Magnova oscilloscope

[KungFuJosh]

Regarding the aliasing "issue" of the Magnova, it seems the Keysight/Agilent is truly unique in this area!

So cant blame the Magnova too much

Can't blame them at all. Only 1 company allegedly has the "anti alias" thing going, and then you have to ask how. 🤷 ...which in this case is probably a topic of conversation for a dedicated thread.

I wouldn't be sad if the Magnova had something like this implemented if the data presented is real. However, if Keysight is doing something like Fnirsi would (fake waveforms), no thanks.

[2N3055]

Which one is your all around favorite?  For general use.

For general use I mostly fire up SDS6104H12. That is an early 12bit version of 6000A with only 2 ADCs. It has 12" screen and many advanced functions. Like SignalScan. It does not have Eye/Jitter function and was up to 1GHz only.
Next in line would be 3000xHD. For most people that is sweet spot. It has active probe interface and is very nice scope. Quite fast too. Has FFT magnitude and phase, decoding from math channels, memory channels etc... Very powerful.
If I were doing something similar to what Josh does, 2000xHD hits that spot very nicely. No active probe interface, but 500 MHz and still professional level scope. Very very quiet acoustically. 2000xHD and QuantAsylum 403 sorts you out for audio work 100%.

If I need to decode something, if it is not much and I need to trigger on specific events, I also use 6000, for screen alone. 2000xHD already have impressive set of decodes.
But if I need to decode a lot, I break out Picoscope. Big screen on PC makes all the difference.
But it cannot trigger on protocols at all. I use it for bulk decode. And if I need some specific thing only Picos do: 16 bit, or some math and measurements. They have flexible X/Y plots for instance. Math is very flexible too.

7000A gets attention only once we break 1 GHz.... It is not quiet. It is big. Used only for special stuff others cannot do.

And MSOX3000T gets very little game time. I use it mostly as a second opinion, or some specific thing it does. It is a nice scope though, although it shows its age...

Do you understand the dilemma now?

[kaspich]

Regarding the aliasing "issue" of the Magnova, it seems the Keysight/Agilent is truly unique in this area!



So cant blame the Magnova too much

My Siglent SDS2000 in Normal mode also performs correctly (some modulation appears in Roll mode, but far from 100%).
So, in your video, Dave checked the Siglent 1000 series, which costs around 300 EUR (?) 500 EUR (?). 1/10 of the Magnova price.
The Siglent SDS2000 price was 1/3 of the price of Magnova.
Then Dave is testing Rigol (sorry, China semicrap - just my personal opinion) with 1GHz bandwidth and a price 1/3 of Magnova's price.
Yes, the Agilent 6000 series was costly (around 10K EUR).

I am really sorry, I've used correctly working scopes (sure, there are some exceptions - MicSig for portable tasks, 16bit Pico for some specific measurements, both without high expectations) for last two decades and had no clue that in 2026, in the Luxury segment, aliasing still is a problem
Let's compare Magnova with Owon SDS210, for example, or Firsni hiend

[2N3055]

Regarding the aliasing "issue" of the Magnova, it seems the Keysight/Agilent is truly unique in this area!

So cant blame the Magnova too much

Can't blame them at all. Only 1 company allegedly has the "anti alias" thing going, and then you have to ask how. 🤷 ...which in this case is probably a topic of conversation for a dedicated thread.

I wouldn't be sad if the Magnova had something like this implemented if the data presented is real. However, if Keysight is doing something like Fnirsi would (fake waveforms), no thanks.

They are running rendering for screen in Peak detect mode always. And their triggering has randomized minimum hold off. But if you stop and magnify what was captured it is nonsense. Because Nyquist. It was discussed and nauseam before. And those that insist it is only way of doing things proclaim all other scopes crap. While other scope actually show proper thing mathematically. Keysight is full of little tricks like that.
On other scope you simply use Peek Detect, but yo have control of that.

