Author Topic: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004  (Read 807721 times)

0 Members and 1 Guest are viewing this topic.

Offline maxspb69

  • Regular Contributor
  • *
  • Posts: 156
  • Country: ru
Re: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004
« Reply #2875 on: January 15, 2021, 01:54:33 pm »
Now you understand why I bought both of them in the end ?! Each of them has its own disadvantages and advantages :)

Two different scopes are much closer to ideal than one. Even if this one scope  is Rohde&Schwarz! :)
« Last Edit: January 15, 2021, 01:57:01 pm by maxspb69 »
 
The following users thanked this post: 2N3055

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26682
  • Country: nl
    • NCT Developments
Re: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004
« Reply #2876 on: January 15, 2021, 06:46:32 pm »
Spectrum analysis option on RTM3000 should use more advanced algorithms borrowed from realtime SA (R&S should know how that's done!) like overlaped FFT and many other tricks. I would be surprised if that one shows same problems.
I'm hanging on the edge of my seat to see what Nico reports from his RTM3k. And expect to be jealous.
No. I already did some tests on the RTM3004 and it shows exactly the same behaviour. The RTM3004 is pretty similar to the RTB2004 where it comes how the functionality is implemented.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: 2N3055

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 6407
  • Country: hr
Re: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004
« Reply #2877 on: January 15, 2021, 07:20:00 pm »
Spectrum analysis option on RTM3000 should use more advanced algorithms borrowed from realtime SA (R&S should know how that's done!) like overlaped FFT and many other tricks. I would be surprised if that one shows same problems.
I'm hanging on the edge of my seat to see what Nico reports from his RTM3k. And expect to be jealous.
No. I already did some tests on the RTM3004 and it shows exactly the same behaviour. The RTM3004 is pretty similar to the RTB2004 where it comes how the functionality is implemented.

That is weird.. Definitely needs investigation. Not something I would expect from R&S, really..
I will try to recreate same circumstances on my equipment.
 

Offline maxspb69

  • Regular Contributor
  • *
  • Posts: 156
  • Country: ru
Re: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004
« Reply #2878 on: January 15, 2021, 09:06:37 pm »
Having experimented more with two oscilloscopes (SDS2K+ & RTB2000), I seem to understand why SDS2K+ shows phantom peaks much less often than RTB2000 (although this can also be achieved on SDS2K+). The fact is that the SDS2K+ keeps the maximum sampling rate and decreases it only when there is not enough memory (and it has 10 times more memory). Plus SDS2K+ allows you to control the timebase (hence the length of the captured signal) in the FFT mode. These two facts allow it to display a much more real spectrogram.
RTB2000 really needs the ability to manual control the sampling rate/timebase in the FFT mode, then there will be less problems. RTB reduces the sampling rate unacceptably low when narrowing the span, as a result, if I understand correctly,  aliasing and the display of non-existent harmonics appear in some situations.
This is a payment for a user-friendly interface "like on a spectrum analyzer"
« Last Edit: January 15, 2021, 09:08:34 pm by maxspb69 »
 

Offline Zlotnik

  • Regular Contributor
  • *
  • Posts: 59
  • Country: nl
Re: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004
« Reply #2879 on: January 15, 2021, 09:10:43 pm »
No. I already did some tests on the RTM3004 and it shows exactly the same behaviour. The RTM3004 is pretty similar to the RTB2004 where it comes how the functionality is implemented.

Do you mean the control of sampling rate in fft mode, or the spurs?
 

Offline Zlotnik

  • Regular Contributor
  • *
  • Posts: 59
  • Country: nl
Re: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004
« Reply #2880 on: January 15, 2021, 09:19:05 pm »
The RTB2000 really needs the ability to manual control the sampling rate/timebase in the FFT mode, then there will be less problems. RTB reduces the sampling rate unacceptably low when narrowing the span, as a result, if I understand correctly,  aliasing and the display of non-existent harmonics appear in some situations.
This is a payment for a user-friendly interface "like on a spectrum analyzer"

Agreed. I think this can be done with a "SA like" UI though: the RTB displays the current sampling rate. You could change it by touching it, and once the record length is too long for the fft, reduce the RBW.

However, there seems to be more going on. Some of the examples above were not undersampled, I think. So where do the spurs come from, that appear for signals close to integral (rational?) fractions of the sample rate?
 

