Rigol DS 1054Z behavior: normal or defective?

[tom66]

The probable reason Rigol don't do what you suggest is because it would require using saturating arithmetic, which adds additional logic to the FPGA.  I have considered how to solve this problem for some time and it seems that you realistically need 12-14 bit depth accumulators which is quite memory and logic expensive.  Rigol use an external SRAM chip of ~4Mbit attached to the FPGA to do the render task.  Many FPGAs have only half that much BRAM on the chip itself.   Using SRAM means they can do random read and writes with little penalty but the part they use is at least $20 (from memory) on Digi-Key so no doubt a significant BOM expense.

[ballen]

I'm not sure that I can live easily with this - I have spent most of my life with analog scopes and want to be able to trust what I see and not wonder if it's an artefact of the scope.  So if there is a single voltage I'd like to see a single thin horizontal line.  A thicker line means that there is hash or noise on it.  If a waveform is spending most of its time in a particular region, that region should be brighter not darker.

Do other entry-level digital scopes behave this way?  The second most popular model comparable to the Rigol is the Sigilent SDS1104X-E, which costs about an extra 30 Euros.  Does it not have these peculiarities or other such issues?

[tom66]

Every digital scope is lying to you.

The Rigol also aliases and it appears some firmware versions have a timebase bug causing times to be off between 4-8%.

Other scopes have bugs too.  If you want the scope to behave identically to an analog scope you will, unfortunately, be disappointed.  However, the chance of these bugs causing anything other than curious EEVBlog posts is close to nil.

Despite its limitations the Rigol DS1074Z I have has been a trusty workhorse for the last 5 years.  I used it during my masters' degree, in the office and on personal projects. Most recently, I used it to analyse the performance of a >1Gbit/s serial bus with some little tricks.  Despite it being a sub-$600 scope, I've never found it lacking in performance - it punches well above its weight for its class.

I know there's a trend towards Siglent on this forum and that's fair, they make good kit too (I really liked the SDS5000X we had in for evaluation) but haven't personally tried using any of their lower end scopes.  The Rigol definitely has a few more limitations that Siglent have fixed on their instrument.

[Fungus]

I'm not sure that I can live easily with this - I have spent most of my life with analog scopes and want to be able to trust what I see and not wonder if it's an artefact of the scope.  So if there is a single voltage I'd like to see a single thin horizontal line.  A thicker line means that there is hash or noise on it.  If a waveform is spending most of its time in a particular region, that region should be brighter not darker.

First of all: Don't use dots mode. Ever. There's people here who imagine they can squint their eyes and see something but it's not true. It doesn't tell you anything useful. The true signal comes from applying a reconstruction filter to those dots.

Secondly: Learn to use the settings under acquisition. If it's a simple repetitive signal then you can set "average" mode to get a pixel perfect trace on screen. The other modes can reduce noise, etc.

Thirdly: I don't recall ever seeing a double line on my DS1054Z without weird settings.

[Fungus]

Every digital scope is lying to you.

Usually the lies are small.

Other scopes have bugs too.

Yep. The Siglent bug thread is busy, the Keysight bug thread is busy, the Rigol bug thread is busy.

Despite its limitations the Rigol DS1074Z I have has been a trusty workhorse for the last 5 years.

Yep. But I get accused of things if I say that.

[Miti]

I'm not sure that I can live easily with this - I have spent most of my life with analog scopes and want to be able to trust what I see and not wonder if it's an artefact of the scope.  So if there is a single voltage I'd like to see a single thin horizontal line.  A thicker line means that there is hash or noise on it.  If a waveform is spending most of its time in a particular region, that region should be brighter not darker.

Do other entry-level digital scopes behave this way?  The second most popular model comparable to the Rigol is the Sigilent SDS1104X-E, which costs about an extra 30 Euros.  Does it not have these peculiarities or other such issues?

You're in the first stages of change, see the attached graph. Change the angle of view, don't let few quirks blind you. All DSOs have their issues, even the expensive ones but they are the present and the future. Analog scopes are the past. This thing can do so many things for relatively little money that it's worth forgiving few sins. My 2 cents...

[TurboTom]

First of all: Don't use dots mode. Ever. There's people here who imagine they can squint their eyes and see something but it's not true. It doesn't tell you anything useful. The true signal comes from applying a reconstruction filter to those dots.
...

Sorry Fungus, I tend to disagree with that statement of yours. At times, especially when trying to visualize statistical effects (cumulative phase jitter and the like), dot mode in combination with infinite persitance can be very useful to observe a temproal distribution over prolonged periods of time, while vector mode will just swamp your screen with an undefined mess.

But at those time base settings that are required for this kind of measurement, the "screen buffer overflow effect" (double-line) never occurs.

I agree though, that for just displaying "wiggly lines", dot mode is pretty much redundant.

[bdunham7]

Peter, thank you, this would indicate that my scope is "normal" rather than "defective".  The stripes seem to be discrete ADC output levels, normally hidden in vector mode.

What do you see if you increase the sweep to 5nS/div?  Don't use zoom, use the horizontal position to move the trigger far enough off screen so that you are looking at a flat section of the signal where it is split.

EDIT:  And when you have that set up, then try it in SINGLE mode--just one capture.

[james_s]

I'm not sure that I can live easily with this - I have spent most of my life with analog scopes and want to be able to trust what I see and not wonder if it's an artefact of the scope.  So if there is a single voltage I'd like to see a single thin horizontal line.  A thicker line means that there is hash or noise on it.  If a waveform is spending most of its time in a particular region, that region should be brighter not darker.

Do other entry-level digital scopes behave this way?  The second most popular model comparable to the Rigol is the Sigilent SDS1104X-E, which costs about an extra 30 Euros.  Does it not have these peculiarities or other such issues?

ALL digital scopes have quirks, it's a mistake to view a DSO as a digital version of an analog instrument. It's a different beast altogether and requires different techniques to use. They offer so many additional capabilities that once you get used to it you won't want to go back. It took me a while to adapt but these days the only time I get out my Tek 465 is when I want XY mode.

[tom66]

Worth noting my DS1074Z achieves 50kwfm/s on dot mode,  double what it achieves on vector mode.

Drawing vectors is not free, neither is interpolation, so this is not surprising.  So there is benefit in other ways too, when you don't care about perfect reconstruction but want to see infrequent jitter or more detail from waveform data.

[ballen]

What do you see if you increase the sweep to 5nS/div?  Don't use zoom, use the horizontal position to move the trigger far enough off screen so that you are looking at a flat section of the signal where it is split.
EDIT:  And when you have that set up, then try it in SINGLE mode--just one capture.

Thank you Peter,  this was very instructive!

Ground the input, single mode, push the force button to capture a trace of DC + input noise.

If I do this on the highest-gain setting, the input amplifier noise gets spread among ~8 adjacent ADC levels. If you horizontal-zoom in on these, you can clearly resolve the 8 levels, and can see that the levels near the middle are the most commonly occurring ones - a Gaussian distribution centered on the mean. But then as you zoom out in time (as you say, "horizontal decimation") the most common levels in the middle disappear and only the least frequent "top and bottom" edge levels are visible!

To me this seems like a horrible show-stopping software or hardware bug.  I presume the latter since otherwise it would have been fixed long ago.  Can you really learn to live with this?

EDIT: here is a detailed report about this problem: https://www.eevblog.com/forum/testgear/rigol-ds1000z-series-buglist-continued-(from-fw-00-04-04-03-02)/msg1442914/#msg1442914

Cheers,
   Bruce

[james_s]

It seems you may be expecting too much from a budget instrument aimed at hobbyists. The Rigol is not a $10k Keysight or Tektronix, but thousands of people seem to be using them effectively. Using the highest gain input setting seems like an edge case, my Tek scopes are not super useful at the highest gain setting.

[ballen]

I am using the high-gain setting only because that makes it easy to generate a "signal" by grounding the input. The display problem occurs with real signals anytime there is enough noise present to jitter the ADC.

 

[Fungus]

First of all: Don't use dots mode. Ever. There's people here who imagine they can squint their eyes and see something but it's not true. It doesn't tell you anything useful. The true signal comes from applying a reconstruction filter to those dots.
...

Sorry Fungus, I tend to disagree with that statement of yours. At times, especially when trying to visualize statistical effects (cumulative phase jitter and the like), dot mode in combination with infinite persitance

OK, dot mode combined with infinite persistence might reveal something when you're at the very edge all channels turned on and Gibbs phenomenon is starting to appear but you'll still get better results in vector mode and a single channel.

Dots mode by itself though? Not useful.

[Fungus]

To me this seems like a horrible show-stopping software or hardware bug.  I presume the latter since otherwise it would have been fixed long ago.  Can you really learn to live with this?

I say it again: I never saw a double line on mine at ordinary zoom levels. Other people here are also saying they don't see this under these conditions. If you have a chance to exchange yours then do so.

