Siglent SDS2000X Plus

[rf-loop]

And Keysight proves that it is perfectly possible to have high update rates AND deep memory at the same time.

Keysight own Salesmens University have perhaps own basic math book. 1M memory need one M samples to fill. If sampling speed is 1GSa/s it takes 1ms and even with zero blind time (impossible) 1kwfm/s. But when deep memory is selected and scope is running for high update rates it can not capture deep memory, but salesmens forget to tell it....  Yes there can do example last acquisition trick.... and fool users (I do not know what is Keysight salesmens trick for this).

There is no fooling users. Now you sound like the sales brochure you loath so much.

Just run two acquisition systems in parallel. One for storage and one for display. I have implemented something similar a couple of years ago. Works like a charm.

But how about deep full mem measurements during this high speed update rate with deep memory, how salesmens jump over this lazy fox. Oh... they never promised...

No oscilloscope can measurements / math realtime. And it seems Siglent does ERES and Sin x/x interpolation in software which is even worse. But then again I don't really care about waveform update rates. I rather have an oscilloscope which offers a high rate of productivity (=least amount of twiddling knobs). The latter is where Siglent needs to improve.

Huhh... how can I stob this laughing, lets make scope what have two screens... fast wfms and then slow deep memory and same power switch so they can run "parallel" oh my... well synched. 
But of course it need be fun, now is Spring Festival eve aka New Year eve.

[0xdeadbeef]

But that still doesn't give you acquisition beyond the edge of the screen even if there is enough memory to do so. And Keysight proves that it is perfectly possible to have high update rates AND deep memory at the same time.

Is there even any affordable (<10k) Keysight scopes that qualifies for "deep memory"? Last time I had a look, they had 4Mpts which is not exactly what I would call deep.

In many cases waveform update rate doesn't really matter when debugging digital / mixed signal circuits. Regulary I use normal mode and take single shots of specific events. It would be logical to have the centre of the screen as the centre point of the memory (half left, half right). At the point where this starts to matter you'll need to choose the time/div and trigger offset settings so you capture the part of the signal you want. But with deep memory this isn't difficult. The biggest benefit of deep memory is that you don't have to care so much about the trigger.

I not an a update rate aficionado either but typically Keysight sacrifices absolutely everything for update rate. So I'm puzzled that of all possible manufacturers, Keysight would sacrifice update rate for cpaturing additional data. Are you sure that they show this behavior in normal mode and not only in single trigger mode? And as stated above: single trigger mode is a different topic and it's pointless to compare the behavior of one scope in normal trigger mode with another one in single trigger mode.

[nctnico]

4MPts isn't exactly deep indeed but the way (-I think-) Keysight works it is easely expandable to 40Mpts, 400MPts, 4Gpts. Doesn't really matter.
And Keysight doesn't sacrifice anything. If I'm right they use two parallel acquisition systems. One with the large memory which works like a continuous circular buffer and one with a short buffer (enough to fill the screen) to do high update rates. Once you press stop Keysight scopes wait until the rest of the large memory buffer is filled before allowing the user to scroll left / right. It is a very simple & elegant solution.

[TK]

4MPts isn't exactly deep indeed but the way (-I think-) Keysight works it is easely expandable to 40Mpts, 400MPts, 4Gpts. Doesn't really matter.
And Keysight doesn't sacrifice anything. If I'm right they use two parallel acquisition systems. One with the large memory which works like a continuous circular buffer and one with a short buffer (enough to fill the screen) to do high update rates. Once you press stop Keysight scopes wait until the rest of the large memory buffer is filled before allowing the user to scroll left / right. It is a very simple & elegant solution.

Can you please explain what you mean by: Keysight works it is easely expandable to 40Mpts, 400MPts, 4Gpts. Doesn't really matter.  It sounds interesting

[supperman]

My Tektronix MDO3k seems to capture the full memory depth that you set regardless of time-base. I guess that is the point to setting the memory in the first place.. since that setting would serve no purpose if it was time-base dependent.

If it have 10M memory length and 5GSa/s and IF it always use full memory regardless of timebase how it reach nearly 300kwfm/s capturing speed. Yes of course it can not. With 5GSa/s alone 10Msamle takes 2ms. (I believe there is other mode for max wfm speed)
If user have 1ns/div setting and IF screen is (just for easy numbers) 1 div wide... user can see 10ns and 1999980 ns is out from display and rest are invisible included to true visual blind time.

Yes, if you set the memory to 10m it will take longer to capture and the number of triggers goes down below the rated speed of the scope. It can take 10 seconds or longer per trigger depending on the time base (i.e. 1/10 sample per second). I thought that is well understood and the exact behavior you want it to have.. You want it to be faster then make the memory smaller.. that is why you have the setting.

The memory length setting would serve no purpose if it was overridden by the time base setting. Why would you ever set it to less than the max? (perhaps segmented memory has something to do with it.. but I think in that case you should have 2 settings, memory length per trigger and number of triggers captured)

[rf-loop]

4MPts isn't exactly deep indeed but the way (-I think-) Keysight works it is easely expandable to 40Mpts, 400MPts, 4Gpts. Doesn't really matter.
And Keysight doesn't sacrifice anything. If I'm right they use two parallel acquisition systems. One with the large memory which works like a continuous circular buffer and one with a short buffer (enough to fill the screen) to do high update rates. Once you press stop Keysight scopes wait until the rest of the large memory buffer is filled before allowing the user to scroll left / right. It is a very simple & elegant solution.

So it is something like as think, last acquisition trick... elegant solution yes but...claiming it do deep memory and fast update rate is bit borderline...  I do not deny that it can be useful in some cases.

[nctnico]

4MPts isn't exactly deep indeed but the way (-I think-) Keysight works it is easely expandable to 40Mpts, 400MPts, 4Gpts. Doesn't really matter.
And Keysight doesn't sacrifice anything. If I'm right they use two parallel acquisition systems. One with the large memory which works like a continuous circular buffer and one with a short buffer (enough to fill the screen) to do high update rates. Once you press stop Keysight scopes wait until the rest of the large memory buffer is filled before allowing the user to scroll left / right. It is a very simple & elegant solution.

Can you please explain what you mean by: Keysight works it is easely expandable to 40Mpts, 400MPts, 4Gpts. Doesn't really matter.  It sounds interesting

You can make a circular buffer for this purpose as long as you like. The only limitation is the amount of pre-fill time. If memory is split in half and the 50% mark of the memory is at the centre of the screen (remember the trigger point itself can be anywhere so it doesn't make sense to use the trigger point a reference to a location in the memory) then you have to wait until 50% of the memory is filled before the first trigger event can be processed. With 4Mpts (and in case of Keysight 1Mpts in reality) the prefill time doesn't really matter much.

@rf-loop: no, using parallel acquisition systems isn't borderline anything. It works and gives the user best of both worlds under all circumstances. There are no downsides to it.

[Martin72]

Who can check if this model can use internal signal generator for BP use only without FG license installed.

I can…
But: License is not installed, so far so "good", but my scope got this function in trial-mode, don´t know NOW what happen if the trial is running out.
Maybe I should activate this 27 times to know it - No..

[0xdeadbeef]

4MPts isn't exactly deep indeed but the way (-I think-) Keysight works it is easely expandable to 40Mpts, 400MPts, 4Gpts. Doesn't really matter.

Well, no, for a couple of reasons. We'll see about that.

And Keysight doesn't sacrifice anything. If I'm right they use two parallel acquisition systems.

For sure they sacrifice functionality for update rate. E.g. the first thing they sacrificed was memory depth. In the past they also crippled measurements by using only the onscreen data. Someone in this forum stated this is no longer the case, since they introduced a measurement buffer but my impression is that this is also some kind of workaround instead of measuring and displaying from the actual sample buffer as LeCroy and Tek scopes do.
But anyway: I think it's clear that they only achieve the high update rate by using an expensive specialized ASIC and this makes it impossible for them to easily increase the memory depth. So whatever their current approach is, it can't be easily extendend to 4GPts.

One with the large memory which works like a continuous circular buffer and one with a short buffer (enough to fill the screen) to do high update rates. Once you press stop Keysight scopes wait until the rest of the large memory buffer is filled before allowing the user to scroll left / right. It is a very simple & elegant solution.