Offline maxspb69

  • Regular Contributor
  • *
  • Posts: 156
  • Country: ru
Re: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004
« Reply #2881 on: January 15, 2021, 09:25:51 pm »
My experiments show that spurs appear not only when the signal frequency is close to whole fractions of the sampling rate. Spurs appeared both at 1 MHz and at 960 kHz and at 871 kHz for the UpRamp signal (FFT parameters: span 12 MHz and starting frequency 0). There are some ratios of frequencies that RTB2000 "does not like very much."  ;)

For comparison: I managed to get spurs on the SDS2K+, but there I had to work hard to find the frequency at which they appeared. But in any case, they instantly disappear as soon as you change the timebase by 1 step. (SDS2K+ not allow to manage sampling rate, but keeps it as maximum as possible - 2Gs/s in almost the entire range of timebase)
« Last Edit: January 15, 2021, 09:32:46 pm by maxspb69 »
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26682
  • Country: nl
    • NCT Developments
Re: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004
« Reply #2882 on: January 15, 2021, 09:43:49 pm »
No. I already did some tests on the RTM3004 and it shows exactly the same behaviour. The RTM3004 is pretty similar to the RTB2004 where it comes how the functionality is implemented.

Do you mean the control of sampling rate in fft mode, or the spurs?
Both. There is no control of sampling rate so you'll get the same spurs if a signal is undersampled. I can do some tests if someone comes up with a specific test scenario.
« Last Edit: January 15, 2021, 10:05:01 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Dunckx

  • Contributor
  • Posts: 32
  • Country: gb
Hi Folks,

Thanks to friendly forumites I was greatly aided in my struggle to decide which digital scope I should buy (I've used a dual channel CRT scope for 35 years and this will be my first digital) and had just reached the point where the RTB2002 was top of the list and after I get paid a week from now, was intending to buy one.

I was reading through the literature (always a good idea to RTFM) and found something I could use clarification on.  On page 4 of the RTB2000 Specifications datasheet under Vertical System "maximum input voltage" I see the following: "300V (RMS), max. 400V (Vp) derates at 20dB/decade to 5V (RMS) above 250kHz."

Does this really mean that if I feed it much over 14V pk-pk above 250kHz the magic smoke comes out?  It almost sounds like advertising copy: "Turns into a brick at the press of a button in under four microseconds!"

Surely not?  Please tell me I have this badly wrong.  I know there are such things as 10x probes, but I am exactly the kind of bloke who forgets to switch the probe to 10x AND WHO CONTINUES TYPING WITHOUT REALISING CAPS LOCK IS ON.  When I inflicted this kind of thing unwittingly on my solid state CRT scope, it just sat there and grinned at me until I pulled my finger out.

Hmmmmmmmmmmmmm...
"God help us, we're in the hands of engineers." - Dr. Ian Malcolm, Jurassic Park
 

Offline KaneTW

  • Frequent Contributor
  • **
  • Posts: 805
  • Country: de
All scopes with a 1M||x pf input impedance have that kind of derating. And 50 Ohm scopes are something like 5Vp max. A 10:1/100:1 probe extends that range (see specs).

From both my experience and my interpretation, that derating starts at 250kHz and continues at 20dB/decade down to 5V RMS at about 20Mhz.
« Last Edit: March 14, 2021, 11:17:17 pm by KaneTW »
 
The following users thanked this post: Dunckx

Offline Dunckx

  • Contributor
  • Posts: 32
  • Country: gb
Gadzooks, I really am at the bottom of the learning curve!  I find it hard to believe that these new-fangled things will ever catch on...

Well thanks for that, forewarned is forearmed. :-+
"God help us, we're in the hands of engineers." - Dr. Ian Malcolm, Jurassic Park
 

Offline goaty

  • Regular Contributor
  • *
  • Posts: 202
  • Country: de
RTB2004 history mode does not seem to trigger right for me.

On a single digital signal (WS2812) on logic input 0, trigger set to D0, TTL, History size is 40 with some megasamples, and I have 15 bursts of serial data, I get less then that in history most of the time.

Some segments even are not shown as triggered at the beginning of the burst, but in the middle.

Scope is in Normal mode.

When capturing the whole bursts in one go (timebase according), I can confirm I have always 15 bursts.

Fast capture mode or wassitcalled, does not help.

Strange.

When is firmware 2.4 coming ? ;-)
 

Offline goaty

  • Regular Contributor
  • *
  • Posts: 202
  • Country: de
I was thinking that "Nx Single" lets me set the number of segments when in Single mode, but actually wanting more than one segment, no ?
In Normal mode, I thought, the history gets filled nonetheless. The gaps between bursts (packets) are rather wide, so I shouldn´t even need fast capture mode (2.5us blind time).