[Miti]

Worth noting my DS1074Z achieves 50kwfm/s on dot mode,  double what it achieves on vector mode.

And how do you measure that, by looking at the trigger out?

[bdunham7]

To me this seems like a horrible show-stopping software or hardware bug.  I presume the latter since otherwise it would have been fixed long ago.  Can you really learn to live with this?

No, if the acquisition mode is not set to 'PEAK', then that looks broken.  I wouldn't live with it.

[bdunham7]

Dots mode by itself though? Not useful.

As implemented by Rigol, at least on the one I had, I agree that dots mode is misleading and useless.  I have no issues with it on my Tek models, and so far, my Siglent.  For the vast majority of cases, I'd be perfectly happy with either display method.

[tom66]

And how do you measure that, by looking at the trigger out?

Yes, exactly.  The trigger out frequency roughly doubles in dot mode.

[mawyatt]

I'm not sure that I can live easily with this - I have spent most of my life with analog scopes and want to be able to trust what I see and not wonder if it's an artefact of the scope.  So if there is a single voltage I'd like to see a single thin horizontal line.  A thicker line means that there is hash or noise on it.  If a waveform is spending most of its time in a particular region, that region should be brighter not darker.

Do other entry-level digital scopes behave this way?  The second most popular model comparable to the Rigol is the Sigilent SDS1104X-E, which costs about an extra 30 Euros.  Does it not have these peculiarities or other such issues?

I've been using scopes probably longer that most here have been alive, over 50 years Like you I've grown up with analog scopes, just about every Tektronix model you can think of, and like you had come to "trust" what the Tek scope was trying to tell me, they were and are still marvelous instruments

The most important "feature" I needed from a DSO over the older analog scopes was the ability to capture and store a waveform, and the ability to perform simple mathematics on the waveform. Since a DSO is nothing more that a data capturing device with a built-in display, these two features are natural to the DSO.

Anyone that tells you that not having a accurate and good resolution ADC, a good low noise front end with low end range and an accurate, reliable and believable displayed DSO waveform probably has a technical IQ equal to their age!! The old Tek analog scopes gave you these features (of course no ADC), so why settle for a modern DSO that can't even perform the most important feature of a scope, to show waveforms that are accurate, low noise and believable.

As you probably know there are numerous fanboys and girls here, pushing there own agenda and bashing everyone's else's option if it doesn't align with theirs. Claiming this feature matters (if their DSO has it) and that feature doesn't (if their DSO is lacking). I just wanted a DSO to work like my old Tek analog scopes, give me nice believable displays without too much added noise, allow some simple math functions and be able to reliably capture and store waveforms. I ended up with a DSO more akin to a sophisticated data capturing device than a scope, and delivering so much more that I expected, but still can perform the basic functions of an analog scope with a believable display.

Like you I was exposed to all the DSO BS flying around, and had to sift thru everything to find the truth. I'm sure you'll find your way thru the DSO BS maze and select a DSO that meets your requirements and budget, whichever brand it is!!

My only recommendation is to look towards the upper end in features and cost, even push your budget because a scope is the 2nd most important instrument you will own, a quality DVM is 1st

Best and good luck,

[_Wim_]

I ended up with a DSO more akin to a sophisticated data capturing device than a scope, and delivering so much more that I expected, but still can perform the basic functions of an analog scope with a believable display.

Can you share which DSO you are using? I am currently using a picoScope 5000 series for its high resolution, but the small signal performance could still use some improvement...

[mawyatt]

I ended up with a DSO more akin to a sophisticated data capturing device than a scope, and delivering so much more that I expected, but still can perform the basic functions of an analog scope with a believable display.

Can you share which DSO you are using? I am currently using a picoScope 5000 series for its high resolution, but the small signal performance could still use some improvement...

Ended up selecting the SDS2102X Plus which has exceeded my expectations in just about every way. It's been "enabled" with the kind help of folks here to 350MHz and just yesterday with help from tv84 to 500MHz which just quickly measured to 615MHz

I can't speak about other brands or Siglent scopes, but this SD2102X Plus is a stunningly good little DSO IMO. Wanted the 2104X+, but that was BO so settled for the 2102X+. The ADC they are using is quite good, much better than one would expect from an 8 bit device, I posted some statistics measurements and well you judge for yourself!

Here's some measurements from another thread.

