Magnova oscilloscope

[KungFuJosh]

Hello again,

without beeing able to contribute anything to your decoder discussion - another question came up in the last days:
Is it possible to substitute the reference-labels "R1", "R2" and so on with the user-label? I can only set the label in the reference menu, but can't display it on the screen (same goes for the Channel labels).
So that in the picture I attached, "R2" is substituted with the label i set in the reference-menu.

There's an option to Always Show Labels, or something to that effect.

[kaspich]

Andre and team – thanks for the excellent concept/product!
Overall build quality – perfect! Quality of screen and frontend – impressive!

My shortlist of bugs and recommendations after 2 days of quick testing.
1.   Probes are not correctly identified at startup. If a 50 Ohm termination was set previously, then a 1:10 probe was connected while the scope was off, the probe was identified (1:10 attenuation is set properly), but the 50 Ohm setting remains active;
2.   Some problems when playing with Time Scale (Roll mode was enabled). I managed to re-create a previously mentioned (but marked as fixed) bug when the trace doesn't appear in Roll mode. The problem seems to have disappeared when switched back to Normal mode, but most interesting – Offset calibration data were ignored. The problem disappeared after the restart. The log file was stored.
3.   The FFT module is crashing and freezing in all possible ways. Hopefully, it's a demo release and further testing will be performed (including: logarithmic frequency scale results in freezing; settings of the vertical axis do nothing; frequency labels are not visible, etc.).

Some recommendations:
1.   Please add a time stamp to the name of the log data file. Otherwise, if/when someone is trying to save more than one log file per day, the file name needs to be modified manually.
2.   Please modify the confirmation menu of the "Quicksave" action. When Quicksave is pressed, nothing happens for around a second or even more, then an info message appears at the bottom of the screen. It's not a Quicksave, it's a Slowsave with some disturbing waiting.
3.   "ff/fr" functionality of knobs could be more aggressive (as I see, a 4X multiplier is used when the knob is turned quickly), or even a 2-step (with different multipliers) option could be implemented. These knobs are performing a million times better than on cheap scopes, but still – my favorites remain knobs on old Agilent 6000/7000 series;
4.   BW limitation warning and frequency values (according to the selected bitrate) are missing for HiRes mode;
5.   "undo" option would be awesome. For example, you accidentally moved the trace of channel X, or moved the cursors to the wrong position, etc. A reverting/undo option is very useful. 
 

[KungFuJosh]

Welcome to the site. What firmware version is installed on your scope? This is standard to include when suggesting potential bugs exist.

Thanks,
Josh

[rf-messkopf]

3.   The FFT module is crashing and freezing in all possible ways. Hopefully, it's a demo release and further testing will be performed (including: logarithmic frequency scale results in freezing; [...]

I already reported the crash issue with the log frequency scale to Batronix before Christmas. They were able to reproduce it and said it will be fixed in the next update. That might actually be some regression issue, because I could swear that I used the log frequency scale before with settings that crash now in version 1.7.1.

settings of the vertical axis do nothing

I never encountered that bug. Can you provide some more details?

frequency labels are not visible, etc.

Do you mean the grid axis labels? There is an option (Display -> Label Size) to control their size.

[Andre77]

Andre and team – thanks for the excellent concept/product!
Overall build quality – perfect! Quality of screen and frontend – impressive!

Thank you very much - we really appreciate the kind words and the feedback.

1.   Probes are not correctly identified at startup. If a 50 Ohm termination was set previously, then a 1:10 probe was connected while the scope was off, the probe was identified (1:10 attenuation is set properly), but the 50 Ohm setting remains active;

Thanks for pointing this out. When a probe is connected, both settings - termination and probe attenuation - should be updated automatically, even if the oscilloscope was powered off at the time. We’ll take a closer look at this behavior.

In general, the settings applied via the probe’s readout pin can always be adjusted manually if needed, so they’re not locked in and can be overridden at any time.

2.   Some problems when playing with Time Scale (Roll mode was enabled). I managed to re-create a previously mentioned (but marked as fixed) bug when the trace doesn't appear in Roll mode. The problem seems to have disappeared when switched back to Normal mode, but most interesting – Offset calibration data were ignored. The problem disappeared after the restart. The log file was stored.

