New Rigol DS1054Z oscilloscope

[metrologist]

Note that Sinc is good for sine or some other "smooth" waveforms. It is NOT best in all cases for fast square-rectangle types of signals.

Isn't it also more significant that you cannot turn off sin(x)/x interpolation unless you have 3+ channels on? You mentioned the dots and they seem OK, but also you mentioned who knows?

[bitseeker]

@rf-loop, I like the way that scope (not Rigol) shows the actual samples with brighter dots when drawing in dot mode. That's a nice feature. I agree that it would be nice if it did that in sin(x)/x mode (via contrasting color).

[frozenfrogz]

Can someone please explain the whole sin(x)/x issue to an oscilloscope-town n00b?
I see, that the displayed waveform in vector mode is not a connect-the-dots with straight lines, but an interpolation. Also, the sin(x)/x is greyed out as "on" in one and two channel mode and does not really do much with three, or four channels activated. (I can not see the difference)

I am probing the antenna of a PMR/Freenet dual channel handheld radio. The displayed trigger frequency does not make sense to me, but the measured values for period and frequency for the Freenet channels are plausible (6.7ns / 149MHz carrier) and even on 446MHz for the PMR carrier, at least the order of magnitude is kind of ok-ish (amplitude of course is nowhere near sane values, but then again this is way out of specs).

With all four channels activated, showing dots and persistence set to infinite, I see "slices" in the plot. As if the sampling is done in serial bursts per channel.
The measurements are way off, but to my understanding: Usable bandwidth on single channel is good up to 80MHz with a 10x probe, two channels more like 50MHz because of shared sampling and three / four channels up to 25MHz.
Most probably I am mixing up a lot of stuff here - please be patient, I am still trying to figure everything out

The "vector" plot looks a like a spline interpolation to me, so it is in fact a connect-the-dots, but with a curve instead of straight lines.

This is on software 00.04.04.01.01, board v. 0.1.4, boot v. 0.0.1.4, firmware 0.2.3.11, 100MHz option,... (just for reference)

Please enlighten me
Frederik

[metrologist]

The main point is that the display data is being manipulated when it shouldn't. rf-loop seemed to indicate you cannot get the raw adc value either, so the data always has some kind of manipulation and you do not know what that is because it is some kind of smoothing function that differs from sin(x)/x. And we know all measurements are performed from display data. And with the dropping amplitude at different sweep rates, it just draws the instrument into serious question.

[David Hess]

@rf-loop, I like the way that scope (not Rigol) shows the actual samples with brighter dots when drawing in dot mode. That's a nice feature. I agree that it would be nice if it did that in sin(x)/x mode (via contrasting color).

The ancient Tektronix 2440 series of DSOs support intensifying the actual samples when sin(x)/x interpolation is used but I have yet to find an application where this is useful other than verifying that sin(x)/x interpolation works correctly.  I do not remember any of their later DSOs supporting it.

[ebastler]

The main point is that the display data is being manipulated when it shouldn't. rf-loop seemed to indicate you cannot get the raw adc value either, so the data always has some kind of manipulation and you do not know what that is because it is some kind of smoothing function that differs from sin(x)/x. And we know all measurements are performed from display data. And with the dropping amplitude at different sweep rates, it just draws the instrument into serious question.

I think you are mixing up two very different things here:

Your first point correctly describes the sin(x)/x interpolation issue discussed in the prior posts. The scope should always retain and show the raw sample data in "dot" mode, and -- dependent on whether sin(x)/x interpolation is activated or not -- should use either smooth interpolation or linear "connect the dots" interpolation in "line" mode. Messing with the actual sample values is not a correct implementation, and not documenting what the scope does there makes it worse.

Your second point, in my opinion, has nothing to do with this. Yes, Rigol scopes derive measurements from display data -- as do Agilent/Keysight scopes, I believe. That of course means that measurement values will depend on the sweep rate. Works fine for me: I primarily use a scope to visualize a signal, and I see it as a nice extra that the scope can quantify what I see on the screen. It would not occur to me to ask the scope to measure something that is not properly visible on the screen.