First off DC measurements of the SDS2102X Plus input using statistics of C1

Input KS34465A of 5.00021VDC, DSO reads Mean C1 5.000398VDC (mean of mean)
Input of 10.00001VDC, DSO reads 10.0031VDC
Input of 7.04184VDC, DSO reads 7.04192VDC
Input of 0.99985VDC, DSO reads 999.39mvDC
Input of 100.0073mvDC, DSO reads 99.825mvDC

Now AC measurement with 50% duty cycle square wave of 5.00021V peak value, so Mean & StdDev (RMS) should be 5.00021/2 or 2.500011V

Input at 250Hz KS34465A 2.49993VAC, DSO reads 2.4916VDC (mean) and 2.5085VAC (stddev)
500Hz 2.49993VAC, DSO reads 2.4912VDC & 2.5094VAC
1KHz 2.49886VAC, DSO reads 2.4910VDC & 2.5095VAC
4KHz 2.49452VAC, DSO reads 2.4913VDC & 2.5092VAC
8KHz 2.48868VAC, DSO reads 2.4921VDC & 2.5085VAC

Note: The peak to peak readings of ~6Vpp include waveform ringing.

This is better performance than I would expect from this DSO

Don't know if I got lucky or this is typical performance from the SDS2102X Plus, maybe others could make some measurements


Here's snip on the ENOB & ERB behavior from another thread, which you can compare with other DSO scopes ENOB & ERB measurements by others.

Just did a quick test of the SDS2102X Plus in 8 bit, 10 bit, and ERES (3) modes using this as a means to evaluate Effective Resolution in bits. I'm not saying this is right or wrong way, just the way I chose to measure and calculate. Used 20MHz BW with 50 ohm in-scope termination, 1us/div sweep. Noise was averaged Standard Deviation to remove DC offset effects as mentioned.

Effective Resolution Bits (ERB) = {10*log[(8*(SF/N)^2 +1)] -1.76}/6.02, where N is Standard Deviation noise, SF is Scale Factor in V/div

Scale    500uv   1mv    2mv    100mv    200mv     500mv     1v
8 bit N  23.4uv  24uv  38.6uv  1.47mv  2.95mv    7.55mv    14.56mv
ERB     5.63      6.59    6.90     7.30      7.29         7.26        7.31

10 bit N 16.9uv  17uv   20uv   316uv    432uv      1.83mv    2.32mv
ERB     6.10      7.09     7.85    9.50      10.06       9.72        9.96

ERES N 12uv     12uv   14uv     132uv    190uv      725uv     860uv
ERB     6.59      7.59    8.37     11.60     11.89      11.80      11.39

FFT NF -122      -120   -120     -118      -108        -100        -97
in dBm

Added FFT noise floor in 8 bit mode using 2.5MSPS, 2097512 pts, delta f 1.19Hz and 16 average.

I computed these on my HP32 calculator, so hopefully no mistakes


Here's the thread for some other DSO ENOB measurements by Howard Long.

https://www.eevblog.com/forum/testgear/benefits-of-going-with-all-siglent-setup/msg3281580/#msg3281580



Best,

[Miti]

Ended up selecting the SDS2102X Plus which has exceeded my expectations in just about every way. It's been "enabled" with the kind help of folks here to 350MHz and just yesterday with help from tv84 to 500MHz which just quickly measured to 615MHz

For me, to spend that amount of money, it must make money afterwards. Just for hobby, naaah...
And then I need 4 channels more than 600 MHz.

[james_s]

For me, to spend that amount of money, it must make money afterwards. Just for hobby, naaah...
And then I need 4 channels more than 600 MHz.

~800 bucks? That's not impulse purchase territory for me but it's certainly well within the hobby budget. I've spent more than that on toys on multiple occasions. I do agree that 4 channels is nice to have, although when I think about it I only use more than 2 channels maybe 10% of the time.

[mawyatt]

For me, to spend that amount of money, it must make money afterwards. Just for hobby, naaah...
And then I need 4 channels more than 600 MHz.

~800 bucks? That's not impulse purchase territory for me but it's certainly well within the hobby budget. I've spent more than that on toys on multiple occasions. I do agree that 4 channels is nice to have, although when I think about it I only use more than 2 channels maybe 10% of the time.

I do appreciate making $ with these as an investment, since I still do consulting hopefully this will be true.

As far "toys" go, yes I've spent a few more $, look at my avatar
The red 911 is worth more today than when I purchased it new in 1996 (last of the air cooled), my daily drive until I acquired the black 911 in 2007

I find 2 channels sufficient most of the time also, but have had an occasion or two where a couple extra channels were required.

Best,