Guess I have to visit them in Mühldorfstraße, it´s not far ;-)
« Last Edit: April 22, 2021, 06:45:52 am by goaty »
 

Offline goaty

  • Regular Contributor
  • *
  • Posts: 202
  • Country: de
I´ll repeat it tonight and make a screen capture. I´ll try Auto mode instead, but thought it might not matter to have Normal mode.
I guess I´m doing somthing wrong, maby trigger level, that number of segments varies. Just haven´t found out yet.
Thanks für trying out.
 

Offline goaty

  • Regular Contributor
  • *
  • Posts: 202
  • Country: de
This segmented / history is seriously broken. I thought we talk about 2.5us delay only between re-arm when using fast mode ?
Firmware 2.3

I try to capture this waveform in segments. I can expect about 21 segments.
Whole data at 20 and 50 ms, it´s always like this, not more not less.
1213815-0
1213817-1

First try with history, still running shows this:
Three segments ? seriously ?
No fast segmentation.
One cannot scroll, because its not stopped.
1213819-2

Now try with fast segmentation:
twelve segments, but look at the delay !
1213821-3

Another try:
seventeen segments. What is this thing doing for 43ms on segment -7 ?
Missing my frames !
1213823-4

It´s unusable.
I tried auto mode, which does of course not work at all, because it´s constantly filling history and loosing my data immediatly.
Then manual mode, see results above.
Also tried single with N=20, which shows exactly the same.

Also tried not sinx/x to no avail.

I think my unit is defective.

Or maby not, but disappointing nonetheless. It seems to only (!) work in single trigger and fast segmentation.
I mean this is bursts with 10ms distance, and it cannot write it to memory only in single and fast segmentation ?

Very poor performance R&S

More tests with Trigger to Aux Out show no more than 5ms / 200Hz whatever I configure.
I thought it´s the measurements, but no.
So basically R&S 50000 waveforms per second is 200.
« Last Edit: April 22, 2021, 08:24:11 pm by goaty »
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 6407
  • Country: hr



You should try putting memory (Record length) management in Auto mode...
I keep repeating that manually managing memory is pain in the ass..
It should be left for special purposes only.

If you set your scope to FIXED 6Msps and sample at 2.5GS/sec every capture will be fixed and forced to be how long?
2.4 ms.....

Set it to Auto and try again...
Let us know what happened.
Regards,
Sinisa
« Last Edit: April 22, 2021, 08:25:25 pm by 2N3055 »
 
The following users thanked this post: goaty

Offline goaty

  • Regular Contributor
  • *
  • Posts: 202
  • Country: de
Ok, I made progress.
Something was wrong with the internal memory of the scope I guess.
I did a secure erase and times are now down to more reasonable values.
Have to do more tests.

Before I was stuck at 5ms/200Hz. I tried everything.
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26682
  • Country: nl
    • NCT Developments
Ok, I made progress.
Something was wrong with the internal memory of the scope I guess.
I did a secure erase and times are now down to more reasonable values.
Have to do more tests.

Before I was stuck at 5ms/200Hz. I tried everything.
Likely the memory length is now set to 'auto' instead of a (long) fixed length. If you need high capture rates then set the memory length to a small amount or auto.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: goaty

Offline goaty

  • Regular Contributor
  • *
  • Posts: 202
  • Country: de
Boy this is implemented way too complicated.

I managed to capture now all segments. But you have to

- Have the record length in auto (ok, understood this, if you capture too much it takes too much time)
- Be in normal trigger mode (Why ?)
- Arm a single trigger
- Have the Nx set to larger or equal the number of expected segments.
- Have fast capture mode on (it does not work without even with my 10ms between segments)

Everything else will give you a couple hundred waveforms per second max.

Why is there the huge delay between segments ?
Basically they show in comercial that the history is always on and you can go back in time,
but with that update rate you will miss everything but very slow events.

Also, why not allow fast capture in normal mode ??
It only works in single !

Argggh.

P.S. I have set acquisition depth mostly to auto mode, see pictures where it show 2.5GS/s and 6MSamples.
« Last Edit: April 22, 2021, 09:42:26 pm by goaty »
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26682
  • Country: nl
    • NCT Developments
Boy this is implemented way too complicated.

I managed to capture now all segments. But you have to

- Have the record length in auto (ok, understood this, if you capture too much it takes too much time)
- Be in normal trigger mode (Why ?)
- Arm a single trigger
- Have the Nx set to larger or equal the number of expected segments.
- Have fast capture mode on (it does not work without even with my 10ms between segments)
This is basically how you setup a segmented recording on any oscilloscope. The difference between 'fast sequence' and 'normal sequence' mode is that in the latter the display is updated as well so you can see what is actually being captured instead of looking at a counter going up without any idea whether the data is useful or not. Then again, if your events come in fast you don't waste much time in case you need to re-capture so being able to see what is going on is less of an issue.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: goaty

Offline goaty

  • Regular Contributor
  • *
  • Posts: 202
  • Country: de
Peter thanks for your extensive tests.
I do not use the internal generator of the scope.