That may just be me, and due to the fact that I come from CRT scopes where I had to do the "measuring" myself by eyeballing on the screen grid. But I don't think this can be considered an "incorrect" implementation of measurements; it is just a design decision.

[frozenfrogz]

Ok. So there is something strange about the implementation of the interpolation and the issue with not getting the raw ADC values. Also, (because of this, or something totally different) the amplitude of the displayed waveform changes on different time-bases.
But when is this really an issue? All the examples seem to deal with signals that are somewhat on the edge, or outside of the capabilities of this device. Given the specs of the DS1000Z series, I would assume 1GSa/s is OK for the projected 100MHz (10 samples per cycle - not awesome, but acceptable). But that does not mean, you should get sane representations of a signal at the upper limits of that. A 100MHz scope just is not the right tool to look at 100MHz signals IMO.
Please correct me if I am mistaken, but to get sane results I would choose a scope that has at least twice (or even 3-5x) the bandwidth of the highest frequency that needs to be looked at. Following this, the Rigol DS1054Z (hacked to 100MHz) would be just right to for example look at Atmel AtMegas with up to 32MHz clock frequency.
Expected minimum rise times are around 5ns, meaning 70MHz minimum bandwidth (from: bandwidth = 0.35/rise time).

So here is my question: Does the unexpected behavior of the implemented sin(x)/x matter beneath 50MHz, or is this just about the limits of this device and not being informed about way of readings thereof?
Or am I still missing the point of the discussion?

[metrologist]

I did not comment about the fact of using the display for measurement data, it works fine for me too. But if you are trying to quantify something like pulse overshoot and ringing, and the scope is going to build false peaks you can't get accurate results. How about just a pure sine wave dropping almost 2 divisions pk-pk just because you disabled sin(x)/x or changed your sweep from 200ns? 20% That is improperly displayed on the screen - it is wrong, not a design decision.

[alsetalokin4017]

One thing that jumps out with frozenfrogz screenshots is how far off the hardware frequency counter is. The traces show a period of pretty close to 6.7 ns which gives a frequency of 149 MHz by inspection, all good, and the Measurements agree with the traces. But the HW counter is wayyy off at just over 100MHz. Usually in my experience (at lower frequencies of course)  it is the other way around, the HW counter is pretty accurate but the Measurements can be off.

What is the story here?

[MrW0lf]

[frozenfrogz]

But the HW counter is wayyy off at just over 100MHz.

My guess is, that the trigger expects to see a frequency between 15Hz and 100MHz (plus a little extra, maybe up to 101MHz). In my case the signal frequency is seriously out of bounds and therefor displaying nonsense. Actually the trigger frequency was jumping all over the place, even more with the 446MHz signal.
For me, that is totally acceptable, as the range of projected working frequencies up to 100MHz are detected just fine.

I suspect the trigger is trying to get a phase lock over a specific length of time. With the 446MHz signal it looked like the trigger was reading partials of the carrier frequency (not displayed here, would just show up by accident). It displayed values like 74.xxxMHz - what would actually the fifth subharmonic (446MHz / 6).
But that is just an assumption.

[Fungus]

But when is this really an issue? All the examples seem to deal with signals that are somewhat on the edge, or outside of the capabilities of this device.

Correct. It's not an issue in real use.

Given the specs of the DS1000Z series, I would assume 1GSa/s is OK for the projected 100MHz (10 samples per cycle - not awesome, but acceptable).

Yes. With one channel on at 100MHz this is a total non-issue.

With all 4 channels on you only get 2.5 samples per cycle and this effect will start to show.

The real issue here is people's expectations. A 100MHz oscilloscope with 2.5 samples per cycle isn't the right tool for looking at 100MHz signals, period. Doesn't matter what brand of oscilloscope you choose.

(unless you're only interested in pure sine waves...)

Bottom line: This is worth knowing about but it isn't a reason not to buy a DS1054Z.

it just draws the instrument into serious question.

Anybody saying things like this is failing to see the beautiful green forest because of all the pesky trees in the way.

