Keysight DSOx2004 vs. Siglent SDS1204X-E

[Gandalf_Sr]

One aspect of scopes that often isn't focused on enough is screen update rate or waveforms per second (WPS); Dave has done a review of the Keysight DSOX2000 series ().  Anyway he shows how WPS affects what you see on the scope and more is definitely better, the point being that you may have a big memory chip inside the scope that stores data at 10,000 screenfulls per second but if the screen only shows you 100 WPS then the chance of seeing a pulse that happens once every blue moon is poor.  So even the Rigol DS1054Z with (upgraded) 24 MPts of memory will not do as well as the DSOx2004 in real world tests when trying to see a glitch because it lacks the processing power to update the screen as fast as the KS - at least that's my understanding - Manufacturers of course like to pick numbers to say "my brand has more <X> than the other brand".  If you're trying to capture long sequences like serial data streams then lack of deep memory obviously limits you but segmented memory was invented to cure that, it only uses the memory when there's actual data to be captured.  IMHO, a scope is only really good for checking the signal integrity of a serial data stream and being able to trigger on a certain pattern is useful but to start tracking what's happening with the actual data, you need a real logic analyzer like a Salea Logic 8/16.

[tautech]

It's not about raw performance but the tools you have at your disposal Ted.
A good trigger suite and color grading can help immensely like in the simple exercise I showed here:
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg1370717/#msg1370717
 

[nctnico]

One aspect of scopes that often isn't focused on enough is screen update rate or waveforms per second (WPS); Dave has done a review of the Keysight .  Anyway he shows how WPS affects what you see on the scope and more is definitely better, the point being that you may have a big memory chip inside the scope that stores data at 10,000 screenfulls per second but if the screen only shows you 100 WPS then the chance of seeing a pulse that happens once every blue moon is poor.  So even the Rigol DS1054Z with (upgraded) 24 MPts of memory will not do as well as the DSOx2004 in real world tests when trying to see a glitch because it lacks the processing power to update the screen as fast as the KS

Waveforms/s doesn't depend on the processing power. (Almost) every scope uses an ASIC of FPGA to draw the traces. There are three issues with waveforms/s:
1) The waveforms/s number depends on samplerate and memory depth so it isn't constant.

2) Tests about waveforms/s are more often then not skewed to show more waveforms/s are better in very particular situations.  If you slow down timebase you'll capture more signal and thus increase the chance you'll see the glitch.

3) A high waveforms/s number doesn't guarantee capturing a glitch. A trigger condition does.

Deeper memory however is not something you can work around easely. Segmented recording is handy to capture rare events which happen at long time intervals (say less than 1 times per second) but it isn't a replacement for deep memory because you loose the context of the segments. You can usually look through them one-by-one but the big picture is missing. All in all I'd recommend to get an oscilloscope with at least 10Mpts per channel with all channels enabled.

[0xfede]

One aspect of scopes that often isn't focused on enough is screen update rate or waveforms per second (WPS); Dave has done a review of the Keysight .  Anyway he shows how WPS affects what you see on the scope and more is definitely better, the point being that you may have a big memory chip inside the scope that stores data at 10,000 screenfulls per second but if the screen only shows you 100 WPS then the chance of seeing a pulse that happens once every blue moon is poor.  So even the Rigol DS1054Z with (upgraded) 24 MPts of memory will not do as well as the DSOx2004 in real world tests when trying to see a glitch because it lacks the processing power to update the screen as fast as the KS

Waveforms/s doesn't depend on the processing power. (Almost) every scope uses an ASIC of FPGA to draw the traces. There are three issues with waveforms/s:
1) The waveforms/s number depends on samplerate and memory depth so it isn't constant.

2) Tests about waveforms/s are more often then not skewed to show more waveforms/s are better in very particular situations.  If you slow down timebase you'll capture more signal and thus increase the chance you'll see the glitch.

3) A high waveforms/s number doesn't guarantee capturing a glitch. A trigger condition does.

Deeper memory however is not something you can work around easely. Segmented recording is handy to capture rare events which happen at long time intervals (say less than 1 times per second) but it isn't a replacement for deep memory because you loose the context of the segments. You can usually look through them one-by-one but the big picture is missing. All in all I'd recommend to get an oscilloscope with at least 10Mpts per channel with all channels enabled.

I completely agree with nctnico. To expand this view a little further I must say that when I turn on the scope is because I don't know what I'm looking for. If I'm able to set a peculiar/complex trigger is because I already know exactly what is going on and therefore the DSO is not useful anymore.
Deep memory is very important and I wouldn't add a fixed limit to my wish list; it is way better to calculate memory/samplerate that gives the time that the sample buffer have before it gets filled. 

Best,
0xfede

[2N3055]

Manufacturers of course like to pick numbers to say "my brand has more <X> than the other brand".

You are so correct. And then you fall into same trap and buy Keysight sales pitch...

Problem is that you cannot do some things if your memory is short, segmented or not.
Like there is no replacement for higher bandwidth and sample rate, there is no replacement for memory depth.

Short retrigger time is very good because you have less blind time.
So if you enable long persistence, you will very quickly see on the screen that since you started capturing, there  is something that is not supposed to be there...

And that is great advantage only for some frequencies of glitch appearance.. If glitch is repeating faster than some frequency, it will be detected immediately on fast retrigger scope, but on slower one in a few seconds. It will be slower, but not in a way that will impact your work.
It will also depend on time base...