[KungFuJosh]

They are running rendering for screen in Peak detect mode always. And their triggering has randomized minimum hold off. But if you stop and magnify what was captured it is nonsense. Because Nyquist. It was discussed and nauseam before. And those that insist it is only way of doing things proclaim all other scopes crap. While other scope actually show proper thing mathematically. Keysight is full of little tricks like that.
On other scope you simply use Peek Detect, but yo have control of that.

Thank you, that's what I suspected. And no thanks for the Fnirsi style lies.

[KungFuJosh]

Do you understand the dilemma now?

What dilemma? You have some nice tools in the toolbox. BTW- I'm gonna go out on a limb and say that's still not all your scopes.

[2N3055]

Do you understand the dilemma now?

What dilemma? You have some nice tools in the toolbox. BTW- I'm gonna go out on a limb and say that's still not all your scopes.

Ones I mentioned in response to Grandchuck? Alas no... 

[KungFuJosh]

 

[Andre77]

Hey, you step away from the forum for just over a week on vacation and suddenly all hell breaks loose with pages of new posts… 😉

First of all, apologies that Florian and I didn’t respond earlier – we were both on vacation.

I’m happy to catch up now. A lot of points have already been raised here (thanks to everyone for that), but I’d still like to add my perspective as a representative of Batronix.

An additional bug was identified while testing my generator (my unit works fine).
Generator = ON. F = 50Mhz. Amplitude: 1Vpp
HiRes = OFF
Changing horizontal resolution:
20us and less = everything is fine
20us = amplitude of the signal is jittering
50us = only the DC offset (jittering) is on the screen
100us = jittering amplitude, etc.
In Roll mode:
50ms = jittering DC offset only
100ms = ok
200ms = only DC offset.

Thanks for taking the time to write this up and for sharing your detailed observations. I’m sorry to hear that your experience so far hasn’t been a positive one – that’s obviously not what we want.

However, something caught my attention in your original post above and in several of the follow-up screenshots:

In many of the screenshots, the memory depth was manually limited to 100 kpts. Given the time scales used, this naturally results in a significantly reduced sampling rate and thus undersampling for the signal frequency used. With the default setting (Auto Memory), the sampling rate would remain significantly higher - up to about 200 times higher. Using the “Auto (Max)” memory option would even increase it by a factor of up to 3200.

You likely set the memory depth intentionally and had your setup under control, but anyone quickly skimming through the posts might easily overlook this as the root cause of the low sampling rate and, consequently, the aliasing effect. The effect itself is well known; several videos from various manufacturers have already been shared, so I won’t go into that in detail again.

The default setting, Auto-Mem, would already solve the problem in most of the cases shown here (except in Roll Mode).

For roll mode - and for situations where the aliasing effect still appears because the sampling rate is below twice the signal frequency - the solution is to use the Peak Detect acquisition mode. This operates at the full sampling rate and effectively eliminates the issue.

2. Why is my scope dead? ONE wfps (without trigger)? Seriously? If/when I connect the probe to my PCB, I expect an immediate reaction from my scope. Ok, 10..20..30ms delay is acceptable. That means 20..30 wfps at least. Why does my scope turn to total crap at low time scale settings?

Could you share which settings you were using in that situation?

For example, did you have a fixed memory depth configured? That can significantly extend the acquisition time, as the scope waits until the defined number of points is filled. Also, was “extended capture” enabled in the Acquire menu?

3. hires mode. What do you expect? A4-sized paper sheet near my scope, where all BW limitations are calculated for each samplerate/bitrate settings? Ok, when selected, why is max averaging (that doesn't affect BW because of slow sample rate) not added? Or should users change these settings every time they change the time scale?

I’m not entirely sure if I’m interpreting your point correctly, but I think there might be two things mixed together here. HiRes and Average may look similar at first, but they serve different purposes: HiRes works within a single acquisition (useful for slower non-repetitive signals), while averaging operates across multiple acquisitions and only makes sense for stable, repetitive signals. Automatically enabling maximum averaging together with HiRes could therefore easily lead to misleading results depending on the signal.