To investigate this properly, we’ll need a bit more information, as we haven’t been able to reproduce the issue on our side so far. Could you please confirm which firmware version you’re running? I assume it’s the current v1.7.1, but it would be good to double-check.

3.   The FFT module is crashing and freezing in all possible ways. Hopefully, it's a demo release and further testing will be performed (including: logarithmic frequency scale results in freezing; settings of the vertical axis do nothing; frequency labels are not visible, etc.).

In v1.7.1 there is unfortunately a known bug in the FFT module that can cause crashes when the logarithmic frequency scale is used in combination with certain other settings. We’re already working on this and plan to release an update within the next few days that fixes the issue.

Aside from this FFT-log-scale related problem, the Magnova should behave stably in normal operation.

Regarding the “settings of the vertical axis do nothing”: could you clarify what behavior you’re expecting? When switching between dBm, dBV, dBmV, and dBµV, the visual appearance of the FFT trace itself does not change - only the displayed values and units do (grid labels, FFT cursor readouts, etc.). That is correct and intentional. Or do you mean something else?

As for “frequency labels are not visible”: labels are currently drawn only where there is enough space between adjacent grid lines. This always works well on a linear frequency scale. On a logarithmic scale, especially over very wide ranges (for example, 1 µHz to 1 GHz), the grid lines can become so dense that no labels fit at all. We’ll revisit this and make sure that at least the major (×1/10/100) grid lines are always labeled, even if that means labels overlap slightly.

Some recommendations:
1.   Please add a time stamp to the name of the log data file. Otherwise, if/when someone is trying to save more than one log file per day, the file name needs to be modified manually.

Good point — that makes sense. We’ll add a time component to the default log file name so multiple logs can be saved without manual renaming.

In the meantime, you can already customize the file name pattern locally. For example, setting it to something like logs_{date}_{time}.log will automatically include the time in the file name.

2.   Please modify the confirmation menu of the "Quicksave" action. When Quicksave is pressed, nothing happens for around a second or even more, then an info message appears at the bottom of the screen. It's not a Quicksave, it's a Slowsave with some disturbing waiting.

The Quicksave action itself isn’t faster or slower than saving via the Save / Load menu - there’s no additional delay built in. The message “Quicksave performed...” is shown after the save operation has completed, which is why it appears with a short delay. Saving a screenshot typically takes about 1–2 seconds. If the quicksave save operation involves larger amounts of data (for example, when Save Waveform is enabled), a progress bar is shown so the process is visible.

That said, would you prefer more immediate feedback when pressing the Quicksave button, such as an instant visual or status indication that the save has started?

3. "ff/fr" functionality of knobs could be more aggressive (as I see, a 4X multiplier is used when the knob is turned quickly), or even a 2-step (with different multipliers) option could be implemented. These knobs are performing a million times better than on cheap scopes, but still – my favorites remain knobs on old Agilent 6000/7000 series;
4. BW limitation warning and frequency values (according to the selected bitrate) are missing for HiRes mode;
5. "undo" option would be awesome. For example, you accidentally moved the trace of channel X, or moved the cursors to the wrong position, etc. A reverting/undo option is very useful.

Thanks for the suggestions - all good points. We’ve noted them and will take them into account for future improvements.

[Gadolinium]

There's an option to Always Show Labels, or something to that effect.

Thank you very much for the tip, it is under Display->Marker Label

[kaspich]

yes, I am ruuning 1.7.1
About frequency labels in the FFT module - they are somewhere here (marked with red in the picture attached). Or, maybe, not
About non-responsive values/settings (FFT) - all issues disappeared after initial settings were restored and the linear scale was selected. So, hopefully, it's a well-known issue with the logarithmic scale.

[Andre77]

yes, I am ruuning 1.7.1
About frequency labels in the FFT module - they are somewhere here (marked with red in the picture attached). Or, maybe, not

Got it - thanks for the clarification and for pointing this out in the screenshot. With the FFT trace overlaying the grid, the axis labels can indeed become hard to read. We’ll discuss possible improvements to make this clearer. As a simple workaround for now, slightly shifting the FFT offset upward can help make the labels more readable. You’ve probably already noticed this, but I wanted to mention it for completeness.

