Choosing between entry-level 12-bit DSOs

[tggzzz]

As for the eye movements, it's very funny ... I don't have any problems.

There's nothing funny. Forcing the user to shift focus, unless it is a natural part of the flow in question, is bad, it's that simple. Good UI designers know this and place related elements in such a way that they are all can be seen and read without a single movement of the eyeball. An example would be the central section of a car's dashboard, a wristwatch, or winamp's main window.

An extreme example of that is an aircraft cockpit, where the "normal and good" position of all the meter needles is pointing straight up. That makes it possible to spot which of the 100 measured values is a problem with just a glance.

Glass cockpits lose that advantage, and the workarounds lead to the all-too-frequent last words on cockpit voice recorders: "what is it doing now?".

[shapirus]

If you don't fit the entire waveform vertically on the display, then that causes other problems. When the input front-ends are overloaded the amplifiers become saturated. Exiting saturation is never timely nor clean. Decent old analogue scopes can be better behaved in that it can take more of an overload to cause saturation; nonetheless it can occur.

Admittedly I did not deep dive into the amplifier settling time measurement problem, but what would other situations be where this might be an issue? What you describe can be said in other words as "extreme zoom", correct?

Yes, naturally there is a limit of the maximum vertical offset of a zoomed-in waveform. I don't have other examples, but I tried this with Rigol DHO800. It's not great, but not terrible, either. Probably good enough for a non-specialized scope. I'm attaching an example of the maximum zoom (scale + offset) possible for the upper portion of a 2.4 Vpp (unipolar) square wave's rising edge.

Apart from a limited offset that can be achieved at a given vertical scale, I see no other issues. The trace can be moved, the time base can be changed etc., and the waveform update happens as usual. Yes the trace becomes jittery, I'm not sure why. Probably triggering doesn't cope well at this zoom level.

BTW I have just found a but where the scope shows a bogus trace (going bipolar well outside the visible area) if I exceed a certain vertical zoom level. It happens *only* when the frequency is 1024 kHz, and if I set it to 1 kHz (the other frequency of the two that this generator supports), then it happens in a different manner: the lower part of the pulse is shown correctly, but the vertical scale values are way off: e.g., -770 mV, whereas the pulses don't go below zero, save for some ringing. Attaching these screenshots too.

[awakephd]

Two issues that have been discussed above, that I want to be sure I am understanding correctly:

1) With respect to the trace freezing when it is being moved - am I understanding that this behavior is more-or-less the same across different brands / models?

2) Is the inability to overlay a math function on the main display a limitation only of the Rigol, or does the Siglent also have this limitation?

Thanks, all, for the continued flow of information. I can't say that I completely understand everything that has been discussed - I suspect I won't / can't until I get my hands on a unit and can try things out for myself - but I do feel that I am learning a lot, and am gaining more confidence in knowing at least some of what I am looking for.

[shapirus]

An extreme example of that is an aircraft cockpit, where the "normal and good" position of all the meter needles is pointing straight up. That makes it possible to spot which of the 100 measured values is a problem with just a glance.

Glass cockpits lose that advantage, and the workarounds lead to the all-too-frequent last words on cockpit voice recorders: "what is it doing now?".

Another example from the aviation field was described in an article that I read about the F35's touchscreen-infested cockpit which allegedly is (was?) an issue for the pilots who, with switching to this new aircraft, no longer had the tactile feedback that physical buttons and switches provide, which created uncertainty for the pilot with regards to whether he (she?) did actually activate a given control or not. And that's obvious: you don't know if your touchscreen action worked unless you were looking at it when you did it, and sometimes not even then.

[nctnico]

In some cases you are not interested in amplitude or shape but the time between two edges. In such cases it helps to be able to stretch a signal without causing the amplifiers to overdrive too much so the trigger system gets a steep edge to work with. There are not many oscilloscopes which can do this; typically the ones made for high speed signals work well.

[Mechatrommer]

As for the eye movements, it's very funny ... I don't have any problems.