4. Existing ''peak detection'' in Roll mode doesn't solve problems, mentioned in p.1 (some of the forum users mentioned this as a solution).

I’d be very interested in your exact setup in that case, as this should normally not occur when using Peak Detect. Perhaps you could share a screenshot or provide details of your setup with me or our service team?

If anyone is interested (350MHz + generator), please PM. Location: EU. Purchased one month ago.

Just to mention it: if you decide the Magnova isn’t for you, the return timeline shouldn’t be an issue - we’re quite flexible on that. You’re welcome to return it to us and we’ll refund your purchase, even if the 30 days have already passed. Just reach out to us directly by email.

We can also continue discussing this here or directly via email - whatever you prefer. I’d be happy to take a closer look and see if we can address your concerns.

[Martin72]

And I personally would rather get SDS3104xHD for a bit more money than Magnova. Scope wise it is more powerful scope in several aspects. From BW to math, analysis functions etc.
But if you do not need all that, and 350MHz BW is enough for you, and you want that "luxury" feel, than Magnova is scope for you, if you can afford it.

Luxury is something that isn't really necessary.
From that perspective, anything over €500 would be considered luxury in that segment these days.
Incidentally, the excellent build quality—which is second to none—is simply a given and is unlikely to be a deciding factor when purchasing a scope.
If it is, then you probably don't need one.

[Andre77]

By the way, has it been officially confirmed that the next firmware update will include a hardware counter?
I’d also be interested in zone trigger - will they be implemented soon, or will we have to wait a little longer?
Also, is it possible that the generator will include a PWM signal?
Otherwise, Magnova is a promising platform. The price might be a bit high, but maybe there will be promotional offers...

Yes, the upcoming firmware update will include both the hardware counter and DVM functionality. Both operate independently on all ADC samples, always at the full sample rate and with very high (adjustable) resolution. The update is expected to be released next week.

Zone trigger is already planned, but it will take a bit more time, as there are a few other features with higher priority that are being implemented first.

PWM modulation is not yet supported by the generator module. We may be able to add this feature in the medium term, but that has not yet been decided.

[mawyatt]

Hey, you step away from the forum for just over a week on vacation and suddenly all hell breaks loose with pages of new posts… 😉

First of all, apologies that Florian and I didn’t respond earlier – we were both on vacation.

I’m happy to catch up now. A lot of points have already been raised here (thanks to everyone for that), but I’d still like to add my perspective as a representative of Batronix.

An additional bug was identified while testing my generator (my unit works fine).
Generator = ON. F = 50Mhz. Amplitude: 1Vpp
HiRes = OFF
Changing horizontal resolution:
20us and less = everything is fine
20us = amplitude of the signal is jittering
50us = only the DC offset (jittering) is on the screen
100us = jittering amplitude, etc.
In Roll mode:
50ms = jittering DC offset only
100ms = ok
200ms = only DC offset.

Thanks for taking the time to write this up and for sharing your detailed observations. I’m sorry to hear that your experience so far hasn’t been a positive one – that’s obviously not what we want.

However, something caught my attention in your original post above and in several of the follow-up screenshots:

In many of the screenshots, the memory depth was manually limited to 100 kpts. Given the time scales used, this naturally results in a significantly reduced sampling rate and thus undersampling for the signal frequency used. With the default setting (Auto Memory), the sampling rate would remain significantly higher - up to about 200 times higher. Using the “Auto (Max)” memory option would even increase it by a factor of up to 3200.

You likely set the memory depth intentionally and had your setup under control, but anyone quickly skimming through the posts might easily overlook this as the root cause of the low sampling rate and, consequently, the aliasing effect. The effect itself is well known; several videos from various manufacturers have already been shared, so I won’t go into that in detail again.

The default setting, Auto-Mem, would already solve the problem in most of the cases shown here (except in Roll Mode).

For roll mode - and for situations where the aliasing effect still appears because the sampling rate is below twice the signal frequency - the solution is to use the Peak Detect acquisition mode. This operates at the full sampling rate and effectively eliminates the issue.