About non-responsive values/settings (FFT) - all issues disappeared after initial settings were restored and the linear scale was selected. So, hopefully, it's a well-known issue with the logarithmic scale.

That matches what we’re seeing. We’ve already identified and fixed the root cause related to the combination of the logarithmic FFT frequency axis with other settings. Unfortunately, this bug made it into firmware v1.7.1.
We’ll be releasing an update in the next few days that resolves this issue and includes a few additional minor improvements. Thanks also to rf-Messkopf, who reported the problem already just before our Christmas break - that helped us track it down quickly.

[john74]

Hi again,

another feature reqest from me would be an improvement of the mask test.

At the moment you can setup a mask (from a given signal) and count the pass/fail events.

I think it would be a nice feature if the trace that raised an error would be more visible.
When you not STOP an every error, you are not able to "see" afterwards what event caused the error,
if this event is a rare glitch.

maybe change the current mask color from blue to light grey
(because blue is also the default color from CH3)
and store all waveforms that fails the test with 100% persistance on the screen.
(the existing reset button should clear the screen at any time)

I am not sure how the mask test is implemented at the moment but
a cool feature would also be to draw a random rectangle as "forbidden" area
instead of using a mask that is generated from an existing signal.

[john74]

here is a minor bug from version 1.7.1:


when the table mode of the serial decoder is enabled
and you click at one of the table mode elements to relocate the time offset
the time-stamps of the x-grid disappears.

[MikeP]

I’d like to take a moment to share some positive thoughts about this instrument. I must admit, I only opened the box about a week or more after receiving it — it was a birthday gift to myself. My initial thought was, "Back to the R&S RTB2k again?" However, within ten minutes, my opinion changed radically. I realized what a magnificent piece of engineering I was holding. Knowing that this platform will continue to evolve only confirmed I made the right choice. It has been a month now, and everything is excellent. Kudos to the development team!

 Probe Cable Management. I want to share my solution for the side-facing BNC connectors. Using right-angle adapters turned out to be a great fix. It freed up desk space, and the probe routing is much more convenient compared to a traditional front-panel layout.

 A quick note for potential buyers: pay attention to the stock probes and their market value. I am familiar with and use Testec probes, and the choice of probe model for this scope is excellent. Thank you again.

 Finally, I have a few suggestions for future updates.

  Please consider reducing the number of clicks required to access certain functions. For example, the settings "gear" icon for the Bode Plot menu could be duplicated to the right of the Cursor A and B buttons...

  Please allow saving X-Y results to "Reference" while X-Y mode is active.

 Currently, changing the file name requires digging deep into the menu to avoid overwriting the previous shot. It would be great to have automatic incremental numbering for screenshots. It might be better to exclude the system status/saving notification bar from the captured screenshot itself.

 BR
 M.

[KungFuJosh]

Currently, changing the file name requires digging deep into the menu to avoid overwriting the previous shot. It would be great to have automatic incremental numbering for screenshots. It might be better to exclude the system status/saving notification bar from the captured screenshot itself.

File saving, including screenshots, has advanced naming features, and they're usually auto incrementing by default (assuming they were unmodified).

In most places that there's an option to save a filename, you can click where it says date, time, or counter in addition to any words you want included by default. For example, for a specific project:
ProjectName {date} {counter}

By default I use:  magnova {date} {counter}, which gives me naming such as the attached.

Click the file name line in the path selection popup to change the file naming settings.

Thanks,
Josh

[Andre77]

Hi everyone,

today we are releasing Magnova firmware version 1.7.2.

Thank you for the continued feedback and the many suggestions posted here on EEVblog. We really appreciate the input and actively take it into account when planning and implementing future firmware updates. Some of the improvements in this release are also based directly on ideas and feedback from forum members.

This is a maintenance-focused release. While it does not introduce major new features, it includes several important fixes and improvements. Rather than waiting for the next feature update, we decided to publish this update in between to address these issues as soon as possible.

Optimizations:

• The Bode Plot phase color can now be selected independently.
• The generator duty cycle can now also be set for swept square signals.
• Improved visibility of the pressed button state on buttons with nearly black backgrounds.
• Increased maximum UART decoder baud rate from 9.984 Mbit/s to 100 Mbit/s.
• The right mouse button can now also be used within the Control selection to toggle between X and Y cursors.