Actually that doesn't seem to be possible even from a theoretical point of view. If you're in normal trigger mode and press stop, there is usually no way to continue sampling from the trigger point unless pressing the button happens more or less at the same time as the last trigger event. E.g. assuming the current DOSX3000T Keysights have 4MPts for your hypothetical circular buffer: at 2.5GSa/s, it would be filled within 1.6ms. So if you stop the aquisition in normal trigger mode and the trigger event happened e.g. 10ms before, the circular buffer was filled again multiple times. Sorry, but this just doesn't seem to be a feasible approach.
Again, what you're suggesting is IMHO only possible in single trigger mode and looking into the DSOX3000T manual seems to confirm this. Quote from the manual: "Pressing [Single] instead of [Stop] fills the maximum memory depth". And I think that other scopes fill the whole selected ("maximum") memory in single trigger mode as well (if this is feasible).

Besides, even letting all of this aside, your suggested approach would not be "easily expandable" to any memory depth. As you say, the scope needs to fill the buffer before it lets you have a look, scroll around etc. Now with 4Gpts and a sample rate of 1MSa/s, it would take 4000 seconds to fill the buffer. That's more than an hour. Now with the 4Mpts of a current Keysight, that's not as much of an issue unless you select a 1kSa/s or so (if that's even possible). Actually I would think that a Keysight doesn't allow you to select an acquisition setup that would delay reaction to a key by even a few seconds. Yet with 200MPts of a Siglent SDS2000X Plus, it would take 200 seconds at 1MSa/s to fill the memory. That doesn't sound horribly feasible either. So it's natural that a scope tries to select either memory depth or sample rate to avoid getting locked or slow down to a crawl. Actually, of all scopes, LeCroys tend to allow you setting up acquisition modes that are terribly slow or only fill a part of the screen and Keysight adepts criticize them for exactly this lack of automatically selecting a setup that gives immediate feedback.

[Martin72]

Maybe I should activate this 27 times to know it - No..

No need to do this - the trial-counter decrease when you use the bode plot function, so I guess the option must be installed to use it as internal source.
Would be nice if siglent could do a fix in that case.
Apropos fixing…
You turn on the the bode plot and the awg will turn on too.
But if you disable bode plot, the awg remains to be on.
Clearly a bug.

[nctnico]

So if you stop the aquisition in normal trigger mode and the trigger event happened e.g. 10ms before, the circular buffer was filled again multiple times. Sorry, but this just doesn't seem to be a feasible approach.

You are overthinking it. The circular buffer runs continuously BUT like every scope Keysight uses double buffering! In your example the first buffer is stopped after a trigger event but the second buffer is not. The buffering is not stopped / started at every trigger event. Think about that for a while and you'll see that this works perfectly when using a parallel acquisition system. Remember: I have build such systems several times already using standard FPGAs. There is no need for a special ASIC.

Regarding the prefill time: that is why oscilloscopes have several selections so the user can select a memory depth which is most suitable for the task at hand. After all an oscilloscope is supposed to be a universal tool and some people may have a use to have 4Gpts being sampled at 1Ms/s. I switch between memory depth settings if necessary.

[0xdeadbeef]

Daniel Bogdanoff stated once that the MegaZoom scopes use something they call "ping pong" memory in normal trigger mode. This is why the memory is halved in normal trigger mode vs. single trigger mode - letting aside that the 4MPts are really all the memory there is, so without interleaving, you effectively only have 1MPts per channel in normal trigger mode and 2MPts in single trigger mode. Actually I'm unsure how the interleaving works exactly and there seem to be two MegaZoom ASICs in the DSOX3000, but for the moment I assume you get 1MPts per channel in normal trigger mode with all four channels enabled. Anyway, so much about the deep memory.
Now would this work without a specialized ASIC with onboard memory? Well, maybe it would in the meantime as the ASIC is old now and FPGAs improved over the years. But given that no FPGA scope has yet reached the overall speed (taking into account also decoding speed and responsiveness) of the MegaZoom scopes, this is just an assumption.
Anyway, the term "ping pong" is usually used for a linear double buffer, not for a circular one. Also the sampling doesn't continue for the same trigger, but the two buffers are used for distinct triggers. So the data of the first buffer is displayed on the screen and the 2nd buffer is used to serve the next trigger event. Again, this just shows that they sacrificed everything for fast update rates and short dead times. I just don't see the magic in this.
Now, is this the ultimate solution that would work just the same with 4GPts? No, of course it isn't and wouldn't. Even a double buffering approach means that you can't reuse the same buffer while it's still processed. If you need to fill a gigantic buffer, you can split it in half to double the update rate or in a hundred buffers, but the only benefit is the higher update rate and smaller dead time.
Anyway, even if the DOSX3000 scopes actually captures the whole 2MPts (interleaved) or 1Mpts (non-interleaved) , I wonder if this "more" is even worth discussing looking at the buffer sizes. If at all, the additional data capture seems be more of an artifact of the double buffer strategy (might simplify the buffer handling to use fixed sizes) and doesn't cause a significant performance hit (update/dead time wise) because the buffers are so small anyway.