2. Why is my scope dead? ONE wfps (without trigger)? Seriously? If/when I connect the probe to my PCB, I expect an immediate reaction from my scope. Ok, 10..20..30ms delay is acceptable. That means 20..30 wfps at least. Why does my scope turn to total crap at low time scale settings?

Could you share which settings you were using in that situation?

For example, did you have a fixed memory depth configured? That can significantly extend the acquisition time, as the scope waits until the defined number of points is filled. Also, was “extended capture” enabled in the Acquire menu?

3. hires mode. What do you expect? A4-sized paper sheet near my scope, where all BW limitations are calculated for each samplerate/bitrate settings? Ok, when selected, why is max averaging (that doesn't affect BW because of slow sample rate) not added? Or should users change these settings every time they change the time scale?

I’m not entirely sure if I’m interpreting your point correctly, but I think there might be two things mixed together here. HiRes and Average may look similar at first, but they serve different purposes: HiRes works within a single acquisition (useful for slower non-repetitive signals), while averaging operates across multiple acquisitions and only makes sense for stable, repetitive signals. Automatically enabling maximum averaging together with HiRes could therefore easily lead to misleading results depending on the signal.

4. Existing ''peak detection'' in Roll mode doesn't solve problems, mentioned in p.1 (some of the forum users mentioned this as a solution).

I’d be very interested in your exact setup in that case, as this should normally not occur when using Peak Detect. Perhaps you could share a screenshot or provide details of your setup with me or our service team?

If anyone is interested (350MHz + generator), please PM. Location: EU. Purchased one month ago.

Just to mention it: if you decide the Magnova isn’t for you, the return timeline shouldn’t be an issue - we’re quite flexible on that. You’re welcome to return it to us and we’ll refund your purchase, even if the 30 days have already passed. Just reach out to us directly by email.

We can also continue discussing this here or directly via email - whatever you prefer. I’d be happy to take a closer look and see if we can address your concerns.

This entire post just shows Customer Service and Customer Satisfaction at Batronix is a priority and not just lip service

Hats off to Batronix

Best

[core]

I'll come back to quality a bit, because I would consider the following details as qualities, not luxuries:

- quality encoders and micro-switches. Just a few, but good ones.
My DHO1000 has optical encoders, but they are not very useful, in the sense that quite often it does not take into account the pulses even at very low rotation speed. After which it skips two positions on the next movement. Siglent has some indexing problems, especially with the indexing button for both the SDS2000X Plus and the SDS1000X-E. Both Rigol and Siglent have no malfunctioning problems, but they do not work very smoothly and precisely. They can be used, but a precision encoder is clearly quality, not luxury.

- large screen, high resolution and optical quality - 100% is not a luxury. You can use an external screen if the device allows it, or access with VNC, but a good screen and a high resolution are worth the money.

- "no tantalum or electrolytic capacitors" for me is a guarantee that it will not have problems in the long term. Electrolytic capacitors have always been a nuisance for any electronic device.
I always try to find a teardown to see how it is built. Among other things, if I see a lot of electrolytics it means that the voltage sources need help because they are noisy, or there are problems with the power paths and interference appears everywhere.

- ref-in, ref-out : you never know when you need it, it is good to have them. DHO1000 has them, although it is a cheap oscilloscope, so it is not a luxury

- internal generator at 90Mhz, with 20Vpp : it is clearly a quality

- channe-to-channel isolation 60-70dB
- bandwidth filters for 20, 50, 100, 200Mhz combined with low-noise input stage helps a lot for weak signal analysis. Likewise, RF features mean attention to detail design (dynamic range etc).

- the metal case may be a luxury, but I really like the idea of ​​passive cooling as much as possible

From a hardware point of view it is clearly a quality device.

That said, although there is little chance that someone will open a device with a 5 year warranty, a teardown would be very nice. Just a few photos of the inside 

[rf-messkopf]

- the metal case may be a luxury, but I really like the idea of ​​passive cooling as much as possible

That decision is likely driven by economy of scale reasoning, and folded and laser-cut metal plus powder coating certainly involves significantly lower risk. With a production run of perhaps 100-300 units per year, a metal case might even be cheaper overall. An injection mold for two case halves of approximately 40x25 cm will cost a significant five-figure sum. If there are any penetrations that are not aligned with the opening direction of the mold, like ventilation slots or for connectors on the side, it will be even more expensive. And you cannot make any changes after the mold is manufactured. What if you realize after doing 100 copies that a connector should be moved 2 cm further to the left?

Anyway, for me the metal case and passive cooling is a plus, and it certainly contributes to the quality of the device.

[kaspich]

Andre/team, thanks for the reply. Also, thanks for returining proposal. If it's acceptable to you, I'll make my decision within this week, depending on the future perspective. Some of the problems/issues/bugs (for me - bugs/issues) can be fixed easily (my expectation). So, if I am the only strange guy who sees these problems as problems, I'll return my device, and you will forget an unsatisfied customer. On the other hand, if my recommendations are useful, they may increase your sales. So, this will be a win/win situation.

Let's start with very simple things. All we know - Magnova is a fast/responsive scope. Let's check the situation outside the sweet spot!
Please see the attachment. I am lazy, I didn't use a calculator, Excel formulas - hopefully, if there are no math mistakes.

Agilent 6000 series (20-year-old scope); Siglent SDS2000 series and cheap portable MicSig TO1000 series (both around 10-year-old models) were compared.


Magnova settings:  hires is OFF, memory options: fast and 300M points.
Normal mode, untriggered (anyway, triggered/untriggered doesn't make a fundamental difference).

Data for Siglent were captured for two memory settings: 140M points (slow response) and 140K points (fast response).
MicSig (14M points per channel) and Agilent have ''automatic'' memory management. Agilent has an 8M points memory license.

Honestly (for Magnova), I have no clue what ''coverage'' means (appears when you press the WPFS ''button''), but definitely, not what I expected...

For me, the picture is catastrophic. Around/at least 99% of the time, Magnova is ''blind''! Why??? Reducing the time scale, things are going worse and worse...
1..2..4 wfps refresh rate: worst result in this test, comparable with 50 USD China-made scopes...
Ok, if it's a feature of ''modern digital oscilloscope'', please let me know.

Also, please pay attention to Agilent. Scope of 2005, but have 80..90% of coverage (with coverage I mean ''uptime'').

p.s. ok, let's check the triggered situation (that's a modern digi... ok, you know what I mean). 10MHz signal at the input, trigger level: 80% of peak value. Memory set to ''fast''.
wpfs increased up to 7 @ 1ms (which is less than 10%) and up to 39 @ 100us (which is around 5%). So, 90 .. 95% of the time, Magnova is ''off''. Still horrible results!

[core]

Can you tell us the conditions under which the tests were performed?

- time base (we have it)
- signal frequency
- what is considered blind time? maybe a screen-capture with the measurement?
- the memory setting does not seem to be uniform. From what I know, Siglent takes into account the memory depth required for a screen for acquisition (14 cells in your language from the table). For Magnova it appears that you set 300MPts ... in this case the settings are completely different.

I could perform the tests with Siglent 2000 X Plus and Rigol DHO1000 for comparison, but the conditions should be similar for all oscilloscopes.

[john74]

Magnova settings:  hires is OFF, memory options: fast and 300M points.

if you force the memory depth to 300M points/ch, the waveform update rate can't be higher than 5.3 wfms/s
regardless of the time scale

formula : updateRate = sampleRate / memoryDepth

[2N3055]

I honestly do not understand what numbers are you showing here.
But I know those numbers are dead wrong.

Also I think you are confusing the memory management modes with trigger modes.
There is Auto and manual memory modes, and there are trigger modes: Auto, Normal and Single.

Did you seriously proclaim that triggers per seconds are measure of how often scope triggers in Auto trigger mode?
What do you mean triggered/untriggered makes no difference. In Normal trigger mode?
In Normal trigger mode without signal producing trigger event, scope will not trigger and capture anything at all.
In Auto trigger mode, scope will acquire occasional screen refresh, based on some timeout. It is not going as fast as it can, but will deliberately wait to see if signal will actually trigger something, and only if that does not happen for a very long time, it will capture something randomly just to refresh screen.

Also, again, memory. You put memory in MANUAL 300 MPts mode. Scope will capture 300 MPTS always, including where you are in 2ns/div and there are only 5 points shown on the screen.
So if you are at At 1.6 GS/s it will take 0,1875 seconds per every trigger to capture all that data. That is 5.3 triggers per second ideally, if scope has no blind time at all.

All your other measurements are probably done using same (wrong) method.

You need to put scope in AUTO memory mode. That is what your beloved Keysight does. It says 2 or 8 MPts, but it is a lie. It captures only as many dots as fit on screen. Only if you press Stop it will then trigger new, separate trigger and capture full memory from there.

[kaspich]

Can you tell us the conditions under which the tests were performed?

- time base (we have it)
- signal frequency
- what is considered blind time? maybe a screen-capture with the measurement?
- the memory setting does not seem to be uniform. From what I know, Siglent takes into account the memory depth required for a screen for acquisition (14 cells in your language from the table). For Magnova it appears that you set 300MPts ... in this case the settings are completely different.

I could perform the tests with Siglent 2000 X Plus and Rigol DHO1000 for comparison, but the conditions should be similar for all oscilloscopes.

Core,

please read my post carefully:
a. first test (before p.s.): no signal/signal below the trigger threshold applied; second test (triggered): 10MHz;
b. no, not 320M points. There are TWO results for Magnova: the best one is for ''fast'' memory setting;
c. as I mentioned before - there is an automatic memory management for MicSig and Agilent, so creating identical conditions (regarding the memory usage) is not possible. For Siglent - two results, two memory depth settings (full and 140K points). Ok, maybe it makes sense to perform a ''correct'' test between different Siglent scopes, but the remaining ones (Agilent, MicSig, and modern digital) are miles away...


Anyway, if coverage/uptime parameter (just as an example) is 80..90% for one scope and 0.2..1% for another, is there a need for extra deep analysis? That's a disaster...
As you see, Magnova's wfps is horrible even in ''fast'' mode, in 300Mpoints mode it becomes a ''brick'' (even - response to keys slows down to 1..2..3 seconds).

[kaspich]

I honestly do not understand what numbers are you showing here.
But I know those numbers are dead wrong.

Also I think you are confusing the memory management modes with trigger modes.
There is Auto and manual memory modes, and there are trigger modes: Auto, Normal and Single.

Did you seriously proclaim that triggers per seconds are measure of how often scope triggers in Auto trigger mode?
What do you mean triggered/untriggered makes no difference. In Normal trigger mode?
In Normal trigger mode without signal producing trigger event, scope will not trigger and capture anything at all.
In Auto trigger mode, scope will acquire occasional screen refresh, based on some timeout. It is not going as fast as it can, but will deliberately wait to see if signal will actually trigger something, and only if that does not happen for a very long time, it will capture something randomly just to refresh screen.

Also, again, memory. You put memory in MANUAL 300 MPts mode. Scope will capture 300 MPTS always, including where you are in 2ns/div and there are only 5 points shown on the screen.
So if you are at At 1.6 GS/s it will take 0,1875 seconds per every trigger to capture all that data. That is 5.3 triggers per second ideally, if scope has no blind time at all.

All your other measurements are probably done using same (wrong) method.

You need to put scope in AUTO memory mode. That is what your beloved Keysight does. It says 2 or 8 MPts, but it is a lie. It captures only as many dots as fit on screen. Only if you press Stop it will then trigger new, separate trigger and capture full memory from there.

Just one fast remark - Magnova's memory was/is in Auto (fast) mode. And, yes, all scopes are in Auto mode.
Anything else?

But, please, perform/present your own CORRECT test. Thanks in advance.
Btw, thanks for the clarification. Again - some unknown to me (as well as aliasing, etc.).

And yes, this time I am talking about ''uptime'' as a function of data representation/refresh rate (not storage). Sure, if/when memory depth is more than the "length'' of the screen, additional data is stored and can be evaluated when the scope is stopped. But, there is one more ''small'' problem - if this memory buffer is ''activated'' once per second, I see nothing on the screen. I have no idea when to stop my scope. And, even if memory depth covers 10 ''screens'', this increases ''uptime'' from 0.25% to 2.5% for modern digital (available only after the scope is stopped)... 

Guys, yes, my tests are ''incorrect''. At the same time, if the refresh rate of one scope covers 90% of the time, another one has 1wfps - sorry, there is no need to invest/wast additional time. 

[core]

Regarding the update rate, please try to reproduce Martin72's tests:

https://www.eevblog.com/forum/testgear/magnova-oscilloscope/msg5740649/#msg5740649

There is definitely something wrong with your approach.

Measuring update rate without signal is nonsense, it has no relevance. Each oscilloscope has its own approach to self-triggering.

Regarding "uptime", "blind time", it seems that these notions are a bit mixed and confused. No offense. I'm trying to help.

Magnova might have some problems, but in no case regarding update rate.

If you give up Magnova, this should not be the cause, in any case.

But if you simply can't deal with it, that's it ...

[2N3055]

We already explained before.

Memory mode must be in AUTO.

Trigger mode must be in Normal.

Apply 10-20 MHz signal to input, with vertical set it to nicely fit on screen vertically. Set edge trigger, and make it trigger reliably.

Then measure.

If you set scope to manual memory mode of 300 MPTs you command it to always capture full 300 Mpts. Regardless the timebase setting.

I tried with inexpensive SDS800xHD. At 100µs/div, it gets 200 triggers per second at 2Mpts and 287 triggers/s at 1MPts (1GS/s). Magnova numbers are faster than that scope, and two orders of magnitude out in your measurements.

[kaspich]

We already explained before.

Memory mode must be in AUTO.

Trigger mode must be in Normal.

Apply 10-20 MHz signal to input, with vertical set it to nicely fit on screen vertically. Set edge trigger, and make it trigger reliably.

Then measure.

If you set scope to manual memory mode of 300 MPTs you command it to always capture full 300 Mpts. Regardless the timebase setting.

I tried with inexpensive SDS800xHD. At 100µs/div, it gets 200 triggers per second at 2Mpts and 287 triggers/s at 1MPts (1GS/s). Magnova numbers are faster than that scope, and two orders of magnitude out in your measurements.

All settings were already set 1:1 as you wrote
Some results are mentioned in my post, but I'll add them to my table!

p.s. 3rd column appeared for Magnova. 10MHz input.

My short conclusion:
a. there is a 20-year-old Agilent that have independant FIFO memory buffer (hardware solution) and a fast/independent trigger;
b. there is a ''modern digital'' sampling tool. Ok, I understand that small manufacturers like Batronix don't have access to custom-made FPGAs (this results in well-known problems at high sampling rates). But, sorry, at slow sampling rates, there are software solutions for memory management.

[kaspich]

Regarding the update rate, please try to reproduce Martin72's tests:

https://www.eevblog.com/forum/testgear/magnova-oscilloscope/msg5740649/#msg5740649

There is definitely something wrong with your approach.

Measuring update rate without signal is nonsense, it has no relevance. Each oscilloscope has its own approach to self-triggering.

Regarding "uptime", "blind time", it seems that these notions are a bit mixed and confused. No offense. I'm trying to help.

Magnova might have some problems, but in no case regarding update rate.

If you give up Magnova, this should not be the cause, in any case.

But if you simply can't deal with it, that's it ...

In any case? In my case - yes, it's an argument.
About its own approach - yes.
Come on, scope that refreshes its screen once per second...
Guys, WHAT should happen to tell - yes, this time, Magnova failed

[Sorama]

There is certainly no lack of self confidence.
Probably way too much…