Choosing between entry-level 12-bit DSOs

[nctnico]

With a better, higher order reconstruction filter you can achieve much better results compared to sin x/x at the expense of computational power (I did some experiments with this a long time ago).

i bet your higher order reconstruction filter is one of sin x/x derivative:
https://en.wikipedia.org/wiki/Reconstruction_filter
https://en.wikipedia.org/wiki/Whittaker%E2%80%93Shannon_interpolation_formula

about "anti-aliasing filters which prevent violating Nyquist", we went through that rigol is not properly BW limited at 4CH turned on, and Sinc reconstruction implementation is a bit broken. but your earlier post is too general, so i'm agitated to reply... such as this...

The real problem is that you can only use Fourier series to construct continuous functions. You can't use Fourier series to construct functions with a step in them like a square wave. However, when sampling a square wave like signal (which in the real world can never be a step function) it can turn into a step function in the digital world due to insufficient samples to follow the edges. And as a result you'll get Gibb's ears when applying the sin x / x filter to the sampled signal

some part of it imho is misleading, esp the bolded line. a properly bw limited scope and correct implementation of Sinc derivative filter will not produce gibbs ear, thats not me saying, thats from theory saying (at least what i understand). ymmv.

That is true and I'm not denying that in any way. Actually, I wrote that earlier on. But in the real world DSO manufacturers push the bandwidth up high so you can get as close to Nyquist as possible and still get a decent sine wave. At the cost of some signals getting artefacts (like a square wave getting Gibbs ears). And as I wrote a few posts earlier, if you don't want Gibbs ears, then you need to turn the bandwidth limit on. But this will reduce bandwidth. IOW: the DSO manufacturers leave it up to the user whether they want the most bandwidth or the best signal representation. Given the fixed, maximum samplerate you have to choose between one or the other. And again, sin x / x reconstruction is rather crude so that doesn't help.

[tggzzz]

Again, I may be getting this totally wrong ... I'm trying to follow the discussion, but not sure I am on the right track.

read the basic below... dont be deluded by nobody who says only him can understand sampling theory and the rest cant...
https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem
https://en.wikipedia.org/wiki/Gibbs_phenomenon

The trouble starts because too many people think they understand Shannon/Nyquist, but don't and then mis-apply it in real life situations. Reading and re-reading the definitions rarely helps

The interplay between the s-plane and z-plane maths also causes many people difficulties.

OTOH, deep understanding of those topics allows remarkable tools with "baffling" properties to be built.

[wasedadoc]

And again, sin x / x reconstruction is rather crude so that doesn't help.

Why do you keep writing that?

[bdunham7]

That is near what I wrote  But the problem isn't Nyquist limit; the problem is the real world square wave becoming a step function in the digital domain.

Yes, but not limited to digital.  Gibbs ears aren't really a (direct) phenomon of the digital domain or Nyquist issues at all.  You can get Gibbs ears in an entirely analog system and although they usually only appear after the transition instead of symmetrically around it (due to causality), with some effort you could make them symmetric.  The nature of the Gibbs ears depend on the nature of the filter applied to transition; whether that filter is deliberate or a result of sampling and sinx/x interpolation is another issue.

[2N3055]

With all this Nyquist, Shannon and Gibbs talk, I wonder what they would think which entry level 12 bit scope was better choice..

[KungFuJosh]

Is that why it's a previous employer? 😉😉

Nah!

I left them because after >10 years I was in a fur-lined rut, and because the new CEO struck me as someone I didn't believe would be good for the company.

Both judgements turned out to be prescient

I was just teasing, but thanks for the phrase; I've never heard "fur-lined rut" before. 😉

[tggzzz]

With all this Nyquist, Shannon and Gibbs talk, I wonder what they would think which entry level 12 bit scope was better choice..

"Mu", as per the famous Zen koan.

[Mechatrommer]

The interplay between the s-plane and z-plane maths also causes many people difficulties.

are you mathematician? if i give you any transfer function in laplace domain, can you get it back to time domain from first principle? we engineer/hobby grade aint mathematician, and we will not claim to be... although we are trained basic mathematics. we use professor's implementation/code what is practical. i didnt attend ee course but i bet they will not dig deep enough about laplace transform, only enough as engineer with some commonly used transform in a table. thats how we work. if you claim to be good enough, we would like to see some credentials, proofs or references, cheers.

[tggzzz]

Is that why it's a previous employer? 😉😉

Nah!

I left them because after >10 years I was in a fur-lined rut, and because the new CEO struck me as someone I didn't believe would be good for the company.

Both judgements turned out to be prescient

I was just teasing, but thanks for the phrase; I've never heard "fur-lined rut" before. 😉

I enjoyed the tease, of course.

I think I invented the phrase, but it may have been dredged up from the depths of my unconscious.

Most of the things I've invented turn out to be "unwitting re-inventions", dammit.

[Mechatrommer]

And again, sin x / x reconstruction is rather crude so that doesn't help.

Why do you keep writing that?

i guess he has some misunderstanding or using some different definition..

That is near what I wrote  But the problem isn't Nyquist limit; the problem is the real world square wave becoming a step function in the digital domain.

