Siglent SDS1104X-E vs. Rigol DS1054Z Advice?

[Fungus]

As a side question. How is the noise on the new Rigol DS7000 series? This is much lower han lower Rigol series and Siglent?

I'm not sure anybody has one yet (apart from Dave).

[rf-loop]

As a side question. How is the noise on the new Rigol DS7000 series? This is much lower han lower Rigol series and Siglent?

From Dave's video.
This may give some preliminary idea before further tests (if we can see real professional grade tests and measurements or just entertainment show as usually - this we know later)

Note that in this image there is BW limit on. I have not get any answer what is BW limit 20M, 100M or what? Also it looks like there is not much sample data on image...only 25k one wfm.
If it is 20M... better that I do not say anything if it looks high or low... (need note that 1mV/div is magnified from 4mV/div what is most sensitive full resolution)

[2N3055]

Note that in this image there is BW limit on. I have not get any answer what is BW limit 20M, 100M or what? Also it looks like there is not much sample data on image...only 25k one wfm.
If it is 20M... better that I do not say anything if it looks high or low... (need note that 1mV/div is magnified from 4mV/div what is most sensitive full resolution)

It has two settings : 20 MHz and 250MHz.
But, unfortunately when making DS7000 they made it as bad as Keysight DSOX3000T, which also has 1 mV/DIVand 2 mV/DIV magnified in software from 4 mV/DIV, and pretty high noise.
Both of those scopes are fast, but not low noise scopes. Actually, DS7000, in many ways is "inspired" by philosophy of Keysight DSOX 3000/4000 series: high wfms/sec, hardware decoding, etc etc..
Except they put much more memory on it... Low noise, high sensitivity input channels were not high on a priority list..

[Fungus]

....Keysight DSOX3000T also has 1 mV/DIVand 2 mV/DIV magnified in software from 4 mV/DIV and pretty high noise.
...
Low noise, high sensitivity input channels were not high on a priority list..

Maybe it's simply not a problem for real work. Maybe it's mostly something for volt-nuts to obsess over in forums.

Realistically, how much of a problem is <1mV of noise on (eg.) a 5V system? A 1V system?

Even a lowly DS1054Z has a "high res" mode which will eliminate all random noise on a repetitive signal by averaging multiple waves together. Any noise you see in high res mode is in the signal, not in the 'scope.

[nctnico]

....Keysight DSOX3000T also has 1 mV/DIVand 2 mV/DIV magnified in software from 4 mV/DIV and pretty high noise.
...
Low noise, high sensitivity input channels were not high on a priority list..

Maybe it's simply not a problem for real work. Maybe it's mostly something for volt-nuts to obsess over in forums.

Realistically, how much of a problem is <1mV of noise on (eg.) a 5V system? A 1V system?

Even a lowly DS1054Z has a "high res" mode which will eliminate all random noise on a repetitive signal by averaging multiple waves together. Any noise you see in high res mode is in the signal, not in the 'scope.

If you did any real work with an oscilloscope you'd know the answer: noise does matter because the noise doesn't get smaller when setting the input to a less sensitive setting. So it is easy to get a band of noise across the screen (I hated my Agilent DSO7104 for that even with BW limit on) which obscures the interesting part of a signal. Try to do accurate cursor measurements on a band of noise. Sure high-res mode can do some filtering but you get aliasing artefacts when zooming out and usually you can't use it together with peak detect so you miss other parts in a signal. All in all the less noise an oscilloscope has the more convenient it is to use.

[JS]

....Keysight DSOX3000T also has 1 mV/DIVand 2 mV/DIV magnified in software from 4 mV/DIV and pretty high noise.
...
Low noise, high sensitivity input channels were not high on a priority list..

Maybe it's simply not a problem for real work. Maybe it's mostly something for volt-nuts to obsess over in forums.

Realistically, how much of a problem is <1mV of noise on (eg.) a 5V system? A 1V system?

Even a lowly DS1054Z has a "high res" mode which will eliminate all random noise on a repetitive signal by averaging multiple waves together. Any noise you see in high res mode is in the signal, not in the 'scope.

  High res doesn't do that, average does... In average you can program the number of captures to average. High res uses oversampling or a moving average. In both cases it will filter out the input noise as well as the random noise from your source which you might want to look at. Also, you can't use average for single shot capture.

  Lower sensitivity ranges only differ in not being digitally amplified, so only the first ranges, which are zoom of a higher range gets higher noise, then the noise stays pretty much the same, could change a bit depending on the gain of each stage but can even get higher at some point where the Eon dominates over Ein.

