Author Topic: new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004  (Read 916476 times)

0 Members and 5 Guests are viewing this topic.

Offline exe

  • Supporter
  • ****
  • Posts: 2631
  • Country: nl
  • self-educated hobbyist
Just got mine, looks good. I can't wait to get some spare time and integrate it into Matlab to add the VSA functionality :-+ :popcorn:
Not a 14 bit front end, but still, for low order modulations it should work pretty well with the fast ETH interface!

Could you please elaborate what's so special about 10bit DAC? I tried to find any pictures to comparison, but didn't find any _convincing_ evidence showing that 10-12bit scopes make a huge difference. I also noticed in some specs there are "effective number of DAC bits" which is often not even 8! So, looks like it's not always possible to make effective use of all the DAC bits available.

I'm asking because I'm making a decision -- to buy a cheaper scope from GW Instek (like gds-2074e), or to go with rtb2004. So, any pictures are welcome!
 

Offline coppice

  • Super Contributor
  • ***
  • Posts: 9566
  • Country: gb
Just got mine, looks good. I can't wait to get some spare time and integrate it into Matlab to add the VSA functionality :-+ :popcorn:
Not a 14 bit front end, but still, for low order modulations it should work pretty well with the fast ETH interface!

Could you please elaborate what's so special about 10bit DAC? I tried to find any pictures to comparison, but didn't find any _convincing_ evidence showing that 10-12bit scopes make a huge difference. I also noticed in some specs there are "effective number of DAC bits" which is often not even 8! So, looks like it's not always possible to make effective use of all the DAC bits available.

I'm asking because I'm making a decision -- to buy a cheaper scope from GW Instek (like gds-2074e), or to go with rtb2004. So, any pictures are welcome!
Going from 8 bits to 10 bits or more doesn't make much difference if you only ever look at waveforms on the screen. If you process those waveforms, or use the scope as a capture device to feed waveforms to an external device (like the above user wanting to feed waveforms into Matlab) it can make a big difference.
 

Offline Robaroni

  • Frequent Contributor
  • **
  • Posts: 434
  • Country: us
  • Retired EE
    • Design Specialties
Just got mine, looks good. I can't wait to get some spare time and integrate it into Matlab to add the VSA functionality :-+ :popcorn:
Not a 14 bit front end, but still, for low order modulations it should work pretty well with the fast ETH interface!

Could you please elaborate what's so special about 10bit DAC? I tried to find any pictures to comparison, but didn't find any _convincing_ evidence showing that 10-12bit scopes make a huge difference. I also noticed in some specs there are "effective number of DAC bits" which is often not even 8! So, looks like it's not always possible to make effective use of all the DAC bits available.

I'm asking because I'm making a decision -- to buy a cheaper scope from GW Instek (like gds-2074e), or to go with rtb2004. So, any pictures are welcome!

It makes a noticeable resolution difference. 2 to the 8th = 256, 2 to the 10th = 1024 or 4 times better resolution. So it's much better, go with 10 bit or better 12 bit (4096 = 16 times better).
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 28111
  • Country: nl
    • NCT Developments
You also need to take the effective number of bits into account. There is no use comparing number of ADC bits without looking at the ENOB (which must also include the effects of the jitter of the sampling clock!).
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: Keysight DanielBogdanoff

Offline Fgrir

  • Regular Contributor
  • *
  • Posts: 164
  • Country: us
Could you please elaborate what's so special about 10bit DAC?

The RTB2K screen has 800 pixel vertical resolution so you can see at least 9-bits straight up.  In general for any higher resolution scope I also find the extra bits useful in the following situations:
  • Changing vertical scale after waveform capture, sort of like long memory in the other axis
  • FFT
  • Nonlinear math (If the RTB2K had any useful math capabilities  :()
  • Post-processing as mentioned by others

ENOB can be an issue but usually doesn't come into play until higher frequencies, there is a whole world of low-frequency analysis where the extra bits can be handy.
 
The following users thanked this post: Someone

Offline agdr

  • Supporter
  • ****
  • Posts: 105
  • Country: us
    • agdr Audio
There's an updated datasheet for the RTB out now, I compared it to the last one ...

UART decode spec changed: max bitrate officially reduced to 3 Mbps.   :-\

Wow! So Mike's discovery wasn't a typo.  Didn't seem possible they would design the scope with such a low rate. Hopefully that gets fixed in the rumored "pro" version of the scope. In fact that might be where many of Mike's suggestions wind up! :D
 

Offline exe

  • Supporter
  • ****
  • Posts: 2631
  • Country: nl
  • self-educated hobbyist
The RTB2K screen has 800 pixel vertical resolution so you can see at least 9-bits straight up.  In general for any higher resolution scope I also find the extra bits useful in the following situations:
  • Changing vertical scale after waveform capture, sort of like long memory in the other axis
  • FFT
  • Nonlinear math (If the RTB2K had any useful math capabilities  :()
  • Post-processing as mentioned by others

ENOB can be an issue but usually doesn't come into play until higher frequencies, there is a whole world of low-frequency analysis where the extra bits can be handy.

I got it, thanks.Still, I would be very grateful If someone could share any actual data. Like, two zoomed waveforms, one is smoth and nice from rtb2000, another one crude and steppy from an 8bit scope.

Concerning FFT, the difference is in the noise floor?
 

Offline Fgrir

  • Regular Contributor
  • *
  • Posts: 164
  • Country: us
I got it, thanks.Still, I would be very grateful If someone could share any actual data. Like, two zoomed waveforms, one is smoth and nice from rtb2000, another one crude and steppy from an 8bit scope.

Both waveforms acquired at 1V/div and then zoomed up to 200mV/div.
 
The following users thanked this post: exe, Octane, vokars

Offline TK

  • Super Contributor
  • ***
  • Posts: 1722
  • Country: us
  • I am a Systems Analyst who plays with Electronics
What happens if you capture both originally at 200mV/div?
 

Offline Fgrir

  • Regular Contributor
  • *
  • Posts: 164
  • Country: us
What happens if you capture both originally at 200mV/div?
 
The following users thanked this post: TK, vokars

Offline Fgrir

  • Regular Contributor
  • *
  • Posts: 164
  • Country: us
And while I'm at it, here is the same signal with each scope in High Resolution mode acquired at 5V/div, zoomed up to 200mV/div.
 
The following users thanked this post: Someone, edavid, exe, Octane, vokars

Offline Fgrir

  • Regular Contributor
  • *
  • Posts: 164
  • Country: us
Concerning FFT, the difference is in the noise floor?

The effect of more bits on FFT performance is more complex than that:
https://en.wikipedia.org/wiki/Quantization_(signal_processing)#Quantization_noise_model
 
The following users thanked this post: vokars

Offline exe

  • Supporter
  • ****
  • Posts: 2631
  • Country: nl
  • self-educated hobbyist
Thank you very much! May I ask you to capture a waveform of a, say, 10MHz sine or higher? That's because 1KHz sine is too simple task to capture. I'm curious if 10bits are preserved over the whole bandwidth.
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2238
  • Country: us
And while I'm at it, here is the same signal with each scope in High Resolution mode acquired at 5V/div, zoomed up to 200mV/div.
That some pretty strange distortion on the MSO X3024A.  Here's your same hi-res measurement on a MSO X3104.  Still noisier than the RTB2K, but not distorted.
 
The following users thanked this post: Fgrir, vokars

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 28111
  • Country: nl
    • NCT Developments
I used to get similar distortion on my Tek TDS744A if there wasn't enough noise to make the hi-res oversampling work correctly.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2238
  • Country: us
I used to get similar distortion on my Tek TDS744A if there wasn't enough noise to make the hi-res oversampling work correctly.
Interesting.  Going from 5V to 200mV makes it 1 bit per division, which pretty much matches the period of the distortion.  I think could also be some DNL in the ADC.
 

Offline Keysight DanielBogdanoff

  • Supporter
  • ****
  • Posts: 788
  • Country: us
  • ALL THE SCOPES!
    • Keysight Scopes YouTube channel
I would comment that ENOB is more important than the ADC's banner bit-depth spec. And, no scope will ever have ENOB=BoR because there's going to be noise. That's one of the industry's dirty secrets :).

Also, the capture-then-zoom-in approach will just highlight ENOB over normal ADC bits. To optimize the ADC you should zoom all your signals in full-screen when capturing, which is something I harp on a bit and have shown on our YouTube channel and our blog here:
http://bit.ly/2rVzs6n

If you're trying to compare ADCs, you should compare between full-scale measurements or else the scope noise is playing a bigger part of the measurement than normal. And, you should generally do that comparison at the smallest hardware V/div setting in the oscilloscope.

For High Resolution mode, you also want to make sure that the time/div setting is slow enough to take advantage of the HighRes oversampling mechanism.

I'm curious if 10bits are preserved over the whole bandwidth.

Generally, if a data sheet doesn't specify this, it'll cover the whole bandwidth. That or they're trying to hide a defect or something, but R&S doesn't play that game to my knowledge.


 
The following users thanked this post: egonotto, electrolust, vokars

Offline EEVblog

  • Administrator
  • *****
  • Posts: 38722
  • Country: au
    • EEVblog
Did Dave give up on a in depth review of this scope?

I have scopes which I got before this one that I haven't reviewed yet, there is a big backlog.
People underestimate how much work is involved in doing and "in depth" review of a scope like this, it is a massive task, you really have to be motivated to do it.
 

Offline mtdoc

  • Super Contributor
  • ***
  • Posts: 3575
  • Country: us
I would comment that ENOB is more important than the ADC's banner bit-depth spec. And, no scope will ever have ENOB=BoR because there's going to be noise. That's one of the industry's dirty secrets :).

I understand ENOB.  What's BoR?
 

Offline exe

  • Supporter
  • ****
  • Posts: 2631
  • Country: nl
  • self-educated hobbyist
I understand ENOB.  What's BoR?