Of course, you will have no clue when it happened, what caused it and what is time correlation and so on.. Then you have to devise clever ways how to figure that out.. You then have to start trying different triggers until you get to be able to catch it.

But all that stuff only works on small subset of signals you work with.. simple, highly repetitive stuff..  Clocks are good example...
And glitch has to be repetitive, and you have to be able to recapture signal over and over until you catch it.
What happens if you have a glitch that is seemingly random...

On the other hand, i have a scope that can capture 0.2 seconds worth of signal at 1GSa/sec. It can also do fast segmented capture without screen updates at 1.3M/Wfms second..
Then I run analysis on that particular capture, in one way, and another way, and another... I can do it till' the chickens come home, go to sleep and resume doing it next day and keep on goin' like that for weeks... You can send someone to other town to do the capture, and you analyse it in your home office.

Despite what either fanboys will tell you, neither way is better. It depends on what you do, and how you do it...

So yes, it's not simple if you don't have a lot of money to just buy whatever you feel like at any moment...

Regards,

Sinisa

[Rich@RohdeScopesUSA]

I forgot to ask in my last post but if I do go with the agilent/keysight, do you guys think the lack of memory depth will affect my measurements seriously? From what I read it looks like keysight does their measurements and such post acquisition vs during it like the R&S. Correct me if I'm wrong

Bigger isnt always better.  Found a decent article to explain why.. gist of it is if the processor gets so bogged down if you go to fast and to deep your response will go to utter trash.  Much like all the other options, its best to read up on what that facet does and if its something you care about and if so what do you want out of it.

That said I've never really had to much issue with KS less im really pushing it hard but im usually just fooling about if im doing something like that... (i've managed to super bog some down to where i could go get coffee.. lol) we got a good selection of scopes to work with so mix and match does work out nicely by having a variety.

https://www.eetimes.com/document.asp?doc_id=1279463

Oh the irony - I wrote that article when I was still at Agilent 

In the end, the article's premise still holds true, but it was also written 6-years ago and some things have changed (in particular, FPGAs helping to give both fast update rate and deep memory without the need for an ASIC that limits you to one or the other).  In general, I tend to agree with the folks who have said that update rate is something you can make up for with additional time - memory can't be made up for, but you should really buy the scope that is best suited for your application(s) and budget.

Having said all that, W2DML - please let me know if you would like assistance in demoing a RTB2000.  I'm always a fan of trying out the scope you want to buy first.  There are also some additional features coming to the 2000 in the not too distant future (like advanced math). 

-Rich

[W2DML]

Rich I would definitely like to demo the rtb2000 if that's possible. Its currently the scope that I'm interested in the most.

Thanks,
Dan

[Elasia]

Oh the irony - I wrote that article when I was still at Agilent 

In the end, the article's premise still holds true, but it was also written 6-years ago and some things have changed (in particular, FPGAs helping to give both fast update rate and deep memory without the need for an ASIC that limits you to one or the other).  In general, I tend to agree with the folks who have said that update rate is something you can make up for with additional time - memory can't be made up for, but you should really buy the scope that is best suited for your application(s) and budget.

Having said all that, W2DML - please let me know if you would like assistance in demoing a RTB2000.  I'm always a fan of trying out the scope you want to buy first.  There are also some additional features coming to the 2000 in the not too distant future (like advanced math). 

-Rich

ROFLMAO... small world!     Yeah, FPGA tech has been amazing what they can do now for the price... really interested to see what this siglent does whenever it manages to come in.

Edit: Should throw this out there as well, cant go wrong with RS.  Some of our guys doing remote IoT devices (as in a mobile industrial machine out in the boonies or bfe) use their gear and its good stuff.

[Rich@RohdeScopesUSA]

Rich I would definitely like to demo the rtb2000 if that's possible. Its currently the scope that I'm interested in the most.

Thanks,
Dan

PM'd.

-Rich

[bicycleguy]

Rich I would definitely like to demo the rtb2000 if that's possible. Its currently the scope that I'm interested in the most.

Thanks,
Dan

Fry's Electronics in Ca. USA is still having a clearance sale on there Keysight stuff.  $1374 for a DSOX2014A.  Back in Nov after visiting several stores, I found an open box of same and talked them down to $999.  It was open box because the store had opened it to take two sets of probes to replace stolen ones on the 2 channel floor model.  You can try the floor models if they have any left.  They seemed anxious to get rid of them back then, don't know about now.

Thanks to EEvblog for an additional < $40 for a lan card kit and a few hours studying (more like days) I have a 200mHz 4 channel all options enabled scope!

[W2DML]

That sounds like an awesome deal, sadly there is no Fry's in Long Island. Our Microcenter which is similar doesn't even carry oscilloscopes...

[W2DML]

I just wanted to come back and say I decided to go with the rtb2004. Thanks to Rich, newark, R&S, and everyone here for helping me out. Definitely recommend the scope, it's awesome

[Rich@RohdeScopesUSA]

I just wanted to come back and say I decided to go with the rtb2004. Thanks to Rich, newark, R&S, and everyone here for helping me out. Definitely recommend the scope, it's awesome

Dan - glad everything worked out in the end.  Enjoy your new scope! 

-Rich