36]

...

Why not? Some people prefer knobs.

Thanks,
Josh

I mean the way he manipulates the knobs.
See Martin72 last answer.

[KungFuJosh]

I mean the way he manipulates the knobs.
See Martin72 last answer.

Did you see my video? I was rougher than he was without any issues from my scope.

Whether it's a hw or sw issue on the MXO3, I dunno. But at that price point, the knobs should work without skipping or delay. Some odd behavior like that is acceptable with a new product while they work the kinks out. But frankly, I hold R&S in high regard, as does their price point. I would hope they fix the issue.

[skander36]

I mean the way he manipulates the knobs.
See Martin72 last answer.

Did you see my video? I was rougher than he was without any issues from my scope.

Whether it's a hw or sw issue on the MXO3, I dunno. But at that price point, the knobs should work without skipping or delay. Some odd behavior like that is acceptable with a new product while they work the kinks out. But frankly, I hold R&S in high regard, as does their price point. I would hope they fix the issue.

I don't think you understand what's actually happening. The knob doesn't skip steps. When he suddenly turns the knob, all the steps taken are transmitted to the controller, but the system has to calculate all these steps, processing all the captured data to display the resulting waveform for the new timebase. I don't know how a prediction could be made. For a larger sampling, a longer time is required, plus what other internal processing needs to be done. 5 GSa/s is higher than 1.6 GSa/s so it results in a longer processing time, although I've seen that the Magnova behaves similarly at the timebase. Yes, as I said, it's not the fastest, but it's slower than other oscilloscopes (e.g. Keysight 2DSOX2K is much faster than the Magnova but that doesn't mean it's better), but I don't think any manufacturer will consider the presented case (sudden actuation of the buttons) a priority, especially since there is no delay during normal actuation.

[2N3055]

Just to note, when you change settings like that on timebase, scope actually reconfigures FPGA processing of datastream. That might take a moment, depending on what reconfig is, and what memory size and timebase it is. As for vertical, it also have to reconfig front ends, and rescale all the values etc.
Everybody likes how Keysight 3000A/T/G series are fast in responding to user. But they forget they have puny little ACQ memory, low res screens, and only 8 bits of datapath.

[skander36]

...
Everybody likes how Keysight 3000A/T/G series are fast in responding to user. But they forget they have puny little ACQ memory, low res screens, and only 8 bits of datapath.

Exactly, being faster means being "lightly"with all that this entails.

[KungFuJosh]

So you're saying the processor they chose is inadequate for the MXO3's specs?

[2N3055]

So you're saying the processor they chose is inadequate for the MXO3's specs?

No. Processor might have nothing to do with it. If you are reconfiguring FPGA you simply need to wait for it to happen.
But I don't have internal info how it really works, I am just interpolating based on general principle of how usual architectures do work. There are not many surprises there.
And if MXO3 also keeps/streams a separate copy of data for FFT for instance, data pump and memory plan is even more complicated.

I mention all this because I see everybody's a critic, and have many conclusions, based on zero data and wishful thinking.

In different topic I also saw a person demonstratively selling his brand new, in my opinion good scope, because they claim scope is at fault because they cannot be bothered to switch on an option, and claim that they expect scope to show 350MHz signals while in roll mode...

I dislike that kind of drama and frankly, spoiled children behaviour. I wish it would not pollute discussion here.

Next thing we will see that some "expert" will use cordless drill to rotate timebase  to demonstrate how some scope is bad... at continuous 1250 RPM.

Full disclosure, from what I see from the videos, i DO find MXO3 GUI and response to keys on slower side, mainly because I hate GUI animations. They are useless, deliberate slowdown of the process, a conscious design choice  by GUI designers to make product more crappy for the users because "it looks cool".
But in all honesty, it is not "I cannot do my work" slow. Perceptible delay is not prohibitive delay. Normal people doing actual work don't twiddle knobs at 120 RPM. It is same as whiny gamer kids complaining about 345 or 346 fps on some game, on 15000USD gaming PC, while on a 2000 USD PC people design satellites...

[KungFuJosh]

I totally agree with you...especially about that user who slightly reminded me of another user whom I won't mention either.

I was more responding to what Skander said:

The knob doesn't skip steps. When he suddenly turns the knob, all the steps taken are transmitted to the controller, but the system has to calculate all these steps, processing all the captured data to display the resulting waveform for the new timebase. I don't know how a prediction could be made. For a larger sampling, a longer time is required, plus what other internal processing needs to be done. 5 GSa/s is higher than 1.6 GSa/s so it results in a longer processing time, although I've seen that the Magnova behaves similarly at the timebase.

His comparison is that more processing time is required for the higher specs on the MXO3, and that logic directly relates to processing power, one way or another.

The same can be said about the time-wasting animations that you dislike. Anything that wastes time makes the scope's processing power for general tasks at least appear weaker than it should be. If there's any lag on the animations at all, then it makes it look worse.

I'm not pretending to know how the scope works better than anybody here; I know I don't.

[Rydda]

Hi all.
I just want to add a short comment to this topic:
Every time the setting on the V/div knob is changed, the MXO starts a new acquisition. So if you twirl the knob too fast it will never get the chance to complete an  acquisition, leading to seemingly missing steps etc.
Doing the same test in 'single shot' mode gives IMHO a more correct view of the of the response of the UI to the knob rotation.
/Rydda

[Xiphos]

Hi

I wonder if the lagging issue is both some firmware optimization (including the animation—I agree, not necessary) but also some FPGA or other systems (front end A/D whatever).  I also notice it on the MXO-5 but it rarely is a problem.  Kinda just not instantaneous sometimes….I only notice it when i am confused about blindly turning a knob to check which direction (up or down) I need to go and spam the knob a few clicks…but it catches up quickly.

Every scope has little issues and compromises, but I personally feel the value that R&S offers for the price in the MXO line is outstanding.  It is a great scope.   It is very, very well designed.  It is my go-to scope.  It does almost everything in its function range well. (BTW, I have mostly HP/A/Keysight stuff.)  For my preferences, the other models in similar price/bandwidth range don’t come close.  The FFT is as fast and useful as any other scope up to 2X the price in my opinion.  (And no, I don’t work for R&S and have no relationship.) 

If I had to pick on the MXO the issue I would choose be the unavoidable slowdown in acquisition from horizontal measurements…. It can be annoying…but every scope has quirks.  By the way, have we confirmed that the measurement slowdown issue also exists on the MXO-3?

In terms of testing the MXO-3, besides the measurement slowdown issue confirmation, I also would be curious about the performance of the internal waveform generator.  Harmonics, etc? It is 50Mhz, correct?  Thank you

[le_yum]

it looks they have same issues as 500$ scopes 

[skander36]

...

 By the way, have we confirmed that the measurement slowdown issue also exists on the MXO-3?

...

Wait for the tests from the thread opener although you know the results ...