They're free to spend three times more money on another brand if they think it will help.

[David Hess]

Ok. So there is something strange about the implementation of the interpolation and the issue with not getting the raw ADC values. Also, (because of this, or something totally different) the amplitude of the displayed waveform changes on different time-bases.

Amplitude is only affected by the frequency response up to and including the ADC unless frequency compensation is done digitally which is common on high bandwidth DSOs.  Timebase settings and sample rate should have no effect.  (1)

But when is this really an issue? All the examples seem to deal with signals that are somewhat on the edge, or outside of the capabilities of this device. Given the specs of the DS1000Z series, I would assume 1GSa/s is OK for the projected 100MHz (10 samples per cycle - not awesome, but acceptable). But that does not mean, you should get sane representations of a signal at the upper limits of that. A 100MHz scope just is not the right tool to look at 100MHz signals IMO.

How about making RMS measurements of a noise signal which is completely within the bandwidth of the oscilloscope?  This is one of the easier and more useful measurements that a DSO should be able to make (2) and the DS1000Z series fails completely at it.

(1) High bandwidth DSOs do this but it might explain the odd sin(x)/x results not lining up with the sample points.  Is this only failing on oscilloscopes which are hacked to unlock their bandwidth because the sin(x)/x filter is implementing passband correction?  I doubt it but maybe someone can check.

(2) It is just the standard deviation of a bunch of points.  That it does not work on the DS1000Z series is easily explained by measurements being made on the display record.

[bitseeker]

But when is this really an issue? All the examples seem to deal with signals that are somewhat on the edge, or outside of the capabilities of this device.

Correct. It's not an issue in real use.

Well, it depends on what your real use is. As always, it boils down to knowing your tools and their limitations, quirks, bugs, etc. Hence, I enjoy seeing these investigations even if they are edge cases in order to better understand how and when this scope works or doesn't.

[CatalinaWOW]

But when is this really an issue? All the examples seem to deal with signals that are somewhat on the edge, or outside of the capabilities of this device.

Correct. It's not an issue in real use.

Well, it depends on what your real use is. As always, it boils down to knowing your tools and their limitations, quirks, bugs, etc. Hence, I enjoy seeing these investigations even if they are edge cases in order to better understand how and when this scope works or doesn't.

Agreed.  Knowing the capabilities and limitations of any tool is important.  That is why it is somewhat OK to disparage a tool which does not achieve its published performance.  But it is just silly to say a tool is useless because it has limitations.

[Assafl]

But when is this really an issue? All the examples seem to deal with signals that are somewhat on the edge, or outside of the capabilities of this device.

Correct. It's not an issue in real use.

Well, it depends on what your real use is. As always, it boils down to knowing your tools and their limitations, quirks, bugs, etc. Hence, I enjoy seeing these investigations even if they are edge cases in order to better understand how and when this scope works or doesn't.

Agreed.  Knowing the capabilities and limitations of any tool is important.  That is why it is somewhat OK to disparage a tool which does not achieve its published performance.  But it is just silly to say a tool is useless because it has limitations.

+1 

This thread reads like an Acceptance Test journal of a group that never decided what the criteria for the acceptance test was. For one - it is RMS (which is much better but still...); for one, Fidelity of the data acquisition (needs those unadulterated bits), for another it is spelling errors (+ Pluses okay - but - Pluses??? ), for yet another - the goddamn awful FFT (thank you PC FFT) etc. These (and many others) are all legitimate reasons the Rigol should not pass your Acceptance Test - entirely. REJECTED!

As a household scope tool (like the cordless drill, the Dyson, the wrench kit, screwdrivers... etc.) it is more than ample. I was able to fix quite a few household items (like a few SMPSs, an amplifier, a weird RS232 issue between the Agilent and the TTI power supply, finally do an auto power off for the coffee grinder based on power input, lots of Arduino and other household embedded work, etc.). All with AMPLE capability.

Sure - at the other side of the hobby (I made the mistake of clicking on the metrology forum once - their resistors cost more than a scope...)  - trying to track down noise in really quiet voltage reference has been far less easy with the Rigol - but so what? I've done this in the past: Increase SNR to where the scope works (which of course, is vastly easier said than done).