  You could use a preamp to go to the lowest ranges and keep the noise as the others, but you can't get very much from them, and they are kind of a pain, expensive if you buy one or hard to build properly.

I don't know why scopes don't have other filtering options, if you want to look at an audio signal a 100kHz BW limit could do a lot of good and implementing this in SW could be quite nice. Having one filter or two per decade would be great. High res does something like this but as I said it's not very straight forward knowing what it's doing, and as you don't know you don't know what it could be doing behind, it will change the response as you change the horizontal scale. With math you also can do something like that but it's not great, sone times helps to have a better look at the signal but doesn't quite cut it.

JS

[2N3055]

....Keysight DSOX3000T also has 1 mV/DIVand 2 mV/DIV magnified in software from 4 mV/DIV and pretty high noise.
...
Low noise, high sensitivity input channels were not high on a priority list..

Maybe it's simply not a problem for real work. Maybe it's mostly something for volt-nuts to obsess over in forums.

Realistically, how much of a problem is <1mV of noise on (eg.) a 5V system? A 1V system?

Even a lowly DS1054Z has a "high res" mode which will eliminate all random noise on a repetitive signal by averaging multiple waves together. Any noise you see in high res mode is in the signal, not in the 'scope.

It depends on what you do.. If you work with analog, low level sensors, power supply stuff.. Yes, than it might be important. I agree that for general work you might not need it. And yes. Hires mode helps a lot.  It will also reduce effective sampling frequency for same factor. So you end up with a 5 MHz scope.  Real low noise and sensitivity with full bandwith cannot be replaced sometimes.  I have a real 12 bit scope and it is amazing what detail can you see sometimes...

Point is, I wasn't criticizing, quite the opposite. Many people swear by Keysight DSOX3000T series as "best scope to use" and it has same input sensitivity specs as ds7000.  So it might not be important to many people out there.  Many people need good MSO for their mixed signal work, not for low level analog stuff.

[nctnico]

I don't know why scopes don't have other filtering options, if you want to look at an audio signal a 100kHz BW limit could do a lot of good and implementing this in SW could be quite nice.

GW Instek and MicSig oscilloscopes have freely settable filters (low-pass, high-pass and band-pass) so it already exists and these are very useful for some purposes. Not just for audio but try looking at a signal with a noisy DC-DC converter nearby.

[KungFuJosh]

I'm reading the manual for the Siglent while I'm waiting for it to arrive. A couple things I noticed I like better already.

For example, on the Rigol, to set the scope to default you have to press Storage, and then press Default on the soft buttons.

On the Siglent it's a single "Default" button in a good spot. That works well for me.

One thing I think is stupid that the Siglent shows in the manual:

"Curr means the current memory depth."

Why wouldn't they just write Mem? Cur means a crappy dog. Is that what they want their scopes associated with? Crappy dogs? 

[tautech]

I'm reading the manual for the Siglent while I'm waiting for it to arrive. A couple things I noticed I like better already.

For example, on the Rigol, to set the scope to default you have to press Storage, and then press Default on the soft buttons.

On the Siglent it's a single "Default" button in a good spot. That works well for me.

Have a study up on User definable Default, a very useful feature. 

One thing I think is stupid that the Siglent shows in the manual:

"Curr means the current memory depth."

Why wouldn't they just write Mem? Cur means a crappy dog. Is that what they want their scopes associated with? Crappy dogs? 

Page # ?
Please offer a suggested manual edit ?

[KungFuJosh]

On the Siglent it's a single "Default" button in a good spot. That works well for me.

Have a study up on User definable Default, a very useful feature. 

I've heard it has lots of customizable buttons and such, haven't gotten that far yet in the manuals. But I like the default button, default. I actually use the auto set and default buttons. They're both useful to me. I don't see the benefit of twiddling first and tuning later. I like to hit auto set and twiddle after.

One thing I think is stupid that the Siglent shows in the manual:

"Curr means the current memory depth."

Why wouldn't they just write Mem? Cur means a crappy dog. Is that what they want their scopes associated with? Crappy dogs? 

Page # ?
Please offer a suggested manual edit ?

PDF page 47, listed as page 23 in the document. #8

8. Sampling Rate/ Memory Depth
Display the current sampling rate and memory depth. Sa means the current sampling rate and Curr means the current memory depth.

Obviously it should say "Mem" for memory depth. Why would it say Curr for the word Current when it isn't measuring current? Does that make any sense? The simples answer is the correct one. Sa, or preferably Sam for sample, and Mem for memory.