Bugfixes:

• Fixed an FFT crash that occurred with certain settings in combination with the logarithmic frequency scale.
• Fixed a protocol error in USBTMC, which led to timeouts in some implementations.
• Fixed an issue with digital channels when using zoom.
• Fixed intensity grading in the light theme.
• Fixed an issue where the newly introduced trigger source "Generator" wasn't saved.

[ZPeterZ]

A question about clipping

For some tests on an audio device I first played around with the internal AWG and FFT to get to know the limits. Found harmonics are more than 60dB down and noise is about < -80dB. While focussing on the FFT I changed the AWG sine levels and forgot the waveform display above. Suddenly I got some comb generator like FFT display because the signal was clipped. The generator levels were 100mV and 500mV rms.
Clipping can also be produced easily while changing the generator load, because then the output level is automatically adjusted.

The following question arised:
 - When clipping occurs in the input amplifier, is it somehow protected?
 - Or does clipping at these levels (<= 500mV rms) occure in the digital domain?

And for the FFT scale I probably missed how to adjust the reference point to 0 dB (which may be adjusted to some peak) and name the subsequent devision labels in dB. For work in the LF domain this would be very nice. Any hint?

Software is 1.7.2

Thanks,
Peter

[ZPeterZ]

I attached a screenshot of the clipped 100mV sine wave.


It looks like WYSIWYG - the whole Y-resolution is spended to the viewable waveform window and the FFT can only use this data(?). Then I would have to keep this in mind when minimizing the waveform window.

[KungFuJosh]

It looks like WYSIWYG - the whole Y-resolution is spended to the viewable waveform window and the FFT can only use this data(?). Then I would have to keep this in mind when minimizing the waveform window.

Yes, the visible waveform window is still important. You need to adjust the scope frontend as you normally would to get the best results. The waveform you see there is what's being analyzed.

[KungFuJosh]

Hello! Can you please share the sig gen settings from the waveform in this Magnova post?

https://www.eevblog.com/forum/testgear/magnova-oscilloscope/msg6164699/#msg6164699

Thanks!

See attached.

Thanks,
Josh

[rusoaie]

Hello! Can you please share the sig gen settings from the waveform in this Magnova post?

https://www.eevblog.com/forum/testgear/magnova-oscilloscope/msg6164699/#msg6164699

Thanks!

See attached.

Thanks,
Josh

Thanks!
Even ChatGPT concluded you have an 15GHz oscilloscope after analysing the screenshot 
Congrats, you managed to confuse the AI 

https://chatgpt.com/share/6971096e-b898-8006-b729-28dfd2cfde50

[KungFuJosh]

Even ChatGPT concluded you have an 15GHz oscilloscope after analysing the screenshot 
Congrats, you managed to confuse the AI 

That's a really low bar! 

[ZPeterZ]

Yes, the visible waveform window is still important. You need to adjust the scope frontend as you normally would to get the best results. The waveform you see there is what's being analyzed.

Thanks, that all now seems logical to me. I wonder if in cases of rare clipped peaks which one might overlook, especially while focussing on the FFT Window - would it be overkill to the scope software to display some overload warning? This could prevent misinterpreting a corrupted FFT :-)

At this point it is time to send a big thank you to the community here and the Magnova team, making the BMO better and better!

- Peter

edit: replaced "crashed FFT" by "corrupted FFT"

[KungFuJosh]

Thanks, that all now seems logical to me. I wonder if in cases of rare clipped peaks which one might overlook, especially while focussing on the FFT Window - would it be overkill to the scope software to display some overload warning? This could prevent misinterpreting a crashed FFT :-)

It does show in overload warning- look at the peaks in the main waveform window.

Kidding aside, that's the right way to do it. If I'm running a sweep, I'll make sure the max amplitude doesn't clip before I reduce the main window size.

[ZPeterZ]

It is just because the FFT on the BMO is so much better than that of my other scopes. As a consequence I use the SA only for higher frequencies now...