[nctnico]

I don't think getting deep into the nitty-gritty of implementation details is very helpfull. And I disagree about responsiveness and decoding speed being unique to Keysight. There are many other scopes out there with fast decoding update speeds (R&S for example). Responsiveness of the user interface has nothing to do with acquisition hardware but everything with how the scheduling is done in the software.

[tautech]

I don't think getting deep into the nitty-gritty of implementation details is very helpful.

It is as it helps us understand how each brand addresses their memory depth management and why they each have selected a particular scheme.

A facit that needs further discussion so to better understand Siglents implementation is their continuously running History buffer of some 200 MPts.

[nctnico]

I don't think getting deep into the nitty-gritty of implementation details is very helpful.

It is as it helps us understand how each brand addresses their memory depth management and why they each have selected a particular scheme.

But not up to the point where 'we' are about to write VHDL and C code.

A facit that needs further discussion so to better understand Siglents implementation is their continuously running History buffer of some 200 MPts.

I'm quite sure nobody cares about history mode because it basically is an acquisition mode without clearly defined boundaries. The way it sits now it is just filling memory which could be put to much better use. I have come across several DSOs which have history mode but (unlike segmented recording where you can set clear boundaries) I never really found a good use for history mode. Once you want to define boundaries (like the number of segments and length) you get into segmented recording mode territory immediately. R&S has both on the RTM3004 but history mode is basically segmented mode using a first-in first-out recording mode.

[Martin72]

Lecroy also relies on history mode and on the memorydepth management - I think, when having full memorydepth independend from timebase is an important thing, they would implement it.
But they don´t and I´m sure they could if they want, there must be a reason why some brands doing it and some brands don´t.
Wether they´re so called A-Brands or not.
They'd be fools to weaken themselves against the competition.

[tautech]

I'm quite sure nobody cares about history mode because it basically is an acquisition mode without clearly defined boundaries.

You don't, you've already stated you don't care about wfps and hence blind time.....others do !

The age old discussion of DSO vs CRO has much to do about blind time of DSO's so why would a designer implement a memory management scheme where blind time could impact on scope usability ?

[nctnico]

Lecroy also relies on history mode and on the memorydepth management - I think, when having full memorydepth independend from timebase is an important thing, they would implement it.
But they don´t and I´m sure they could if they want, there must be a reason why some brands doing it and some brands don´t.
Wether they´re so called A-Brands or not.
They'd be fools to weaken themselves against the competition.

Let's put it this way: there are less brands which don't have recording outside the screen compared to brands who have. So far Lecroy and Siglent seem to be the only brands. Lecroy always has been a bit of an odd-ball brand; therefore Lecroy isn't the best benchmark when looking for industry standards. Lecroy doesn't have peak detect for example which also is a limiting factor. Don't ask me for the reason. It could be a limitation of their acquisition hardware in order to make it cheaper or a principle. Last week I had a junior engineer come to me complaining a 1PPS signal was intermittant. Turning on peak detect on the oscilloscope he used fixed that; the pulse was too narrow to show up on screen.

[nctnico]

I'm quite sure nobody cares about history mode because it basically is an acquisition mode without clearly defined boundaries.

You don't, you've already stated you don't care about wfps and hence blind time.....others do !

The age old discussion of DSO vs CRO has much to do about blind time of DSO's so why would a designer implement a memory management scheme where blind time could impact on scope usability ?

Now you are moving the goal posts trying to make it look like it is never good enough for me. People are smart enough to see through that. It has been pointed out many times before that modern triggering features (like zone triggering and runt triggering) are much more useful for catching rare events than just staring at a screen. Besides that allowing variable memory depth allows the user to select between long (partly off-screen) acquisitions or high waveform updates so if people feel inclined to stare at a screen then they can still do so. In my RTM3004 review I have already showed that deep memory combined with clever waveform display techniques can make a rare event stand out like a sore thumb even without resorting to specific triggers.