Yes, but not limited to digital.  Gibbs ears aren't really a (direct) phenomon of the digital domain or Nyquist issues at all.  You can get Gibbs ears in an entirely analog system and although they usually only appear after the transition instead of symmetrically around it (due to causality), with some effort you could make them symmetric.  The nature of the Gibbs ears depend on the nature of the filter applied to transition; whether that filter is deliberate or a result of sampling and sinx/x interpolation is another issue.

the gibb's ears referred is not your real life pre-ringing and post-ringing circuit. we are talking about gibb's ears produced by mathematics that was not there in real life in the first place, please dont add unnecessary complexity to ever off-derailed entry level topic. cheers.

[tggzzz]

The interplay between the s-plane and z-plane maths also causes many people difficulties.

are you mathematician? if i give you any transfer function in laplace domain, can you get it back to time domain from first principle? we engineer/hobby grade aint mathematician, and we will not claim to be... although we are trained basic mathematics. we use professor's implementation/code what is practical. i didnt attend ee course but i bet they will not dig deep enough about laplace transform, only enough as engineer with some commonly used transform in a table. thats how we work. if you claim to be good enough, we would like to see some credentials, proofs or references, cheers.

It comes as no surprise to hear you haven't got an electronic (or other?) engineering qualification.

I know I'm third rate... I know a few people[1] who I regard as an order of magnitude more competent than me, and who I listen to very carefully when we disagree. I also know they know other people who they regard as an order of magnitude better than them. Hence I am third rate, at best.

[1] one was descended from two people who each had a different Nobel prize; lovely chap.

[wasedadoc]

That is near what I wrote  But the problem isn't Nyquist limit; the problem is the real world square wave becoming a step function in the digital domain.

Yes, but not limited to digital.  Gibbs ears aren't really a (direct) phenomon of the digital domain or Nyquist issues at all.  You can get Gibbs ears in an entirely analog system and although they usually only appear after the transition instead of symmetrically around it (due to causality), with some effort you could make them symmetric.  The nature of the Gibbs ears depend on the nature of the filter applied to transition; whether that filter is deliberate or a result of sampling and sinx/x interpolation is another issue.

Lots of previous discussion on that at https://www.eevblog.com/forum/testgear/split-from-rigols-new-dho800-oscilloscope-unbox-amp-teardown/.  (And my post with photo of analogue filter response https://www.eevblog.com/forum/testgear/split-from-rigols-new-dho800-oscilloscope-unbox-amp-teardown/msg5140797/#msg5140797 .)

[nctnico]

And again, sin x / x reconstruction is rather crude so that doesn't help.

Why do you keep writing that?

Because it is. As I wrote before you can use much longer filters which result in much better curve fitting so you can reconstruct frequency content very close to Nyquist. The computational power you have available is the limit.

[2N3055]

With all this Nyquist, Shannon and Gibbs talk, I wonder what they would think which entry level 12 bit scope was better choice..

"Mu", as per the famous Zen koan.

Such an useful answer... Thank you.

[tggzzz]

With all this Nyquist, Shannon and Gibbs talk, I wonder what they would think which entry level 12 bit scope was better choice..

"Mu", as per the famous Zen koan.

Such an useful answer... Thank you.

That answer is appropriate in many circumstance, this included. That's why it continues to be relevant after 12 centuries. 

[Martin72]

Somehow this has gotten completely out of hand....
That was the first post:
https://www.eevblog.com/forum/testgear/choosing-between-entry-level-12-bit-dsos/msg5423063/#msg5423063
So the topic was which of the two 12-bit entry-level scopes.
Because there are only these two, either DHO800(900) or SDS800X HD.
I think since hardly anyone who posts here can say anything about it because they don't know both of them, this is so off topic.
If I can't say anything about it, I look for and find something I can say something about.
That's not helpful unless the topic creator has found a solution and the topic itself is over.
Is that the case?
Then the topic can be closed here, we have plenty of basic topics on oscilloscopes.
If not, let's start again and point out the advantages/disadvantages of the two scope series.
I could help, as I had both or still have one of them.

[KungFuJosh]

If not, let's start again and point out the advantages/disadvantages of the two scope series.
I could help, as I had both or still have one of them.

What's your opinion between the two again? 😉

[Mechatrommer]

It comes as no surprise to hear you haven't got an electronic (or other?) engineering qualification.

i know how to read signal integrity from 1 bit scope, and i'm surprised by how many people who dont understand laplace transform... sorry no link provided as i am fourth rank in our junkyard lab crap...

And again, sin x / x reconstruction is rather crude so that doesn't help.

Why do you keep writing that?

Because it is. As I wrote before you can use much longer filters which result in much better curve fitting so you can reconstruct frequency content very close to Nyquist. The computational power you have available is the limit.

you missed our fun with gf providing data to various Lanczos width and its constructability BW ... i'm still struggling to understand the code though as to where and how to set an optimum width to get constructability level at Sr / 2.2...

[shapirus]

If not, let's start again and point out the advantages/disadvantages of the two scope series.
I could help, as I had both or still have one of them.