Also I think if Rigol can do fast capture in Normal trigger mode, so should R&S.
That was dumb, it´s the generator display, sorry.

I did not manage what you did in the second picture (RTB2004-PulseBurst-1-21-Sample.png), where there is 10ms delta and fast-capture off.
There, I get huge variations in trigger time. I think it might be if I have a second channel on, then off again,
which I did to measure something on the other input. I switched trigger and channel on and off, and
sometimes got more than 100ms trigger arm time on the counter (aux out is trigger).

What firmware version are you running ?

I´ll probably write R/S support to see if they have anything to say to the issues:
- When Auto-Trigger, a segmented single does not wait for trigger but automatically triggers every 500ms
- Huge variations in re-arm even if dot-mode and almost everything switched off.

Edit: Fair enough they write in the manual that fast segmentation only works for single trigger.
« Last Edit: April 23, 2021, 07:36:01 am by goaty »
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 6407
  • Country: hr
@Goaty

there are few things you don't seem to understand about how scope works. Any scope. And then you need to think about some of your statements and you will find them illogical.
New generation scopes are very sophisticated (both powerful and somewhat complicated) and takes time to get used to it.

If you allow me to explain:

Auto trigger mode is behaving EXACTLY as it should. It waits for trigger for some short time and then if there was no trigger event, it AUTO triggers without reason. It is equivalent of you pressing "force trigger" every few hundred milliseconds.  To wait for trigger unconditionally you use NORMAL mode.  (People get confused that AUTO means scope will magically discover appropriate trigger settings by itself. It doesn't. It simply takes blind capture every now and then. Purpose of that is that scope will show something on the screen even if you setup trigger wrong or there is no trigger event)
From RTB 2000 Manual:

"Auto" The instrument triggers repeatedly after a time interval if the trigger
conditions are not fulfilled. If a real trigger occurs, it takes precedence. This mode helps to see the waveform even before the trigger
is set. The waveform on the screen is not synchronized, and successive waveforms are not triggered at the same point of the waveform.

"Norm" The instrument acquires a waveform only if a trigger occurs, that is, if
all trigger conditions are fulfilled. If no trigger occurs, no waveform is
acquired and the last acquired waveform is displayed. If no waveform
was captured before, nothing is displayed.

I said on many occasions that it is unfortunate that what is called AUTO trigger should be called TIMEOUT trigger and there would be less confusion..

You should really read a manual from front to back, these scopes are sophisticated enough that simply playing with it is not enough.

Segmented mode (fast one) is always single burst. How else would it be?   You set it to capture 20 segments, and it does that and stops. Otherwise it wouldn't be 20 i you let it run forever. Also if you press SINGLE or RUN it is the same. It would capture next 20 trigger events (without screen updates) and then stop. On Keysight you can press SINGLE or RUN and it is the same.

When you use History mode (segmented with screen updates) retrigger time will be somewhat jittery. Reason is screen refresh, that happens on regular interval for screen, and it will make a scope "pause" a little while screen is being rendered. It takes time to take all that data from buffer, decimate it from megapoints to kilopixels and render it for screen. And those little "pauses" will happen at seemingly "random" intervals in correlation to your signal, because those two events are not correlated.
Because of that, there is "fast" segmented mode (one without screen updates) to make retrigger time and captures deterministic and as fast as possible.

And as Nico said, if you need retrigger time in msec, by the time you blink, you'll have hundreds of captured buffers already there waiting.

« Last Edit: April 23, 2021, 08:15:32 am by 2N3055 »
 
The following users thanked this post: Cthulhoid, goaty

Offline goaty

  • Regular Contributor
  • *
  • Posts: 202
  • Country: de
I did not manage what you did in the second picture (RTB2004-PulseBurst-1-21-Sample.png), where there is 10ms delta and fast-capture off.
There, I get huge variations in trigger time.

Maybe activated measurements are to blame? They also need time and probably slow down the trigger.

Peter
Ah no, I have disabled all and it did not change.
 