There's nothing funny. Forcing the user to shift focus, unless it is a natural part of the flow in question, is bad, it's that simple. Good UI designers know this and place related elements in such a way that they are all can be seen and read without a single movement of the eyeball. An example would be the central section of a car's dashboard, a wristwatch, or winamp's main window.

An extreme example of that is an aircraft cockpit, where the "normal and good" position of all the meter needles is pointing straight up. That makes it possible to spot which of the 100 measured values is a problem with just a glance.

Glass cockpits lose that advantage, and the workarounds lead to the all-too-frequent last words on cockpit voice recorders: "what is it doing now?".

FFT i dont need to be overlayed since it has different axis, frequency vs time domain no point to overlay imho, its just math like A+B or A-B i need to overlay so i can see correlation of magnitude and phase wrt to signal A and B, but to be able to do both split and overlayed display is more clever, it will give freedom to user whichever is convenient... and suck of DHO800, its only plot in Sinc interpolation ON, we cant turn it off we cant switch to Line vector plot or dots... thats why i said DHO800's GUI is probably designed by a kid who is clueless about signal processing... ymmv.

[tggzzz]

If you don't fit the entire waveform vertically on the display, then that causes other problems. When the input front-ends are overloaded the amplifiers become saturated. Exiting saturation is never timely nor clean. Decent old analogue scopes can be better behaved in that it can take more of an overload to cause saturation; nonetheless it can occur.

Admittedly I did not deep dive into the amplifier settling time measurement problem, but what would other situations be where this might be an issue? What you describe can be said in other words as "extreme zoom", correct?

Yes, naturally there is a limit of the maximum vertical offset of a zoomed-in waveform. I don't have other examples, but I tried this with Rigol DHO800. It's not great, but not terrible, either. Probably good enough for a non-specialized scope. I'm attaching an example of the maximum zoom (scale + offset) possible for the upper portion of a 2.4 Vpp (unipolar) square wave's rising edge.

Apart from a limited offset that can be achieved at a given vertical scale, I see no other issues. The trace can be moved, the time base can be changed etc., and the waveform update happens as usual. Yes the trace becomes jittery, I'm not sure why. Probably triggering doesn't cope well at this zoom level.

BTW I have just found a but where the scope shows a bogus trace (going bipolar well outside the visible area) if I exceed a certain vertical zoom level. It happens *only* when the frequency is 1024 kHz, and if I set it to 1 kHz (the other frequency of the two that this generator supports), then it happens in a different manner: the lower part of the pulse is shown correctly, but the vertical scale values are way off: e.g., -770 mV, whereas the pulses don't go below zero, save for some ringing. Attaching these screenshots too.

That particular use-case is indeed fairly specialised; certainly not something for a beginner to worry about

Overload recovery is operating outside specifications and hence is very device specific and not predictable; you can't extrapolate from one scope to another. That also implies you have to go looking for problems before you can see whether there are or aren't any. You have to visually check at many timescales and with different waveforms; a triangle wave can be useful since the eye is good at spotting deviations from a straight line.

Those first two traces don't show anything useful, as far as I can see.

The final trace is 60mV/div. I'm talking about 1mV/div

0.1% is 1/1000, so if there are 10 divs vertically on the screen, you are overloading the input by a factor of 100.

BTW, my scope is a Philips PM3400: 200ps risetime (1.7GHz), horizontal timebase is 10ps/div to 20µs/div(!), transistors in the signal path are BC107s (f_T = 0.2GHz). A very strange beast indeed; it took me a while to realise it wasn't badly broken Really good value at £25

[tggzzz]

1) With respect to the trace freezing when it is being moved - am I understanding that this behavior is more-or-less the same across different brands / models?

That behaviour is not inherent in a digital sampling scope; it is an artefact of an implementation. Hence you have to suck-it-and-see.

Nonetheless, manufacturers of expensive scopes make sure you realise their scopes are very responsive

[tggzzz]

An extreme example of that is an aircraft cockpit, where the "normal and good" position of all the meter needles is pointing straight up. That makes it possible to spot which of the 100 measured values is a problem with just a glance.

Glass cockpits lose that advantage, and the workarounds lead to the all-too-frequent last words on cockpit voice recorders: "what is it doing now?".

Another example from the aviation field was described in an article that I read about the F35's touchscreen-infested cockpit which allegedly is (was?) an issue for the pilots who, with switching to this new aircraft, no longer had the tactile feedback that physical buttons and switches provide, which created uncertainty for the pilot with regards to whether he (she?) did actually activate a given control or not. And that's obvious: you don't know if your touchscreen action worked unless you were looking at it when you did it, and sometimes not even then.

That's true in modern cars too There are serious moves to legally ban touchscreen operation for the principal controls that are needed when driving.

Find a Tesla showroom, tell them you are driving along a road and you need to adjust the ventilation to clear mist on the windscreen. Get them to show you how to do it. What happens is tragic/comic.

They will all fail to do it within a reasonable time/distance; point out that while "heads down" you ploughed into a queue of children at a busstop.

One had to move icons from somewhere to the taskbar(!).

Another failed, said you could use voice commands - and only succeeded on turning on the seat heater!

[shapirus]

2) Is the inability to overlay a math function on the main display a limitation only of the Rigol, or does the Siglent also have this limitation?

I don't have a Siglent to confirm, but there was one suggestion in one of previous posts that yes it can display a math trace just like a regular channel in the main window.
If this is really so, then it is a big advantage, if you are ever going to use math to correlate a direct input trace with a math trace -- for example, monitoring an IC's input signal on channel 1 and a difference, if you don't have a differential probe, between two nodes in the output section, as an A-B math operation using channel 2 and channel 3 signals as operands (in other words, software emulation of a differential probe).

As a Rigol DHO804 owner, I can say that it creates a feeling of an undercooked prematurely released product to begin with, and of a toy, even if a very capable toy, on top of that. No it's not bad, not even close. It just really sucks that a very decent hardware has been created only to run such a crappy software.

The only reason why I bought it was that getting the Siglent here, at least at even a remotely reasonable price, was and still is not an option, whereas I could get the Rigol for $342 delivered from Aliexpress (sale + publicly available coupons). Frankly, that's about the price higher than which I would not be willing to pay for it (and that on the condition that it is hackable to DHO924 sans the hardware difference), especially now that the Siglent has been released and reviewed.

From what I have read in the respective Siglent review thread, it has its (minor) issues too, but it's a much more mature and serious product overall. Whoever designed the Siglent apparently understood the requirements and expectations of a competent user better than their counterparts at Rigol -- or maybe the latter did not care about that audience at all.

Another nice feature that I'd like to have and that is missing in Rigol, but, I believe, is present in Siglent is that the latter allows you to enter an arbitrary probe multiplier factor. Possible use cases are: 1) DIY probes made of a piece of coax and a resistor (e.g., 21:1 ratio with a 1k resistor and 50 Ohm feed-through terminator); 2) current transformers of whatever mV/A or V/mA ratio; 3) fine tuning of non-ideal 10x or whatever probes to improve amplitude measurement accuracy.

p.s. I am not sure that I would have been fully satisfied with Siglent, either. Reading reviews is one thing, actual usage is another. Take my words with a grain of salt. My critical views are amplified by the fact that I can't stand incompetency in software and UI development, especially in commercial products: do your job well or GTFO. Rigol fails this check.

[tggzzz]

In some cases you are not interested in amplitude or shape but the time between two edges. In such cases it helps to be able to stretch a signal without causing the amplifiers to overdrive too much so the trigger system gets a steep edge to work with. There are not many oscilloscopes which can do this; typically the ones made for high speed signals work well.

The time taken to recover from overload will, in general, depend on the degree of overload and the duration of overload.

You have to be very sure that "slow" recovery hasn't moved the edges.

No point in having a good trigger if the input is so distorted that the edge is in the wrong place.

[shapirus]

and suck of DHO800, its only plot in Sinc interpolation ON, we cant turn it off we cant switch to Line vector plot or dots... thats why i said DHO800's GUI is probably designed by a kid who is clueless about signal processing... ymmv.

Yes good point! Yet another "toy vs professional tool" aspect: I have no other explanation why Rigol does not have a dot trace visualization mode than that the sofware was designed by someone having no clue what an oscilloscope even is. Hell, the dots are already there, was it really so difficult to add a checkbox that disables the drawing of interpolated curves between them? Is it a competence issue or a political decision?

[shapirus]

Those first two traces don't show anything useful, as far as I can see.

Their purpose was to demonstrate the bug that I had found. Note the values on the vertical scale. Then, realize that the input signal is a unipolar square wave, never going below zero (except for a small amount of ringing). Beyond a certain zoom level the displayed values on the vertical axis become screwed -- and not only the displayed ones, but the measured ones as well, as demonstrated in the "zoom-bug-1khz.png" screenshot (look at the measurement widget in the sidebar on the right).

[Mechatrommer]

and suck of DHO800, its only plot in Sinc interpolation ON, we cant turn it off we cant switch to Line vector plot or dots... thats why i said DHO800's GUI is probably designed by a kid who is clueless about signal processing... ymmv.

Yes good point! Yet another "toy vs professional tool" aspect: I have no other explanation why Rigol does not have a dot trace visualization mode than that the sofware was designed by someone having no clue what an oscilloscope even is. Hell, the dots are already there, was it really so difficult to add a checkbox that disables the drawing of interpolated curves between them? Is it a competence issue or a political decision?

either designed by kid, or purposely crippled, but why is beyond my wisdom...

[2N3055]

Two issues that have been discussed above, that I want to be sure I am understanding correctly:

1) With respect to the trace freezing when it is being moved - am I understanding that this behavior is more-or-less the same across different brands / models?

2) Is the inability to overlay a math function on the main display a limitation only of the Rigol, or does the Siglent also have this limitation?

Thanks, all, for the continued flow of information. I can't say that I completely understand everything that has been discussed - I suspect I won't / can't until I get my hands on a unit and can try things out for myself - but I do feel that I am learning a lot, and am gaining more confidence in knowing at least some of what I am looking for.

1. yes
2 I allready answered that. That is only Rigol limit.

[shapirus]

Oh and a cherry on the top: yes, DHO800/900 can into NTP. But it can't into setting the time zone via any UI normally accessible by the user: it is hardcoded to Asia/Shanghai!
Yes you can modify the boot script (the good) to set it to whatever time zone you need, but it has been reported at least twice (myself included) to render the device unbootable (the bad), in my case temporarily -- it later resolved on its own, but I've never tried to change the time zone since then. What about the ugly? Well, you'll know when you look inside the boot script.

[KungFuJosh]

1) With respect to the trace freezing when it is being moved - am I understanding that this behavior is more-or-less the same across different brands / models?

2) Is the inability to overlay a math function on the main display a limitation only of the Rigol, or does the Siglent also have this limitation?

1) Yes.
2) Rigol limitation only.

I highly recommend you thoroughly study @Performa01's excellent thread demonstrating the capabilities of the SDS800XHD series: https://www.eevblog.com/forum/testgear/sds800x-hd-review-demonstration-thread/

[Fungus]

Oh and a cherry on the top: yes, DHO800/900 can into NTP. But it can't into setting the time zone via any UI normally accessible by the user: it is hardcoded to Asia/Shanghai!

AFAIK the 'scope doesn't make any claims about timekeeping, in fact they removed all time displays in the first firware update (presumably because not many users have their 'scope connected to the network so it was useless).

NTP is just a bonus.

[tautech]

I have one such scope, and when there is a 1Vpp square wave, I can "zoom in" vertically to see the end-of transition ringing at 1mV/division. Good luck with that with any other type of scope!

Welcome to 2024. 

Best you study the thread Josh linked.
Just 1 image from it using vertical zoom with a Math zoom as well.

[tggzzz]

I have one such scope, and when there is a 1Vpp square wave, I can "zoom in" vertically to see the end-of transition ringing at 1mV/division. Good luck with that with any other type of scope!

Welcome to 2024. 

Best you study the thread Josh linked.
Just 1 image from it using vertical zoom with a Math zoom as well.

Who is Josh?
What thread?

Why does the LHS display show 20mV/div?

Presuming that is a 2.5Vpp signal, 0.02/2.5 is 1%, a factor of 10 worse than 0.1%

How much do you sell that scope for?

[tautech]

Who is Josh?

The one that implies he does Kungfoo

What thread?

This post:
https://www.eevblog.com/forum/testgear/sds800x-hd-review-demonstration-thread/msg5392730/#msg5392730

How much do you sell that scope for?

They start at $339

[vk6zgo]

Rigol DHO800 sucks in this respect. My definition of "sucks": the waveform update freezes and remains stopped while the vertical (or horizontal) offset encoder is being turned. Analog scopes do not suck: you can turn the offset dials for as long as you want, and the trace(s) will continue to update in real time as if nothing is happening. Do any digital scopes behave this way too?

Do you mean something like this?  (apologies for the unprofessional video)
 


Edit:  input is a sine wave that is being edge-triggered continuously

No not like that. That scope also stops triggering while you are moving waveform up and down. Look at trigger light.
Feed a nonuniform signal (like one of data protocol streams from demo board) and move waveform up and down. Wth stable sine signal you cannot see it stopped changing.

All recent Siglent scopes behave exactly like that R&S you show. Actually, 3000xHD and 6000A restart a smidge faster after you let it go.

But this is exactly what I was arguing all the time: there is no problem.
It it just stubborn nostalgia of "it is not analog" without any usefulness per se.

As the waveform is shifted vertically, the trigger point is lagging the vertical movement of the display, but it does catch up.
An analog 'scope if it is DC coupled & using simple edge triggering will lose triggering if the DC offset control is similarly adjusted.

With the DSO in the video, the display seems to be moving in discrete steps, but fast enough that the waveform appears to be shown twice.
Whether this is just persistence of vision or some artifact of the YouTube video, I have no idea

[BillyO]

The choice is obvious.  The one I choose!!!   

[Fungus]

I can zoom in on the top corner of a square wave, too...



And then add an FFT of that...

[2N3055]

Rigol DHO800 sucks in this respect. My definition of "sucks": the waveform update freezes and remains stopped while the vertical (or horizontal) offset encoder is being turned. Analog scopes do not suck: you can turn the offset dials for as long as you want, and the trace(s) will continue to update in real time as if nothing is happening. Do any digital scopes behave this way too?

Do you mean something like this?  (apologies for the unprofessional video)
 


Edit:  input is a sine wave that is being edge-triggered continuously

No not like that. That scope also stops triggering while you are moving waveform up and down. Look at trigger light.
Feed a nonuniform signal (like one of data protocol streams from demo board) and move waveform up and down. Wth stable sine signal you cannot see it stopped changing.

All recent Siglent scopes behave exactly like that R&S you show. Actually, 3000xHD and 6000A restart a smidge faster after you let it go.

But this is exactly what I was arguing all the time: there is no problem.
It it just stubborn nostalgia of "it is not analog" without any usefulness per se.

As the waveform is shifted vertically, the trigger point is lagging the vertical movement of the display, but it does catch up.
An analog 'scope if it is DC coupled & using simple edge triggering will lose triggering if the DC offset control is similarly adjusted.

With the DSO in the video, the display seems to be moving in discrete steps, but fast enough that the waveform appears to be shown twice.
Whether this is just persistence of vision or some artifact of the YouTube video, I have no idea

Discrete steps are artefact of acceleration. If you turn knob faster it will start going faster.
There are 50-60 refreshes of screen in a second. There is limit how fast can you follow button with that alone.
On Siglent SDS3000xHD and SDS6000A, like I said, it is even smoother a bit. But it is all good and works perfectly well on this R&S scope.