What's your opinion between the two again? 😉

And why. The why is essential.

[Antonio90]

I was having a lot of fun reading the (off) topic.
But you are right, very few posts try to answer the main question, and very few people has had both.

I can only say the SDS800X-HD is surprising me for the better. I think I like it more than the DSOX1204G. I can't really compare it with anything else though.

Also, to the OP, I don't find it problematic, cumbersome, slow or fiddly to do basic probing around, at all. Unless you just don't have space in your bench and need to mount it on a VESA arm, i'd try it. The Bode plot is great, too, although it needs an external AWG (additional expense).
Why don't you ask specific questions? Maybe Martin and others can answer.

[awakephd]

Somehow this has gotten completely out of hand....
That was the first post:
https://www.eevblog.com/forum/testgear/choosing-between-entry-level-12-bit-dsos/msg5423063/#msg5423063
So the topic was which of the two 12-bit entry-level scopes.
Because there are only these two, either DHO800(900) or SDS800X HD.
I think since hardly anyone who posts here can say anything about it because they don't know both of them, this is so off topic.
If I can't say anything about it, I look for and find something I can say something about.
That's not helpful unless the topic creator has found a solution and the topic itself is over.
Is that the case?
Then the topic can be closed here, we have plenty of basic topics on oscilloscopes.
If not, let's start again and point out the advantages/disadvantages of the two scope series.
I could help, as I had both or still have one of them.

Martin, I posted a question in another thread in response to a statement you made - something to the effect that you would say more (I think about comparing the two?) in another thread, but that your conclusions might be surprising. I can't seem to find where that was; I'll keep looking. But as I recall, my question was whether your conclusions had changed any in the several months since that time.

[wasedadoc]

And again, sin x / x reconstruction is rather crude so that doesn't help.

Why do you keep writing that?

Because it is. As I wrote before you can use much longer filters which result in much better curve fitting so you can reconstruct frequency content very close to Nyquist. The computational power you have available is the limit.

How can you have much longer filters than the infinitely long sin(x)/x ?  A modified version can be as long as you want within the available computational power.  Give us the details of one of your filters and show that it is better than can be achieved with the well-known 'windowed' versions of sin(x)/x allowed the same computational power as yours.

[Mechatrommer]

looking for "credential", i found an example how BW limited scope can miss things out... you can quick read the possibly fabricated story until the Analog Discovery "recommended for beginners" screen capture... how it missed 96MHz oscillation... the fix? (or verification) use SA (probably an expensive one) and a 300MHz scope... so the conclusion? if you want to see oscillation or something wrong in your circuit, get a SA and 3x the suspected oscillation freq DSO... so for example... if oscillation is at 300MHz, get a 1GHz DSO (and SA too and other TEs mumbo jumbo) one, so your measurement is metrologically sound... thats the (spending) more, for less (novel or fancy technique work) just as we typically did or recommended Practical Traps with a One Transistor Audio Amplifier, Solderless Breadboards, and Oscilloscopes unless your name is Jim William, your tin can novel technique will be commonly accepted as crap by professionals (or retired there is)... ymmv.



 

 

what could possibly go wrong with emitter follower on breadboard unless purposely impedance mismatch at long input lead, no?

[tggzzz]

looking for "credential", i found an example how BW limited scope can miss things out... you can quick read the possibly fabricated story until the Analog Discovery "recommended for beginners" screen capture... how it missed 96MHz oscillation... the fix? (or verification) use SA (probably an expensive one) and a 300MHz scope... so the conclusion? if you want to see oscillation or something wrong in your circuit, get a SA and 3x the suspected oscillation freq DSO... so for example... if oscillation is at 300MHz, get a 1GHz DSO (and SA too and other TEs mumbo jumbo) one... thats the (spending) more, for less (novel or fancy technique work) just as we typically did or recommended Practical Traps with a One Transistor Audio Amplifier, Solderless Breadboards, and Oscilloscopes unless your name is Jim William, your tin can novel technique will be commonly accepted as crap by professionals (or retired there is)... ymmv.
...
what could possibly go wrong with emitter follower on breadboard unless purposely impedance mismatch at long input lead, no?
...

A few points about that rant:

• that is a complete misrepresentation of the article, in too many ways to bother to describe. One possible reason is that mechatrommer has completely misunderstood the article, but I will leave other people to infer reasons
• the few technical points are wrong:
  
  • DSO-vs-analogue scopes is irrelevant; it is a red herring mechatrommer has invented
  • there is no impedance mismatch
  • £25 analysers could have been used, but weren't since the £350 SA was easier to use
• what is the benefit of make it difficult to read?

By all means read the article and see whether mechatrommer's points have any validity whatsoever!

[Mechatrommer]

typical strawman fallacy, taking my words out of context. didnt i use question mark? or words possibly/probably?  where did i mentioned "analogue scope"? or even comparing it with DSO? i know the article is about why you should not use breadboard... which to some of us is pretty invalid anyway, just because of one FU scenario. but my highlight is on BW limited scope demonstrated in the article, not really anything to do with the main topic specifically.