Offline goaty

  • Regular Contributor
  • *
  • Posts: 202
  • Country: de
@Goaty
Auto trigger mode is behaving EXACTLY as it should. It waits for trigger for some short time and then if there was no trigger event, it AUTO triggers without reason. It is equivalent of you pressing "force trigger" every few hundred milliseconds.  To wait for trigger unconditionally you use NORMAL mode.  (People get confused that AUTO means scope will magically discover appropriate trigger settings by itself. It doesn't. It simply takes blind capture every now and then. Purpose of that is that scope will show something on the screen even if you setup trigger wrong or there is no trigger event)
From RTB 2000 Manual:

"Auto" The instrument triggers repeatedly after a time interval if the trigger
conditions are not fulfilled. If a real trigger occurs, it takes precedence. This mode helps to see the waveform even before the trigger
is set. The waveform on the screen is not synchronized, and successive waveforms are not triggered at the same point of the waveform.

"Norm" The instrument acquires a waveform only if a trigger occurs, that is, if
all trigger conditions are fulfilled. If no trigger occurs, no waveform is
acquired and the last acquired waveform is displayed. If no waveform
was captured before, nothing is displayed.

I said on many occasions that it is unfortunate that what is called AUTO trigger should be called TIMEOUT trigger and there would be less confusion..
Thank you for clarifying this. I was under the impression that in Auto, if there is no trigger, I still see the waveform on the screen (capture), but it is untriggered all the time (like freerunning).
Also I do not see the trigger LED, which led me to believe, that also the history must therefore be NOT filled.

@Goaty
Segmented mode (fast one) is always single burst. How else would it be?   You set it to capture 20 segments, and it does that and stops. Otherwise it wouldn't be 20 i you let it run forever. Also if you press SINGLE or RUN it is the same. It would capture next 20 trigger events (without screen updates) and then stop. On Keysight you can press SINGLE or RUN and it is the same.

When you use History mode (segmented with screen updates) retrigger time will be somewhat jittery. Reason is screen refresh, that happens on regular interval for screen, and it will make a scope "pause" a little while screen is being rendered. It takes time to take all that data from buffer, decimate it from megapoints to kilopixels and render it for screen. And those little "pauses" will happen at seemingly "random" intervals in correlation to your signal, because those two events are not correlated.
Because of that, there is "fast" segmented mode (one without screen updates) to make retrigger time and captures deterministic and as fast as possible.

And as Nico said, if you need retrigger time in msec, by the time you blink, you'll have hundreds of captured buffers already there waiting.
If I press SINGLE it captures what I set with Nx=..., then stops. If I press RUN, it does not stop for me, it runs continuously. Or did I misunderstand ?

I understand it will jitter because it´s probably not fully asynchronous, which it IMHO shuld be (acquisition and display processing),
but I was mislead by the "up to 50000wfms/s" which I have not been able to even closely reach.
It does not matter in most cases, but for trying to capture bursts with long pauses, one has to know the re-arm time, or use fast-capture. But then you have to know the Nx...

I guess I have a lot more reading and experimenting to do.
Thank you for taking the time to write !!
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 6407
  • Country: hr

Thank you for clarifying this. I was under the impression that in Auto, if there is no trigger, I still see the waveform on the screen (capture), but it is untriggered all the time (like freerunning).
Also I do not see the trigger LED, which led me to believe, that also the history must therefore be NOT filled.

Yes, it is basically freeruning. Not exactly, it will wait for trigger event for a short time, and then give up and fire off capture just because....  ^-^ But basically freerunning.


If I press SINGLE it captures what I set with Nx=..., then stops. If I press RUN, it does not stop for me, it runs continuously. Or did I misunderstand ?

I understand it will jitter because it´s probably not fully asynchronous, which it IMHO shuld be (acquisition and display processing),
but I was mislead by the "up to 50000wfms/s" which I have not been able to even closely reach.
It does not matter in most cases, but for trying to capture bursts with long pauses, one has to know the re-arm time, or use fast-capture. But then you have to know the Nx...

I guess I have a lot more reading and experimenting to do.
Thank you for taking the time to write !!

The difference between SINGLE and RUN with segments is actually R&S unique feature (not a bug  ^-^).

With all other scopes you capture 20 segments and stop. Same as you get with SINGLE on R&S. But on R&S, if you press RUN, it will not stop after 20 segments, but it will keep acquiring circular buffer of LAST 20 segments. So at any time, you STOP, and in buffer you will have last 20 triggers before STOP.  It's kinda like counting forward and backward...

As far as WFM/sec that is most bogus thing in industry... Most of the scopes achieve those published numbers only in specific settings, and it is not really important.  Blind time, (retrigger time) is important and very few publish it at all, and those that do don't publish comprehensive tables of all possible combination. So it is up to user to characterize it if they need it to be exact.

You're welcome, not a problem, as long as I have some time...  ^-^
 
The following users thanked this post: goaty


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf