REVIEW - Rigol DS2072 - First Impressions of the DS2000 series from Rigol

[marmad]

Well they are listed as RP3300A.

I can't even find an RP3300A probe listed anywhere online - it's like a cheaper version of the RP3300 that they've produced specifically to include with the DSO. Kind of sucks for you, man. 

Perhaps they had it planned from the beginning - but didn't start producing them until later - so had to include the more expensive RP3300; I've no idea. But the online datasheet (from June 2012) lists the standard probe as RP3300 - and so does the User Manual (May 2012). So it might be you have a very early model.

Edit: What was your calibration date?

[darrylp]

Well they are listed as RP3300A.

I can't even find an RP3300A probe listed anywhere online - it's like a cheaper version of the RP3300 that they've produced specifically to include with the DSO. Kind of sucks for you, man. 

Perhaps they had it planned from the beginning - but didn't start producing them until later - so had to include the more expensive RP3300; I've no idea. But the online datasheet (from June 2012) lists the standard probe as RP3300 - and so does the User Manual (May 2012). So it might be you have a very early model.

Edit: What was your calibration date?

its a very new model.

dated 4th June 2013

the manual, has a satin feel to the outside cover... and I can only see my single set of finger prints on it...  I'd bet the manual had never been opened before,  the glued edge, hasn't had a crease to it yet.

and the english half of the manual, definitely says Mar 2012 .... of course cant read the other half ;-)

[marmad]

its a very new model.

dated 4th June 2013

the manual, has a satin feel to the outside cover... and I can only see my single set of finger prints on it...  I'd bet the manual had never been opened before,  the glued edge, hasn't had a crease to it yet.

and the english half of the manual, definitely says Mar 2012 .... of course cant read the other half ;-)

You can always demand the RP3300 probes - saying that you read the probe specs in the online material before you bought it. Those materials post-date the manual.

[alank2]

You can always demand the RP3300 probes - saying that you read the probe specs in the online material before you bought it. Those materials post-date the manual.

I agree - this is exactly what I would do.

[marmad]

and the english half of the manual, definitely says Mar 2012 .... of course cant read the other half ;-)

Look at all the materials - and the dates - at Rigol's main site. They ALL list the RP3300 (even the RP3300 User's Guide!) - so if I were you, I would demand an exchange.

[zibadun]

the problem is that the Rigol has user-selectable sample lengths - the Agilent does not. And it seems Rigol didn't figure out an elegant way to deal with this (e.g. forcing the sample length to a specific size).

So the issue here is that Rigol gave you enough rope to hang yourself?  While Agilent thought the user does not know any better and picked the "correct" sample rate for you

If you don't think it's possible, you should be getting an Agilent owner to prove me wrong. For example:
@Hydrawerk - can you please demonstrate? For example, send a 100kHz sine into the DSO - and adjust the horizontal scale until the sampling rate shows 100kSa/s - then grab a screen shot and post it here?

Ok . I'm curious what Agilent will show for 100 khz signal sampled at 100 ksps, at 10ms time base. but sounds like this picture will not be possible to obtain (since Agilent does not allow use override ).

[marmad]

So the issue here is that Rigol gave you enough rope to hang yourself?  While Agilent thought the user does not know any better and picked the "correct" sample rate for you

  I'm not sure you're completely getting it. In general, there is no difference between, for example, sampling at 2GSa/s and keeping every 10th sample - or reducing the clock to the ADC and sampling at 200MSa/s - they are identical when doing NORMAL sampling.

But for anti-aliasing purposes - which is a very specific need (I don't give a shit about anti-aliasing much of the time, such as when I'm using faster time base settings or close to the full sample rate) - then there is an advantage to oversampling and decimating, because you can vary the sample you keep - thus achieving random decimation and STOCHASTIC sampling.

I want the Rigol to operate exactly as it does now - unless I turn on Anti-Aliasing - at which point I want it to do 'real' anti-aliasing with stochastic sampling. 

Ok . I'm curious what Agilent will show for 100 khz signal sampled at 100 ksps, at 10ms time base. but sounds like this picture will not be possible to obtain (since Agilent does not allow use override ).

It has nothing to do with overriding - you just reduce the timebase until the Agilent gives you 100kSa/s.

Edit: The problem with the Agilent X2000/3000 series' is that you can't turn OFF anti-aliasing -so while it works perfectly for eliminating the aforementioned aliasing, it also makes waveforms look 'lumpy' (due to the random decimation) when you 'zoom' in on frequencies that aren't represented with many samples (as in attached image). BTW, the 100kHz signal sampled at 100kSa/s on the Agilent should look exactly like the image (unzoomed upper portion).

So ideally, you want to be able to turn on anti-aliasing when probing unknown signals at slower sample rates - then turn it off when you've settled on the correct time base for the signal. 

[jsykes]

Well they are listed as RP3300A.

I can't even find an RP3300A probe listed anywhere online - it's like a cheaper version of the RP3300 that they've produced specifically to include with the DSO. Kind of sucks for you, man. 

Perhaps they had it planned from the beginning - but didn't start producing them until later - so had to include the more expensive RP3300; I've no idea. But the online datasheet (from June 2012) lists the standard probe as RP3300 - and so does the User Manual (May 2012). So it might be you have a very early model.

Edit: What was your calibration date?

its a very new model.

dated 4th June 2013

the manual, has a satin feel to the outside cover... and I can only see my single set of finger prints on it...  I'd bet the manual had never been opened before,  the glued edge, hasn't had a crease to it yet.

and the english half of the manual, definitely says Mar 2012 .... of course cant read the other half ;-)

 
With respect to Rigol manufacture dates, I have read this somewhere. In checking the 5 different Rigol instruments that have been in my hands this seems to hold true. For example: serial # DS2A1436xxxxx. The 14 represents 2012 ( the 14th year Rigol has been in existence). This agrees with their website stating they were established in 1998.  http://us.rigol.com/html/about/history.shtml
The 36 represents the 36th week of 2012.
RigolNA also told me that a while ago, some units were sent back to be reworked. I received a 4022 that had been reworked but still was supplied with the pre-rework cal certificate. At turn-on, the menus were set for Chinese. I'm not sure if any of the 2000 series were reworked.
 
Hope this helps,

[darrylp]

Well they are listed as RP3300A.

I can't even find an RP3300A probe listed anywhere online - it's like a cheaper version of the RP3300 that they've produced specifically to include with the DSO. Kind of sucks for you, man. 

Perhaps they had it planned from the beginning - but didn't start producing them until later - so had to include the more expensive RP3300; I've no idea. But the online datasheet (from June 2012) lists the standard probe as RP3300 - and so does the User Manual (May 2012). So it might be you have a very early model.

Edit: What was your calibration date?

its a very new model.

dated 4th June 2013

the manual, has a satin feel to the outside cover... and I can only see my single set of finger prints on it...  I'd bet the manual had never been opened before,  the glued edge, hasn't had a crease to it yet.

and the english half of the manual, definitely says Mar 2012 .... of course cant read the other half ;-)

 
With respect to Rigol manufacture dates, I have read this somewhere. In checking the 5 different Rigol instruments that have been in my hands this seems to hold true. For example: serial # DS2A1436xxxxx. The 14 represents 2012 ( the 14th year Rigol has been in existence). This agrees with their website stating they were established in 1998.  http://us.rigol.com/html/about/history.shtml
The 36 represents the 36th week of 2012.
RigolNA also told me that a while ago, some units were sent back to be reworked. I received a 4022 that had been reworked but still was supplied with the pre-rework cal certificate. At turn-on, the menus were set for Chinese. I'm not sure if any of the 2000 series were reworked.
 
Hope this helps,

my serial starts DS2A152101...  so thats like may 2013, about right i'd say for a 4th june 2013 assembled and tested ready to go.

--
 Darryl

[Teneyes]

@Xyphro
The DS2000 does have display programs that cause a skip in resolution .
Below I show dots patterns on the 500uV/div scale , with just noise
I did capture 5 traces of dots, by setting Trigger level away from the noise ,so no triggers
set for SINGLE
and did 'FORCE' trigger to collect traces of dots on the display with infinite presistance

There are black bands on the vertical resolution between 2 rows of dot Data.

displays
5 traces of dots
multiple traces of dots
Auto on traces of Dots

Counting the rows , the display show 25 rows of dots/div 
with 8 divisions vertically, means 200 discrete values can be displayed
We know the rest of the 256 values are above and below what is displayed
 256 values is 8 bits, but how they get there is internal,

The steps you often see are the jumps across the black bands of the 400 Pixel LCD Display
resolution.,
The DSO uses all the 400 pixels to connect between dots (best algorithm )

[Xyphro]

Thanks, Teneyes, lucky to see that you also have blank lines.

That there are blank lines is not really an issue, I think we are all on the same level of understanding. The ADC generates 256 steps and the display is 480 pixels high (minus some pixels for top and bottom bar).
The procedure which lies between sample data and display data reshapes the data (interpolation / Decimation / ...) and pixels might not translate back 1:1 to ADC values, so depending on the level of interpolation/decimation, in between values can be shown on the display.

But it is funny to see that this lead to the discovery that the 500uV mode is in fact no "true" 500uV mode and only a digitally scaled 1mV version.
They could also offer a 250uV or 125uV setting with this method.

[marmad]

But it is funny to see that this lead to the discovery that the 500uV mode is in fact no "true" 500uV mode and only a digitally scaled 1mV version.
They could also offer a 250uV or 125uV setting with this method.

While it's completely possible that 500uV is scaled from 1mV, I don't understand why Rigol wouldn't list this in the specs - just as Agilent does for the X-Series: "** 1 mV/div and 2 mV/div is a magnification of 4 mV/div setting."

But since I don't have a clue as to how the Rigol handles gain at that setting, I'm afraid you just counting discrete values of different files is not, for me, definitive proof - just as you posting that the LSB of display memory data was always 0 wasn't definitive proof either.

I'll admit that the evidence seems to point in that direction - but since I'm certain Rigol won't lie about the specifications, I'll wait to hear what their response is before I assume that this is fact.

[Xyphro]

@marmad:
It can still be a problem of the CSV data export within the scope, but if this would not be the case, I have no another explanation for the same amount of values in histogram.
It's also a question if not mentioning something is really lying...

I did another test by setting the amplitude of my signal generator to a value that I just get clipping in the values of the CSV file @ 1mV/DIV setting. When capturing at 500uV/DIV I get the same amount of clipping.

What would be a proof for you?

[marmad]

@marmad:
It can still be a problem of the CSV data export within the scope, but if this would not be the case, I have no another explanation for the same amount of values in histogram.
It's also a question if not mentioning something is really lying :-)

Well, it probably can't be technically classified as lying  - but Rigol (as opposed to other Chinese manufacturers) has always seemed to be straightforward with their specifications; following accepted practices. If you look around at specs of DSOs which use scaling for lower vertical settings, I think it's a generally accepted practice to make a note of it in the specs.

I've also emailed my dealer about it. He's an expert Rigol technician - so either he'll know, or he'll get the answer from Rigol.

What would be a proof for you?

Well, I assume you're probably right - but I want to hear Rigol's response. I'm sure they won't lie about it when asked - since it could cause trouble for them if they did.

[Teneyes]

The procedure which lies between sample data and display data reshapes the data (interpolation / Decimation / ...) and pixels might not translate back 1:1 to ADC values, so depending on the level of interpolation/decimation, in between values can be shown on the display.

Yes , The DSO can interpolate extra data point between 7 Bit data samples to create data  values for the Display.
But, I wonder why I see 8 bits in the Waveform data file witch is before the display program handler

[Teneyes]

SO, I generated a ramp to test the 500uv & 1mV resolutions and see what the output to the CSV File is like. 
Below are the pix
  1 DSO display

  2 A section of Excel csv file from 1mV/div, with added columns for:
                           Fix Point Format (volts)
                           uVolts and
                           change between last data point in the CSV file 

  3 A section of Excel csv file from 500uV/div, with added columns for:
                           Fix Point Format  (volts)
                           uVolts and
                           change between last data point in the CSV file 

Note" the section of data is near the 0.0 Volts value, ie @ sample 1400 in 2.8K


I  see for 1mV/div the resolution is 40uV in the data samples (delta = n*40)
And for  500uV/div the resolution is 20uV in the data samples (delta = n*20)


Also
@Xyphro
  I do not fully understand your Histogram  and I would like to see the histogram with the Axis switched  (0-200) on 'X' and show bars for number of samples occurring for each voltages level.  maybe try "hist(sig500uv, 256)";

[Xyphro]

I got a confirmation from Rigol, that the 500uV mode is a scaled mode.

[EV]

I got a confirmation from Rigol, that the 500uV mode is a scaled mode.

So we have been cheated! 

[marmad]

I got a confirmation from Rigol, that the 500uV mode is a scaled mode.

For my own work I don't particularly care, but they should REALLY note that in their documentation.

Would you mind posting their exact response?

[Teneyes]

I am glad , I have 500uV/div, with 200uV Noise @BW=20MHz
and 20 uV resolution, how ever it is derived.

[EV]

What do you mean 'WE'.. .., 

I am glad , I have 500uV, with 200uV Noise @BW=20MHz
and 20 uV resolution, how ever it is derived.

I am glad and pleased too! 

[egonotto]

Hello,

I did not understand how the resolution in 500uV/Div normal mode can be 20uV?
If it was scaled 1mV/Div I would expect a 40uV resolution.

Best Regards
egonotto

[Teneyes]

I did not understand how the resolution in 500uV/Div normal mode can be 20uV?
If it was scaled 1mV/Div I would expect a 40uV resolution.

take a series of 40 uV points and interpolate 
   I guess  take     200, 200, 240, 240, 280, 280
           and make  200, 220, 240, 260, 280, 300
 

[egonotto]

Hello Teneyes,

this would be a kind of filtering and the BW should go down.
And it would be no scaling.

Best Regards
egonotto

[Wim13]

this would be a kind of filtering and the BW should go down.

Maybe EV or Wim can do a Sweep with DS2000 set @ 500uV, I only did from 4-40 MHz

I dont get the clou, but the bandwidth at 500 uV is the same as at other levels, what
do i have to do...