Choosing between entry-level 12-bit DSOs

[ebastler]

Post an example scenario, see if OP can relate to it. 

The OP had specifically mentioned PWM applications, and the whole point of PWM is to represent analog signals, right? So I can see merit in a good, quantitative FFT; e.g. to look for bleedthrough of the carrier frequency or for distortion/harmonics caused by limited resolution. And when you want to optimize your post-PWM lowpass filter, Bode plots might also come in handy.

If one were purely after digital signals, I would question whether a 12-bit DSO is required at all. A fast 8-bit scope with a good logic analyser option -- or even a pure USB-based logic analyser -- might be the better choice. But I don't think that is the OP's situation. As a hobbyist's "swiss army knife", one of the new entry-level 12-bit DSOs should do the job nicely.

[csuhi17]

You measured more than me, you could tell.
Which is the better case, which gives you more freedom?
when measuring at a workbench.
1. When the AVG and the scope are integrated and the reference point of the AVG is connected to the reference points of the scope's channels.

2. Or when you have separate AVG and scope with separate reference points.

Returning to Siglent, I did not see that anyone mentioned the many little things it contains.
Vertical zoom, a wave displayed as a point or vector, smaller icons, so maybe a bigger workspace, the green channel, I see now that it has sin(x)/x and linear interpolation...
If you need it later, then a more usable FFT.

I cannot comment on the frequency of Siglent updates, Rigol has not been updated for 4 months.

[Fungus]

Post an example scenario, see if OP can relate to it. 

The OP had specifically mentioned PWM applications, and the whole point of PWM is to represent analog signals, right? So I can see merit in a good, quantitative FFT; e.g. to look for bleedthrough of the carrier frequency or for distortion/harmonics caused by limited resolution. And when you want to optimize your post-PWM lowpass filter, Bode plots might also come in handy.

Let's see if OP relates to that.

The thing is... OP might get the idea from that the Rigol doesn't have an FFT, or it's somehow broken, which it isn't. It has a very good FFT and very simple to use.

I've used mine use it to look at PWM-based audio signals, it worked perfectly. It also has a very fast update rate, which is valuable (much faster than Siglent's, AFAIK, unless Siglent has changed anything).

If one were purely after digital signals, I would question whether a 12-bit DSO is required at all. A fast 8-bit scope with a good logic analyser option

Do you know one in this price range? MSO5000 costs around $1000 once you factor in the probes, etc.

-- or even a pure USB-based logic analyser -- might be the better choice.

I wouldn't want anything that doesn't have real time display.

But I don't think that is the OP's situation. As a hobbyist's "swiss army knife", one of the new entry-level 12-bit DSOs should do the job nicely.

Yep. But I don't there's a clear "best".

Me? I'll take a better UI over features I'll probably never use. Optimize the things you do all day, every day... not the things you hardly ever do.

[ebastler]

Me? I'll take a better UI over features I'll probably never use. Optimize the things you do all day, every day... not the things you hardly ever do.

I agree that this is probably the most significant difference between the DHO800 and the SDS800X HD series: Rigol's DHO is easier to get started with, since the large dialogs present all relevant controls for a given mode at once, where Siglent sometimes has to resort to multi-level menus. Siglent's SDS, on the other hand, is the more "serious" scope, offering some additional functionality which Rigol does not have.

Whether that makes Rigol's UI "better" overall is debatable. Personally, I found that I got used to Siglent's UI very quickly, and I do appreciate that they use the limited display area a bit more space-consciously than Rigol do.

Regarding the "more serious" functionality, I would not necessarily focus on special modes like Bode plots. Having a trace history always available in the background; being able to capture frames in much more rapid succession; having full control over sampling rates so I can choose to trade time resolution for a much larger number of stored frames; having a high-resolution trigger counter always visible are examples of the differences I consider more important. The Siglent appears better thought-through to me in many little details. (Not surprisingly, since it is built on a software platform that has already gone through a few iterations.)

And regarding the rarely-used features, I do find it reassuring to have them available. It's not so much a matter of "optimizing" these, but of being able to do e.g. automated Bode-Plots at all or not, right?

(Edit: Typos.)

[ebastler]

The thing is... OP might get the idea from that the Rigol doesn't have an FFT, or it's somehow broken, which it isn't. It has a very good FFT and very simple to use.

Since you specifically brought that up: I'm afraid Rigol's FFT is broken in a few respects.

At least one of the window functions (Flattop, the important one if you want true peak heights) is wrong, resulting in nasty artefacts in the spectrum. And the displayed spectral resolution of the FFT is wrong on multiple counts: It is designated as "RBW" but is actually meant to be a frequency step; it is unclear which unit it is meant to be in; and if you switch your scope's time base back & forth, the "RBW" can get confused and does not return to its previous value. (The last point assumes it behaves the same as on the DHO1074 I had.)

FFT on the DHO (and all other Rigol scopes?) also lacks Averaging and Peak Hold modes. That does not make it "broken", but limits its usefulness.

[Martin72]

From DS1000Z to MSO8000, all FFT features are the same, even their biggest model (DS70000)do not have averaging or peak.
And that the FFT sometimes doesn't do what it should do is a known fact, at least with the 1000Z, MSO5000 and now DHO.
I don't know whether this is different with the 7000,8000 and DS70000.
But it's doubtful, I'm afraid Rigol hasn't cared that much about the FFT thing for years.
Which is surprising, since they don't offer any (my guess) bad SA.
Maybe this is a different department and they should talk to each other.

[Mechatrommer]

The thing is... OP might get the idea from that the Rigol doesn't have an FFT, or it's somehow broken, which it isn't. It has a very good FFT and very simple to use.

Since you specifically brought that up: I'm afraid Rigol's FFT is broken in a few respects.
At least one of the window functions (Flattop, the important one if you want true peak heights) is wrong, resulting in nasty artefacts in the spectrum ....

since i suspect this thread is going to be another hanging rig's vs sig's dick contest based on FW bugs/flaws etc that yet to be fixed somewhere soon (or simply dont use flat top window?). at least you can search available open source code FFT library in the net, learn it and hence you got some idea what all these windowing delusion all about. even wiki got good explanation on them without getting into the mess of the C/C++ syntaxes.. heck why do you want flat top so much if you know it has a bug, when you have another windowing option such as backman or hamming etc?

and then if you want to get further... try to implement it from downloadable raw data from scope to PC. this has been proven in rigol and i havent seen someone did it with siglent, the argument i usually heard, heck who want to connect scope with usb cable to PC? the same persons keep playing their scope with even longer and bigger ethernet cable, i'm not sure how people process things but all these relevant only if, FFT really matter so much, if it does, you can get a decent affordable used SA in ebay up to GHz...

5Mpts FFT from Rigol DHO924S (hacked from DHO804) PC plotted if its really matters.. (what class 1MPts? snort) btw i have no project requiring close inspection of spectral analysis, so i cant prove anything as a case study... only this random noise floor plot with VisaDSO ymmv...

[ebastler]

at least you can search available open source code FFT library in the net, learn it and hence you got some idea what all these windowing delusion all about. even wiki got good explanation on them [...] heck why do you want flat top so much if you know it has a bug, when you have another windowing option such as backman or hamming etc?

I am not sure what this has to do with searching source code libraries, but in any case there is no need for a rant on your part. I explicitly stated in my post why one may prefer (or even need) to use the flattop window: "Flattop, the important one if you want true peak heights." And you can read that in Wikipedia too if you prefer:

Flat-top window
A flat top window is a partially negative-valued window that has minimal scalloping loss in the frequency domain. That property is desirable for the measurement of amplitudes of sinusoidal frequency components.[17][42]

https://en.wikipedia.org/wiki/Window_function#Flat_top_window

[Mechatrommer]

I am not sure what this has to do with searching source code libraries

if you diy your own fft app, you can control it and verify signal all you like, more freedom. not relying on toyish implementation of FFT in cheap resource limited scope. ymmv.

Flat-top window
A flat top window is a partially negative-valued window that has minimal scalloping loss in the frequency domain. That property is desirable for the measurement of amplitudes of sinusoidal frequency components.[17][42]

https://en.wikipedia.org/wiki/Window_function#Flat_top_window

imho you should read all, esp intro at the beginning to get idea of the purpose, pro and con... there are many kind of signal, some signal suitable for different window, and i believe signal attenuation due to windowing can be estimated, esp if you have a known signal source as reference. ymmv.

[awakephd]

Again, many thanks to all. I appreciate that (at least so far) the posts have not veered too far into flame-war territory. (Only a bit warm from time to time. )

To further unpack my limited understanding of my needs: I'm not sure if I made it clear that I am operating on a pretty low-level hobby basis. My knowledge of electronics is entirely self-taught, often by trial-and-error as I cobble together a circuit.

Here is a specific example, if you are bored and in need of a good laugh at my feeble efforts. My most recent (and still current) project is a replacement for the controller in a "touchless trash can" - specifically an older version of this model: https://www.bestbuy.com/site/insignia-18-gal-automatic-trash-can-with-recycle-and-waste-divider-stainless-steel/6409629.p?skuId=6409629. No, I am not shilling for them; keep in mind that I am making a new controller because the old one died, and while it was very convenient when it was working, it is way too expensive to throw the whole thing away over the failure of $5 worth of electronics - not exactly a ringing endorsement of the product!

I designed a replacement circuit built around an ATTiny84 and an SR602 IR motion sensor. I worked up a PCB design and ordered it from one of the usual suspects - they did a nice job, and I populated it with no trouble. And re-populated it. And did some more re-work, and some more, as I have debugged the design. The main bug was using the wrong footprint for the SOT23 transistors (!!!), but fortunately I am able to turn them upside down to fix that issue. It took several tries to settle on a suitable resistor value to drive the transistors. Did I mention trial-and-error? I kinda, sorta know what I am doing, but especially with anything analog, I generally am partly guessing on the values to use. I'm at the point where I am about ready to send off a revised PCB design - I can almost fudge the first prototype into doing everything I need it to, but I have realized that I really need to reverse the polarity on the switches so that they are normally low rather than normally high - or at least, I think I need to do that in order further to reduce power consumption. The new design will let me jumper-select which polarity to use, so I can try it both ways.

So where might a scope come into play for this example? The lid has a spring on it which allows a relatively low-power gear-reduction motor to open it and close it. To determine how long to apply the motor, there is a sensor wheel that interrupts an led-and-photosensor as the motor spins. The trick is to get the right count of pulses to represent the motion upwards or downwards. With my old Tektronix, I could get a nice square wave off the sensor when the motor was disconnected from the lid - power up the motor and get a continuous pulse train. That let me know the timing I was working with for the pulse train, but it did nothing to help me know how many pulses to count when powering the motor up or down. That's where I was really wishing for a DSO, so that I could manually lift the lid and see how many pulses were generated.

Or at least, I am thinking that a DSO would have helped me do this. Again, I've never used one, so I don't know if my expectations for what it can do are accurate! In the end, I have trial-and-errored until I found a suitable count to use.

Note again that more sophisticated things, such as FFT's and Bode plots and function/waveform generators and logic analyzers, are not even on the horizon for what I have done on this project. I do more-or-less understand what an FFT is, but I don't know of any reason I would need it for this sort of project - though if there is a way that it would be helpful, I definitely want to learn! Meanwhile, I do not know what a Bode plot is, so have absolutely zero clue whether or how it could have been helpful in this example. As several have mentioned above, it is highly unlikely that I will ever spend the considerable extra to add the logic analyzer or waveform generator to any DSO I buy, so for me that potential advantage of the Siglent SDS804 is moot. Of course, if someone wants to give me a fully equipped multi-signal capable scope, I won't turn it down ...

I with I could take the advice to buy both scopes and then return one. Three major problems with that approach are 1) convincing my spouse; 2) finding enough time in the first 30 days to do the testing / comparison; and 3) convincing my spouse.

If you have read all the way to the bottom of this post, you deserve an award!

[awakephd]

Oh, I forgot to address one other point - a couple of times posters have mentioned that an 8-bit scope might be more than enough for my needs. Yes, very likely so - I'm open to suggestions!

My impression is that at this point, the 8-bit entry level scopes are either less-favorable brands, some of which I cannot figure out how to pronounce (FNRZKI or something like that?) or models that are 10 years old, and don't offer much if any advantage in price. Is there any reason to buy a Siglent SDS1104X-E instead of the newer, 12-bit, 2G-sample, SDS804X-HD? Especially when, at least from what I'm seeing, the 1104 is 10% more expensive than the 804? And surely even less reason to buy a Rigol 1054 in place of the new DHO804?

Please, please, correct me if I am wrong - that's why I started this thread, because I now know just enough to have some sense of how very much I do not yet know!

[ebastler]

Is there any reason to buy a Siglent SDS1104X-E instead of the newer, 12-bit, 2G-sample, SDS804X-HD? Especially when, at least from what I'm seeing, the 1104 is 10% more expensive than the 804? And surely even less reason to buy a Rigol 1054 in place of the new DHO804?

Only if you are on a very tight budget, and can find a used SDS1104X-E or DS1054Z for a much lower price than new, I would say.

Otherwise I completely agree with you thinking: Get one of the recently introduced 12-bit models. As mentioned earlier, I would consider them the "Swiss army knife" for the hobbyist. They should have you covered not only for today's needs, but also when you take your hobby projects further over the next few years.   

[tautech]

With my old Tektronix, I could get a nice square wave off the sensor when the motor was disconnected from the lid - power up the motor and get a continuous pulse train. That let me know the timing I was working with for the pulse train, but it did nothing to help me know how many pulses to count when powering the motor up or down. That's where I was really wishing for a DSO, so that I could manually lift the lid and see how many pulses were generated.

Or at least, I am thinking that a DSO would have helped me do this. Again, I've never used one, so I don't know if my expectations for what it can do are accurate! In the end, I have trial-and-errored until I found a suitable count to use.

Many recent models offer a Measure Pulse count, either + and perfect for your needs.

Meanwhile, I do not know what a Bode plot is, so have absolutely zero clue whether or how it could have been helpful in this example.

While not useful for your immediate task Bide plots are useful for filters, stop/pass band etc.
You do need a AWG to provide a frequency sweep for the scope to plot the result of the filter.
This is something you could need further down the track as skill and understanding grows.

[KungFuJosh]

You're correct to stick with the new 12bit stuff, there's no point in looking at outdated 8bit entry level scopes.

Today, I think you'd be happy with either of the scopes you're looking at. I think Siglent Bodes better for long-term growth if you get interested in learning and having additional potential without needing to replace the scope.

I'm not sure if it applies to this scope or not, but there is a DIY thread for making one of the Siglent logic probes if that's something you'll ever care about.

[ebastler]

I'm not sure if it applies to this scope or not, but there is a DIY thread for making one of the Siglent logic probes if that's something you'll ever care about.

That would be a DIY of the passive probe used with the higher-end scopes (that have the electronics for the logic analyser built in), right? Or is there actually a DIY version of the active external unit (FPGA-based) which is required for the entry-level SDS scopes?

[Mechatrommer]

My impression is that at this point, the 8-bit entry level scopes are either less-favorable brands, some of which I cannot figure out how to pronounce (FNRZKI or something like that?)

FNIRSI? dont touch that shit. you created the thread with the right spot already, rigol or siglent, nothing else..

I designed a replacement circuit built around an ATTiny84 and an SR602 IR motion sensor...

your story is not much different than i am, except your is better. i didnt even know what a pcb silkscreen is when i first came here. i suspect when you get your first DSO is like extra excited like a kid get an expensive toy. but this game can be intoxicating, since there are so much more you can do with DSO, stick with what you have now is like only using partially the potential of your DSO. i was with DS1052E back then and its several aspects more crippled than newer DSO, with today possibility, DSO+AFG+LA is cheaper than before... you can use your imagination the level of "toxicity" you are exposing yourself into

another advice. if you get the latest 12bit version, it can do you good in long run. used/older model may have shorter lifetime due to usage by old owner, or simply lack of modern features. sure 8 bit is perfectly fine for the job you described, but if for some reason you want to change mind, you'll regret it if it only can save you a few bucks. ymmv.

what is bode plot? and LA? i dont want to get to it, it probably could harm you financially, or simply you are not ready for it. you can google for yourself if you want to.

[tggzzz]

A few semi-random observations.

Some low-end scopes have a large capture buffer, but the post processing (FFT, digital decode, etc) only operates on what's displayed on the screen. That is a significant limitation.

If FFTs are a significant use case, make sure you understand the significance of the ENOB on the frequency spectrum.

For simple assessment of digital waveform signal integrity, 8 bits is perfectly sufficient. The first 1GHz HP digitising scope I used had, IIRC, 6 bits.,

Some of the low-end digitizing scopes I've played with have GUIs that are appallingly laggy compared with analogue scopes.

[awakephd]

I was with you up to "ENOB" - I confess that I don't have a clue what that is, nor what its significance is.

I did a little reading on Bode plots just a bit ago - I now see where they could be very useful in some circumstances. Am I correct in thinking that the "magic" of the DSO Bode plot is that it gives you the whole plot across the frequency range automatically? As opposed to measuring amplitude and phase shift at, say, a dozen discrete points along the frequency range, and creating a rough graph? (I may have just revealed how little I understood of what I read ... !)

[tggzzz]

I was with you up to "ENOB" - I confess that I don't have a clue what that is, nor what its significance is.

I did a little reading on Bode plots just a bit ago - I now see where they could be very useful in some circumstances. Am I correct in thinking that the "magic" of the DSO Bode plot is that it gives you the whole plot across the frequency range automatically? As opposed to measuring amplitude and phase shift at, say, a dozen discrete points along the frequency range, and creating a rough graph? (I may have just revealed how little I understood of what I read ... !)

ENOB is effective number of bits, less than the number of bits coming out of the ADC. (See my .sig )

You know what a Bode plot is from university lectures: log frequency, log amplitude plus phase. An AWB plus post processing captured signals allows them be calculated and displayed. For an example, download the Digilent Waveforms GUI; you can play with it without having the hardware.

[Mechatrommer]

I did a little reading on Bode plots just a bit ago - I now see where they could be very useful in some circumstances. Am I correct in thinking that the "magic" of the DSO Bode plot is that it gives you the whole plot across the frequency range automatically? As opposed to measuring amplitude and phase shift at, say, a dozen discrete points along the frequency range, and creating a rough graph? (I may have just revealed how little I understood of what I read ... !)

you can plot bode by hand with eyeballing magnitude and phase. built in bode plot automates that so you are correct, this for area of filters and amplifiers, not cup of tea for many. i made bode plotter PC SW just for experimentation for DS1052E and Hantek DDS3X25 AWG back then, when built in bode plot on cheap DSO was so alien. on latest DHO800, hacking the LA port and probe too is now possible, it wasnt on earlier model, what a wonderful time.

[Fungus]

Oh, I forgot to address one other point - a couple of times posters have mentioned that an 8-bit scope might be more than enough for my needs. Yes, very likely so - I'm open to suggestions!

There is literally no point in buying a new 8-bit 'scope now.

They're not cheaper, and they all feel like antiques compared to this new generation.

[Fungus]

If FFTs are a significant use case

They aren't!

[Fungus]

While not useful for your immediate task Bide plots are useful for filters, stop/pass band etc.

Or you can inject white noise and look at the FFT. Much simpler.

Yet another oscilloscope thread derailed by the eternal "FFT" discussion, even though it's stated as NOT a requirement.

[Martin72]

There is literally no point in buying a new 8-bit 'scope now.

They're not cheaper, and they all feel like antiques compared to this new generation.

However, this is not due to the 12Bit, but to the fact that Siglent/Rigol has also introduced a new UI in the low-cost range with the 12Bit.
They could have done the same with 8bit.

[Fungus]

There is literally no point in buying a new 8-bit 'scope now.

They're not cheaper, and they all feel like antiques compared to this new generation.

However, this is not due to the 12Bit

Correct. It's because of the touch screen, etc.

The 12 bits and low noise is just icing on the cake.