Claiming that any of the above "rejections" IN ANY WAY - reduces the usability of Rigol as a "household scope tool" (for uses similar to the ones stated above) - is just wrong, misleading and foolish. Similarly claiming that the Rigol can replace the Picoscope or your 10k NI DAQ setup - or that it is as friendly as an R&S or a Lecroy - is just as foolish.

Rant: Why do some people need to categorically dismiss a product because of limitations that affect their use cases? Why not say "I found it too limiting for my needs" and explain why so we can all be careful and kowtow the thought care and rigorousness of presenting the evidence - rather than the offensive "Rigol is useless shit" (which also bears bad on the poster - being of a rather childish disposition)?



 

[BravoV]

Rant: Why do some people need to categorically dismiss a product because of limitations that affect their use cases? Why not say "I found it too limiting for my needs" and explain why so we can all be careful and kowtow the thought care and rigorousness of presenting the evidence - rather than the offensive "Rigol is useless shit" (which also bears bad on the poster - being of a rather childish disposition)?

Why ? Few that I could think of ...

- Some competing brand's distributors find this kind of big popular thread is quite annoying, probably made they can not move their competing products fast enough out from their warehouse, while new stock are piling up, but their new potential customers are keep shouting .. "I want Rigol !!" and keeps pointing out this thread. 

- Disgruntling personalities that use this thread for venting out, its nothing about this scope really.

- DS-1054Z envy, as they made wrong decision buying competing brand, while keep drooling why this thread keeps growing, while their scope is left out cold never discussed and popularly explored like in the forum discussion.

[Karel]

Having read all the comments in this thread, my conclusion is that, for most hobbyists,
the Rigol DS1054Z is still the recommended scope in it's price class.

[Howardlong]

Having read all the comments in this thread, my conclusion is that, for most hobbyists,
the Rigol DS1054Z is still the recommended scope in it's price class.

Agreed, it's like trying to find the perfect partner or perfect house. Neither exist, you have to compromise, and where you compromise depends on your priorities over a number of factors, different people will have different priorities.

One thing is absolute, this scope is certainly not a toy as some have tried to convey. It is a serious instrument, and like all tools you have to know how to use it and be able to interpret the results, and, that will be through experience and some learning. Not all tools work the same way. For some things, yes, it'll be the wrong tool, nobody's saying otherwise, like you wouldn't use a micrometer to measure a mile. But as a general purpose scope it is difficult to see how overall it can be bettered at the price point.

[xrunner]

Having read all the comments in this thread, my conclusion is that, for most hobbyists,
the Rigol DS1054Z is still the recommended scope in it's price class.

There you go folks. The summary we've been waiting for.

Now, we can close the thread after 145 pages. Thank everyone for attending! 

[ebastler]

Now, we can close the thread after 145 pages. Thank everyone for attending! 

No, sorry, can't do that.
We need to keep the thread open for the periodic "does the hack still work with the latest firmware?" questions. 

[Fungus]

Having read all the comments in this thread, my conclusion is that, for most hobbyists,
the Rigol DS1054Z is still the recommended scope in it's price class.

On the 2nd of April 2017, yes.

Even the up-and-coming new low-end 'scopes from other manufacturers don't quite unseat it. They either cost a lot more or have less channels/features.

To unseat the Rigol they'll need to be hacked to release their upgrade options. So far that's looking unlikely.

[Fungus]

One thing is absolute, this scope is certainly not a toy as some have tried to convey. It is a serious instrument,

Yep. The things that people really could be complaining about simply aren't there. Build quality is good, there's no lockups that need power cycles, nothing like that. It works, it does the job.

[coppice]

Having read all the comments in this thread, my conclusion is that, for most hobbyists,
the Rigol DS1054Z is still the recommended scope in it's price class.

For most professionals the DS1054Z is just fine, too. The majority of professional engineers are currently working on things where any high speed signals are confined within a single package, where a scope can't get at them.

[jonovid]

saving up for this one