May be Bits of Resolution?

PS thank you, guys, for detailed answers about ENOB! This helped me a lot.
 

Offline Fgrir

  • Regular Contributor
  • *
  • Posts: 164
  • Country: us
That some pretty strange distortion on the MSO X3024A.  Here's your same hi-res measurement on a MSO X3104.  Still noisier than the RTB2K, but not distorted.
That is interesting that yours is so much better.  I don't want to hijack the thread into a long discussion on the MSOX Hires mode so I won't post any more screens on this, but I get a similar result on all four input channels, with two different function gens, switching to 50-ohm termination, etc.  Maybe a difference on the 1GHz front-end?  My unit is a year out-of-cal, but I'd be surprised if cal would address this at all.  I'm inclined to agree with nctnico that there just isn't enough noise on my 5V/div range to make averaging useful.

To optimize the ADC you should zoom all your signals in full-screen when capturing

I agree, but this was just meant to illustrate the utility of having the extra bits available if you decide you need to zoom in after you've already captured a waveform.  I love my MSOX3024, but sometimes there is no substitute for dynamic range.
« Last Edit: May 26, 2017, 06:47:09 pm by Fgrir »
 

Offline Someone

  • Super Contributor
  • ***
  • Posts: 5016
  • Country: au
    • send complaints here
That some pretty strange distortion on the MSO X3024A.  Here's your same hi-res measurement on a MSO X3104.  Still noisier than the RTB2K, but not distorted.
That is interesting that yours is so much better.  I don't want to hijack the thread into a long discussion on the MSOX Hires mode so I won't post any more screens on this, but I get a similar result on all four input channels, with two different function gens, switching to 50-ohm termination, etc.  Maybe a difference on the 1GHz front-end?  My unit is a year out-of-cal, but I'd be surprised if cal would address this at all.  I'm inclined to agree with nctnico that there just isn't enough noise on my 5V/div range to make averaging useful.
Try doing a user cal, you'll need enough cables and splitters to make the octopus harness.
 

Offline blacksheeplogic

  • Frequent Contributor
  • **
  • Posts: 532
  • Country: nz
Didn't seem possible they would design the scope with such a low rate. Hopefully that gets fixed in the rumored "pro" version of the scope.

I don't think it's a bad scope, but with the current pricing it's now too close to the 3000T. If there is a pro version it will be interesting to see how they position it against the 3000T without lowering the current price of the 'non-Pro' version significantly.
 

Offline Rich@RohdeScopesUSA

  • Frequent Contributor
  • **
  • Posts: 460
  • Country: us
Didn't seem possible they would design the scope with such a low rate. Hopefully that gets fixed in the rumored "pro" version of the scope.

I don't think it's a bad scope, but with the current pricing it's now too close to the 3000T. If there is a pro version it will be interesting to see how they position it against the 3000T without lowering the current price of the 'non-Pro' version significantly.
I normally try not to intervene because this forum does such a good job of correcting misinformation itself, but your comment caught my attention. 

Clearly the RTB2000 was designed to compete with the 2000x and DPO2000 from Tek.  And with that you get a 10-bit ADC, significantly longer memory, a large, high resolution touch display, significantly lower noise, more SR and (available) bandwidth, 1mV/div in HW with full bandwidth, similar to better update rate, boots faster, standard ethernet, etc, etc.  There are few to no specs that those products beat the RTB2000 on and they are priced similarly. 

Even when comparing to the x3000T, the RTB2000 gives you a 4x the vertical resolution (10-bit ADC), significantly more memory, a larger, higher resolution display, lower noise, 1mV/div in HW with full bandwidth, standard ethernet and in most cases it costs half the price.  I wouldn't call that too close.  Does the x3000T have benefits?  Sure - more available bandwidth/SR, 50Ohm input and higher update rate.  But again, the RTB2000 wasn't designed to compete with the x3000T hence the reason it doesn't have those features. 

In the end, value is clearly a personal belief (of which I personally think we stack up extremely well on value), but I wouldn't say the current pricing is too close to the x3000T. 

-Rich
« Last Edit: May 27, 2017, 01:11:13 pm by Rich@RohdeScopesUSA »
 
The following users thanked this post: Someone

Offline TK

  • Super Contributor
  • ***
  • Posts: 1722
  • Country: us
  • I am a Systems Analyst who plays with Electronics
It looks like initial "I have to have it" impulse is fading as the last 3 RTB2004 units that showed up on eBay are still there and prices are being reduced.

R&S should have targeted the 3000T (for features, capabilities, measurement) and not the low end of the Keysight X Series like 2000X.  Don't forget that Keysight has the trade-in incentive program offering up to 30% discount and if you are lucky you can get one 1000X scope as well (as some forum member reported)

I think the current 2000X 2 channel pricing is in the $1000 territory, I saw some new units being sold by Fry's for $950 (local in store pick only), so maybe Keysight is getting rid of the 2000X inventory.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf