Magnova oscilloscope

[kaspich]

You're not comparing apples to apples, and you should pay attention to what shows in your screenshots...and fix your screenshots to show the related data, rather than unrelated settings.

Look at the sample rates you have set in the Siglent scope vs. the Magnova. They're not set the same way.

ok, I'll try very slowly.
Agilent 6000 series. 8M memory option. 300MHz. 20-year-old scope.
Regardless: Normal mode or Roll mode, for ALL timescale settings, completely correct picture.

Siglent 2000 series. 70MHz version.
For Normal mode, correct picture for ALL timescale settings!

Tested on different frequencies.

Magnova: 5+ different pictures (all completely wrong) depending on timescale settings.

Cool explanation. Now look at your screenshots and compare the sampling rates and see if your results might make sense.

Which part was too complicated?

[S13]

Can confirm that my Siglent 1000X-HD and Micsig MHO1 show the same aliased signals when undersampling, just like your Magnova. Only when using Peak Detect can you avoid the aliasing.
I question that the Agilent would actually be doing any better, you might just think you are seeing a correct result because the clocks of the generator and scope are not synchronized? Maybe try to fine tune the frequency and see if you can get the Agilent to alias?

As far as I understand, if/when samplerate is set to 1MHz (for example), it's 1/1000 of the ADC capabilities.
I assume, 99% of the time, ADC is on ''hold'' as well as all remaining FPGA/CPU. So (just idea/proposal), why not change the samplerate periodically (automatically/hiddenly) to a higher rate to capture the whole spectrum/trace of the input signal?

What you are describing is basically Peak Detect.


Here is your 50MHz signal on a Micsig MHO1 running at 50MS/s:
 


The frequency counter is correct, but the wave on screen shows only a 12Hz (aliased) sinewave. Peak detect solves this.
Interestingly, the Micsig also has an "Envelope" sampling mode with adjustable number ranging from 2 - infinity. Im not quite sure how this sampling mode works, but it also reconstructs the wave on screen properly, just like Peak Detect. The next picture is for envelope-16 mode:



Anyone know what envelope mode does?

[KungFuJosh]

Which part was too complicated?

For me, none of it. You've got inconsistent settings, and you're not comparing equal setups.

Why do you have offsets applied only to the Magnova?

Are you comparing the HARDWARE counters in the other scopes, to the measured / soft counter you currently have in the Magnova?

The only overlap in timescale is 50ms in the last two comparisons.

Make all settings / timings / setups the same before you compare. Remove all offsets/delays or make them consistent across the board.

[kaspich]

Can confirm that my Siglent 1000X-HD and Micsig MHO1 show the same aliased signals when undersampling, just like your Magnova. Only when using Peak Detect can you avoid the aliasing.
I question that the Agilent would actually be doing any better, you might just think you are seeing a correct result because the clocks of the generator and scope are not synchronized? Maybe try to fine tune the frequency and see if you can get the Agilent to alias?

As far as I understand, if/when samplerate is set to 1MHz (for example), it's 1/1000 of the ADC capabilities.
I assume, 99% of the time, ADC is on ''hold'' as well as all remaining FPGA/CPU. So (just idea/proposal), why not change the samplerate periodically (automatically/hiddenly) to a higher rate to capture the whole spectrum/trace of the input signal?

What you are describing is basically Peak Detect.


Here is your 50MHz signal on a Micsig MHO1 running at 50MS/s:
 
(Attachment Link)

The frequency counter is correct, but the wave on screen shows only a 12Hz (aliased) sinewave. Peak detect solves this.
Interestingly, the Micsig also has an "Envelope" sampling mode with adjustable number ranging from 2 - infinity. Im not quite sure how this sampling mode works, but it also reconstructs the wave on screen properly, just like Peak Detect. The next picture is for envelope-16 mode:

(Attachment Link)

Anyone know what envelope mode does?

Yes, I tried fine-tuning my generator. Interference on Agilent is barely visible (just maybe 1% of the peak value). So, I see how small ''jittering'' is waving left/right. And - that's it.
It seems peak detection is added/enabled automatically.

[kaspich]

Which part was too complicated?

For me, none of it. You've got inconsistent settings, and you're not comparing equal setups.

Why do you have offsets applied only to the Magnova?

Are you comparing the HARDWARE counters in the other scopes, to the measured / soft counter you currently have in the Magnova?

The only overlap in timescale is 50ms in the last two comparisons.

Make all settings / timings / setups the same before you compare. Remove all offsets/delays or make them consistent across the board.

ok, my first question is - how could an offset affect such interference/aliasing?
Or, maybe - could (taking into account Roll mode bug, reported by another user, but not reacted by Magnova team, as well as my report about Roll mode, too). Sorry for my sarcasm.
 
So far, I had an illusion that the dev team has its own interest in fixing bugs/problems. Or should I prove? Maybe write a whitepaper?
From my perspective, the situation is pretty simple. There are aliasing problems (well-known). Question is - is this situation acceptable for this product? This is the decision of Andre/dev team. So simple.
If the dev team decides to add a peak algorithm or other solutions (there are several options), I'll be happy. If not, I'll be disappointed. So simple.

[KungFuJosh]

Here's an example that seems like expected behavior to me, and to those who previously explained it better than I can.

Internal AWG with system synchronized to external GPSDO.

Initially AWG set to 900kHz, then 1MHz, then 1.1MHz.

[S13]

Yes, I tried fine-tuning my generator. Interference on Agilent is barely visible (just maybe 1% of the peak value). So, I see how small ''jittering'' is waving left/right. And - that's it.
It seems peak detection is added/enabled automatically.

Maybe that is what they are doing, but then again always running Peak Detect adds noise to the waveform, so im not sure if Agilent can get away with that? (or maybe they can, in the 8 bit era and small screen sizes a bit of additional noise on the waveform at longer timescales is not a problem?)

[rf-messkopf]

If the dev team decides to add a peak algorithm or other solutions (there are several options), I'll be happy. If not, I'll be disappointed. So simple.

It is already there, just like in any other modern digital scope I have seen. Tap on Sample below the screen, and then hit Peak Detect in the dialog box.

[rf-messkopf]

Again, apologies for the trouble - this should not have made it out into the field. We’ll get this sorted as quickly as possible.

No worries. These things happen to the best of us.

I'll shoot you an email.

Just a small shoutout to Batronix.

Today I received a replacement AWG module that works just fine. Huge thanks for the smooth process, and for the thoughtful gift of the Batronix mug. It will definitely come in handy in the workshop. 

[john74]

So far, I had an illusion that the dev team has its own interest in fixing bugs/problems.

what you see is not a "bug" ... please watch this video completely:
(I find this a very good explanation)

[nctnico]

This is not a question of cost, but about the limitations of sampling systems. If the scope needs to decimate at large time/div settings, it has to discard samples, i.e., throw away information. This doesn't come for free. Therefore the scope gives you options to handle that, e.g., the simple Sample mode, or Peak Detect, or Boxcar.

As I said above, in theory another option might be to dither the sampling clock, which would turn aliasing products into noise. But that would affect the wanted signal components as well, and add jitter to them.

Actually there is a trick to decimating long records to a visible display quickly which is adopted by many oscilloscope manufacturers. I think this is based on using subsampling at irregular intervals to prevent aliasing issues. However, this method (however it is implemented) can still be tricked into showing aliasing using slow frequency sweeps. But for signals with a fixed frequency it works fine.

[rf-messkopf]

This is not a question of cost, but about the limitations of sampling systems. If the scope needs to decimate at large time/div settings, it has to discard samples, i.e., throw away information. This doesn't come for free. Therefore the scope gives you options to handle that, e.g., the simple Sample mode, or Peak Detect, or Boxcar.

As I said above, in theory another option might be to dither the sampling clock, which would turn aliasing products into noise. But that would affect the wanted signal components as well, and add jitter to them.

Actually there is a trick to decimating long records to a visible display quickly which is adopted by many oscilloscope manufacturers. I think this is based on using subsampling at irregular intervals to prevent aliasing issues. However, this method (however it is implemented) can still be tricked into showing aliasing using slow frequency sweeps. But for signals with a fixed frequency it works fine.

Isn't subsampling at irregular (random) intervals similar to dithering the sampling clock? As far as I can see, that would turn aliasing components into noise as well, but also affect the wanted signal components.

But with this discussion, we're slowly starting to hijack the thread. I suggest that we move basic discussions about the function of digital oscilloscopes to a separate thread if there is interest in that topic.

[kaspich]

Yes, I tried fine-tuning my generator. Interference on Agilent is barely visible (just maybe 1% of the peak value). So, I see how small ''jittering'' is waving left/right. And - that's it.
It seems peak detection is added/enabled automatically.

Maybe that is what they are doing, but then again always running Peak Detect adds noise to the waveform, so im not sure if Agilent can get away with that? (or maybe they can, in the 8 bit era and small screen sizes a bit of additional noise on the waveform at longer timescales is not a problem?)

If you are inspecting a high-frequency signal in low timescale mode, accuracy could be reduced (and will be reduced because all small details are ''too close" to each other). Most important - shape/tendencies should be visible! So, a small amount of ''noise doesn't make sense!

[kaspich]

If the dev team decides to add a peak algorithm or other solutions (there are several options), I'll be happy. If not, I'll be disappointed. So simple.

It is already there, just like in any other modern digital scope I have seen. Tap on Sample below the screen, and then hit Peak Detect in the dialog box.

ok, please try it

[kaspich]

So far, I had an illusion that the dev team has its own interest in fixing bugs/problems.

what you see is not a "bug" ... please watch this video completely:
(I find this a very good explanation)

please read my other posts.

[kaspich]

This is not a question of cost, but about the limitations of sampling systems. If the scope needs to decimate at large time/div settings, it has to discard samples, i.e., throw away information. This doesn't come for free. Therefore the scope gives you options to handle that, e.g., the simple Sample mode, or Peak Detect, or Boxcar.

As I said above, in theory another option might be to dither the sampling clock, which would turn aliasing products into noise. But that would affect the wanted signal components as well, and add jitter to them.

Actually there is a trick to decimating long records to a visible display quickly which is adopted by many oscilloscope manufacturers. I think this is based on using subsampling at irregular intervals to prevent aliasing issues. However, this method (however it is implemented) can still be tricked into showing aliasing using slow frequency sweeps. But for signals with a fixed frequency it works fine.

exactly. One of the methods. If these burst packs are selected randomly (''noisy''), this trick also works for sweep (works for ALL type periodic signals).

[kaspich]

Andre and team, this post is dedicated to you.
This scope has good potential. But. There are some problems.
Let's start with target users.
Digital guys? 350MHz, 1.6GHz samplerate? In 2026? For 5K EUR? No.
12bit frontend, large display - analog guys? And there is a problem.
I am mostly an analog guy. Maybe I am unique, but... This scope is not for me.
why?
1. problems mentioned above. When I connect the probe to my test object, I should see a true and trustworthy picture. Unfortunately, I CANNOT trust this scope.
My major requirement for scope: if the input signal is within the allowed range (in this case: DC to 350MHz), the scope should provide a correct (as possible) picture. Without exceptions!
Pretty often, I need to evaluate slow deviations/changes of a high-frequency signal (for example, self-oscillation, interference from the digital part, etc.). This scope acts like cheap China-made craps (trust me, I have some experience working with different scopes in the last 40 years).
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?
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?
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).

If the community of this forum believes that everything is perfect (and just some small bugs need to be fixed), I say sorry for my posts.
Guys, several times you mentioned ''modern digital scope''. Seriously??? My 20-year-old Agilent performs correctly (at least if we are talking about aliasing issues). My 10-year-old China-made Siglent performs correctly in Normal mode. This ''modern'' scope totally fails. Maybe it's some kind of innovation?
So far, this scope is not for me. p.s. and definitely, it's not ''German quality'' (I mean software).

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



   

[S13]

Sometimes the conclusion is that a different scope just suites you better. No arguing in that.
Agilent, the scopes you are comparing with, sure makes nice scopes with great technology. I used to love the MSO6034A we had at the office. So quick and responsive, it just worked. But those things do come at a premium price.
Nowadays im happy with more affordable scopes, that surely don't perform as the old Agilent, but do offer more features. And i accept that these scopes just aren't up there with the high end stuff, in terms of stability, reliability and such.

Imo that doesnt make the Magnova a bad scope by any means. But maybe a bit steeply priced for something that is still partially in development?

I will emphasize that the issue with aliasing is quite common with scopes, hence the youtube video posted earlier. Siglent is no saint in this matter unlike what you suggest, they have aliasing (i checked on my 1000X-HD). And Micsig en R&S all suffer from it. Rigol probably too. Dont know about Tek, but Lecroy most likely too since they overlap with Siglent. Most of us learn to manage it without much hassle. Keysight even has a page explaining it, and how to prevent it:
https://www.keysight.com/used/us/en/knowledge/glossary/oscilloscopes/what-is-aliasing
But it might be nice to know if HP/Agilent/Keysight use a specific anti-aliasing sampling strategy. Could even be patented i suppose?

Avoiding "Store Every N-th Sample": Unlike some oscilloscopes (e.g., certain Tektronix models) that, at slow sweep speeds, simply store every n-th sample, Agilent scopes use their acquisition memory more intelligently. This "decimation" method discards data, creating large gaps between stored samples and making the scope highly susceptible to aliasing.
Agilent oscilloscopes process the high stream of data from their ADCs differently. Instead of discarding samples, they use methods like averaging, peak detection (min/max), or random equivalent time (R.E.T.) sampling to store information from the samples that occur between the final data points. This preserves more signal information and significantly reduces the risk of high-frequency components slipping through undetected.

This quote is AI generated, so take it with a grain of salt...

[KungFuJosh]

That doesn't make any sense. As it stands, there's nothing in the price range that competes with the Magnova's feature set. It will be even better after the next firmware update.

Essentially, all modern digital scopes work roughly the same way, but Magnova should lose points for the same behavior seen on scopes from brands/models that cost significantly more?

[core]

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...

[Martin72]

A bit high
I was actually wondering the opposite—how do you manage to get a scope like that for such a low price...

[Sorama]

if you are into development and production of electronics, you know that devices are not particularly expensive to put into the market.

The real (and only) problem is that manufacturers take way too much profit on it.

I know cause I am one of them :-)

[KungFuJosh]

if you are into development and production of electronics, you know that devices are not particularly expensive to put into the market.

The real (and only) problem is that manufacturers take way too much profit on it.

I know cause I am one of them :-)

That's not how it works. Making one product with high margin doesn't mean every other product has a high margin. Certainly not at this price point for what they're selling. An R&S scope for 6 or 7 figures, sure. That's got some margin on it.

It costs money to pay people to update and support a product. That's besides the hardware side of costs.

It also costs money to license products for sale in different countries on this scale. I don't think you do this scale work.

[Sorama]

if you are into development and production of electronics, you know that devices are not particularly expensive to put into the market.

The real (and only) problem is that manufacturers take way too much profit on it.

I know cause I am one of them :-)

That's not how it works. Making one product with high margin doesn't mean every other product has a high margin. Certainly not at this price point for what they're selling. An R&S scope for 6 or 7 figures, sure. That's got some margin on it.

It costs money to pay people to update and support a product. That's besides the hardware side of costs.

It also costs money to license products for sale in different countries on this scale. I don't think you do this scale work.

Actually , I do.
Well, I did.
Own development in my country, production in Eastern Europe (and later in my country) , sale through partners in Germany, Spain, Italy, the Netherlands and France.

I’m pretty sure low cost devices still have a profit margin of magnitude of 4 to 5 the production cost.

[KungFuJosh]

I’m pretty sure low cost devices still have a profit margin of magnitude of 4 to 5 the production cost.

Yes, but that's not what an oscilloscope like this is. Low cost might be true of some Fnirsi garbage, but not a scope like this.

...and as somebody who is still making stuff, but not remotely close to the scale of Batronix, the costs for everything went up. Once with Covid, and again with the moron here and his tariffs.