[Martin72]

There must be advantages and disadvantages given, when some makes it possible to use full memdepth over all and some don´t.
I did measurings since 17yrs at work, 5 days a week - Maybe the most of them wasn´t so complex, but surely above the needings for the most hobbyists.
In this time I never had the case that the way lecroy does it´s business won´t fulfill our needs.
Therefore I won´t blame the managing of the memory depth like siglent does as a true failure.
There must have been advantages and disadvantages to do so.
And we´re talking about 200Mpts memory maximum.
Have a look to a keysight DSOX2014a....
Costs more than double, having 100Kpts Memory.
Or tektronix TDS2014C....Costs over 3000 and have 2.5 kpts memory .
Or lecroy WS3014z..... 10Mpts Memory and costs 4486€....
So what we´re talking about…
Spin the time-wheel for saying 10yrs backwards...which memorydepth was possible...and how much did it costs…
How could engineers and technicians lives with it ?
Keeping this in mind for me the actually discussion is somekind of worthless.

[nctnico]

Somehow you managed to pick all the oscilloscopes with the least amount of memory in their price range. The Megazoom ASICs from Keysight date back to about 20 years ago and for the time they had superior memory depths. And there is a whole slew of scopes with deep memory nowadays: Tektronix TBS2000 series has 10Mpts per channel IIRC, R&S RTB2000 has 20Mpts, GW Instek GDS1000B and GDS2000E series have 10Mpts (20Mpts in segmented mode), Yokogawa DLM3000 series has 50Mpts in their base model (up to 500Mpts/channel), etc.

We are not living 10 years ago. We live now and large memory makes life easier. Acquire once and analyse is much quicker compared to acquiring multiple times and try to solve a puzzle from different measurements. Sure people managed to solve problems in the past too but at the expense of needing more time. And the latter is where my point is at: good tools save time and you should always be looking for better tools even if you are satisfied with the tools you have. You might stumble onto something which is better. I upgrade my equipment (which I need/use to make a living) all the time in order to work more efficiently.

Over the years I have owned several high-end logic analysers. The first one had 1kpts or 2kpts per channel and no 'compression' so it really needed carefully crafted trigger conditions and making screendumps to get a complete picture of a puzzle. The next one had 128kpts per channel. A huge improvement because it allowed making longer acquisitions and doing analysis on the signals where cause & effect where related. The next one has 32Mpts per channel and compression. I can capture several video frames of data to a TFT screen with 1ns resolution with it. Doing one capture gets me all the timing data I ever need to diagnose the interface.

[Martin72]

Somehow you managed to pick all the oscilloscopes with the least amount of memory in their price range.

Not somehow and you know it.

Remember the price of a SDS2104X Plus ?

Right, appx 1400€ incl. tax….
What do you get from the so called A-Brands for this money ?
And therefore again, what we´re talking about....

1400 might be very, very low cost, but for hobbyists, most of them, it would be a huge money to spend for.
When we discuss and compare, we always should compare in the pricerange it costs.
I grab the models which nearly came close to the specs of the new siglent.
You might working with "high end models", to having this point of view in general.
Again, what do you get from the A-Brands for 1400€ ?

[nctnico]

Well... R&S has the RTC1000 series and an RTB2002 is in a similar price bracket. Tektronix TBS2000 series is also very nice to use although no decoding. But Siglent isn't an A-brand so it would be comparing apples and oranges anyway. Other choices with mature firmware would be GW Instek or MicSig (I own scopes from the last two brands). I would take any of these over the SDS2000X just because of the automatic memory depth and clunky decoding.

[Martin72]

Well... R&S has the RTC1000 series and an RTB2002 is in a similar price bracket. Tektronix TBS2000 series is also very nice to use although no decoding.

RTB2000, 100Mhz, 4 channel will cost about double.

RTC series....got only 2 channels and small display, no comparison.
Tektronix TBS 2104( 4 channel , 100Mhz)...costs nearly double than the siglent (2200 against 1400) .

I can understand your intentions, but be aware of what we´re talking about.

[nctnico]

Again: you can't compare Siglent to A-brand pricing. For starters the service level of the A-brands is in a completely different league. If you want to compare the SDS2000X with A-brand prices then compare the price Lecroy asks for the SDS2000X with other A-brands.