[Andre77]

For some tests on an audio device I first played around with the internal AWG and FFT to get to know the limits. Found harmonics are more than 60dB down and noise is about < -80dB. While focussing on the FFT I changed the AWG sine levels and forgot the waveform display above. Suddenly I got some comb generator like FFT display because the signal was clipped. The generator levels were 100mV and 500mV rms. Clipping can also be produced easily while changing the generator load, because then the output level is automatically adjusted.

The following question arised: When clipping occurs in the input amplifier, is it somehow protected?

The input is protected within the specified limits, which are well above the 100–500 mV RMS you mentioned. This is well within the safe operating range of the analog front end.
According to the Magnova datasheet:
- 50 Ω input: 5 Vrms, 30 Vpk max
- 1 MΩ input: 300 Vrms, 400 Vpk max, CAT I

Or does clipping at these levels (<= 500mV rms) occure in the digital domain?

Like most oscilloscopes, the Magnova adjusts the analog front-end gain so that the selected vertical range uses nearly the entire input range of the analog-to-digital converter (ADC).
For example, if you set 10 mV/div (80 mV full screen), the oscilloscope amplifies the signal so that an 80 mV signal largely fills the ADC input range. This is done to achieve the best possible vertical resolution of the ADC.

Clipping occurs in the analog front end.

And for the FFT scale I probably missed how to adjust the reference point to 0 dB (which may be adjusted to some peak) and name the subsequent devision labels in dB. For work in the LF domain this would be very nice. Any hint?

The FFT of an oscilloscope typically displays a power or voltage spectrum (e.g., dBm or dBV/Vrms) across frequency bins.

From your description, it sounds like you’re looking for a gain/loss display in decibels referenced to a user-defined level. For example, you could set a 1 Vrms reference and show everything relative to that. Conceptually, this would mean normalizing the FFT by subtracting the reference value from each bin.

The scope does not offer a dedicated “set 0 dB reference voltage” function for the FFT. However, the Magnova can subtract FFTs from each other, with the result displayed in dB. This effectively provides a relative spectrum, although this approach may be more complex than needed for simple use cases.

I wonder if in cases of rare clipped peaks which one might overlook, especially while focussing on the FFT Window - would it be overkill to the scope software to display some overload warning? This could prevent misinterpreting a corrupted FFT :-)

That’s a valid point.

Some kind of overload or clipping indicator would be helpful, especially when focusing on the FFT, as it's easy to miss an occasional clipped peak in the time domain. It's easy to misinterpret a distorted FFT if you don't notice that the ADC has been saturated.

One idea would be to highlight the upper and/or lower grid boundary in red whenever a clipping event occurs. This would provide a subtle yet effective visual warning without cluttering the display.

Thank you very much for your suggestion!

[ZPeterZ]

Andre, I thank you for clarifying!

From your description, it sounds like you’re looking for a gain/loss display in decibels referenced to a user-defined level. For example, you could set a 1 Vrms reference and show everything relative to that. Conceptually, this would mean normalizing the FFT by subtracting the reference value from each bin.

The scope does not offer a dedicated “set 0 dB reference voltage” function for the FFT. However, the Magnova can subtract FFTs from each other, with the result displayed in dB. This effectively provides a relative spectrum, although this approach may be more complex than needed for simple use cases.

I will try that. Seems that I am still scratching only the surface of possibilities implemented.

[KungFuJosh]

One idea would be to highlight the upper and/or lower grid boundary in red whenever a clipping event occurs. This would provide a subtle yet effective visual warning without cluttering the display.

Two more ideas:
1: Have an option in the measurement pane that shows min/max values and include the red highlight feature mentioned above to illustrate when clipping occurs. This could be done in a dedicated new combo  Min/Max measurement, or independently on Min and Max measurements that already exist, or using the Peak-to-Peak or Amplitude measurements that already exist...or all of them.

2: Have an optional Min/Max value overlay to optionally show those values (with the red highlight when clipped)- this would be a good alternative to the boundary highlight, and could be optional in both the FFT and waveform windows.

I think an important part to include is to be able to easily reset the clipping highlight at any time with user input (tapping any highlight, for example, should reset all references to that specific min or max overload).

All of these options would be great. Let's pile more on to the to-do list!

Thanks,
Josh