[bitseeker]

Josh, does copy editing Siglent's manual count as "free debugging?" (You know I had to ask.)

[wpwrak]

Josh, does copy editing Siglent's manual count as "free debugging?"

He's not talking about an error in the manual. "Curr" is what appears on the screen, and I agree that "Mem" would be much clearer.

[tautech]

Josh, does copy editing Siglent's manual count as "free debugging?"

He's not talking about an error in the manual. "Curr" is what appears on the screen, and I agree that "Mem" would be much clearer.

Hmmm, I see both points of view as in Curr as pertaining to Mem depth with the 'current' setting and the same applies to the sampling rate wherein it changes too due to scope settings and/or active channels.

Whatever they're called there will be some confusion for a DSO newbie. Curr, Pres, Exist, Mem and so on as each is not a static value. Curr clearly points to 'with current settings'.

Interesting point though, maybe it should migrate into the main thread about these scopes so those that have posted there can get a chance to add their 2c.

[Fungus]

If you work with analog, low level sensors, power supply stuff.. Yes, than it might be important.

In those cases "low noise" isn't enough. To make it useful you also need really good DC offset ability, etc.

(I suspect this is the real reason the DSOX3000T/DS7000 designers haven't made any special efforts to reduce noise - opamps that can recover at >5Ghz when massively overloaded are difficult to find so there's no point in trying to please the "low noise" market segment)

And yes. Hires mode helps a lot.  It will also reduce effective sampling frequency for same factor. So you end up with a 5 MHz scope.

Jitter aside, the rise times, etc., will be identical in hires mode so I'm not sure where that number came from.

Real low noise and sensitivity with full bandwith cannot be replaced sometimes.

True, but if you really need that then you have to buy a 12-bit 'scope.

The bottom bit of a high-speed ADC is always a bit random so an 8-bit ADC leads to approx. 1mV of noise.

[Fungus]

I'm reading the manual for the Siglent while I'm waiting for it to arrive. A couple things I noticed I like better already.

For example, on the Rigol, to set the scope to default you have to press Storage, and then press Default on the soft buttons.

On the Siglent it's a single "Default" button in a good spot. That works well for me.

Be sure to report back on how annoying it is to choose/toggle the on-screen measurements,

[KungFuJosh]

Josh, does copy editing Siglent's manual count as "free debugging?" (You know I had to ask.)

  Hahahaha, that's a good point. But I'll give them this one for free if I don't have to stare at something that dumb on the screen.

[KungFuJosh]

Be sure to report back on how annoying it is to choose/toggle the on-screen measurements,

I know what you're referring to, but I don't know if I'll use most of them? The Siglent's menu is customizable from what I've read, and when you press Measure, FFT is supposed to be one of the first things that's supposed to show up. We'll see. If I find it annoying, I'll let you know. I didn't press any of the left side buttons on the Rigol, even though the little pictures were cute.

[KungFuJosh]

Josh, does copy editing Siglent's manual count as "free debugging?"

He's not talking about an error in the manual. "Curr" is what appears on the screen, and I agree that "Mem" would be much clearer.

Hmmm, I see both points of view as in Curr as pertaining to Mem depth with the 'current' setting and the same applies to the sampling rate wherein it changes too due to scope settings and/or active channels.

Whatever they're called there will be some confusion for a DSO newbie. Curr, Pres, Exist, Mem and so on as each is not a static value. Curr clearly points to 'with current settings'.

Interesting point though, maybe it should migrate into the main thread about these scopes so those that have posted there can get a chance to add their 2c.

Both settings are "Curr" - what other point of view is there? Setting 1 is "Current Sample Rate" and setting 2 is "Current Memory Depth." There's a couple of different ways to abbreviate both settings that make sense. Neither have anything to do with crappy dogs. I speak a little Chinese. I know what left over Chinglish looks like.

Xiexie,
Josh

[2N3055]

Jitter aside, the rise times, etc., will be identical in hires mode so I'm not sure where that number came from.

"High Resolution"
The average of n captured sample points is recorded as one waveform
sample. Averaging reduces the noise, the result is a more precise
waveform with higher vertical resolution.

"Average" The average is calculated from the data of the current acquisition and
a number of consecutive acquisitions before. The method reduces
random noise. It requires a stable, triggered and repetitive signal.
The number of acquisitions for average calculation is defined with
"No. of Averages"

That is from Lecroy manual but description is accurate.

[bitseeker]

Josh, does copy editing Siglent's manual count as "free debugging?" (You know I had to ask.)

  Hahahaha, that's a good point. But I'll give them this one for free if I don't have to stare at something that dumb on the screen.

Right on. That's why we do what we do. It's all good. And it's good to see you've got a good sense of humor, too.

Cheers!

[bugi]

Jitter aside, the rise times, etc., will be identical in hires mode so I'm not sure where that number came from.

"High Resolution"
The average of n captured sample points is recorded as one waveform
sample. Averaging reduces the noise, the result is a more precise
waveform with higher vertical resolution.

"Average" The average is calculated from the data of the current acquisition and
a number of consecutive acquisitions before. The method reduces
random noise. It requires a stable, triggered and repetitive signal.
The number of acquisitions for average calculation is defined with
"No. of Averages"

That is from Lecroy manual but description is accurate.

"... description is accurate." ?

Those descriptions are not complete, are quite simplified, and claiming that the "averaging" gives a more precise waveform is IMHO a bit misleading, or it should at least explain the cases where the result will not be more precise. As an extreme example, a suitable long average over a suitable stable periodic signal would give a flat line; yeah, it is more precise in showing the average value, but completely incorrect in showing the input waveform, for which oscilloscopes are mainly meant

Those descriptions give a good start for some people, but one can not really go very far with only that info. With (quite) a bit of math knowledge, one can derive the missing effects by oneself, but, when the descriptions are at the level shown above, they are clearly aimed at people who would not be able to do those derivations (Even beginners on this category of math do not need to be even mentioned what averaging will do for noise, specifically to random noise, but there they are.)

An example of missing info: What does the averaging of sequential samples (the "high resolution" above) do to bandwidth or step response?
An example on potential difference between theory and explained feature: Averaging multiple (well triggered) waveforms/acquisitions (the "average" above) would theoretically allow both reducing noise and giving higher resolution (with some limitations), not only reducing noise. So the result depends on the implementation.

So, yeah, I would not call them "accurate" descriptions

On the terminology:

As is obvious from above, both methods use averaging (but samples are chosen differently), and both can/could theoretically give higher resolution (and yet do not have to give higher resolution) and both can give lower noise, so which term to use for which?  Lecroy has chosen that particular way, and it is good they explained it in the manual, even if that crudely.  Other people might have ended up using the terms in a different way.

IMHO, better terms would be by the method, e.g. "moving average" and "multi-waveform average" or something like that, because those are unambiguous (or at least less ambiguous), though the even still the chosen effects and trade-offs are unknown unless explained in a manual (especially, if they give also higher resolution and not just noise reduction). But then again, the less-mathematically aware users would then not know what basic effects those will have (unlike "high resolution" which directly tells something to everyone). Without other explanation, "high resolution" effect could be achieved by multiple methods, each with different set of side-effects.  Lecroy (and I think many other scopes) decided to mix and match, one feature termed by the method, other by the apparently most important effect.  It gets even weirder when one mode is "hires" and another "eres"... which one does exactly what and how?... I wonder how many knows the answer without peeking into the manual (or scope's help popups).

[2N3055]

Those are stupid commercial names given by scope manufacturers.
Those explanation are not wrong, therefore they are accurate. They are simplistic, not mathematically rigorous nor meant to be  "accurate algorithmic description".

You understood correctly (accurately) that first name refers to running average filtering. Also that other name represents making averages between separate trigger events.

Which have different consequences:

"High Resolution" -  "moving average" - Represent FIR lowpass filter... Therefore it will filter out high frequency spectra above cutoff frequency. Noise will decrease basically in higher spectra above cutoff. It will work fine on random, non- repetitive signals, albeit with bandwidth of scope reduced to filter cutoff response...

"Average" - "multi-waveform average" "multi-screen average" - It will average N number of capture buffers, mapping point on point in different captures. It is repetitive sampling, works only on highly repetitive signals. It keeps full bandwidth of scope. There will be some noise averaging if noise is Gaussian, but will usually have additional noise in time domain from trigger jitter.

[KungFuJosh]

Translation for simple people like me:

Hi Res: Scope lowers bandwidth and averages some crap to pretend to be better quality.

Average: Scope removes data and averages some other crap to pretend to be better quality.

[tautech]

Translation for simple people like me:

Hi Res: Scope lowers bandwidth and averages some crap to pretend to be better quality.

Average: Scope removes data and averages some other crap to pretend to be better quality.

Despite how you translate it at some time each of these tools will be found to be indispensable !