Building my own scope

[David Hess]

your sampling scope would not even catch the runt pulse, and your 1GHz bandwidth 40MS/s scope you love so much would not even notice half of the data... so what to use ? Mr. Tggzzz with a sharp pencil and a paper roll will do the trick i guess... while everyone else is happily using the 1GS/s scopes which are in fact not usable because the almighty omnipresent and omnipotent Mr. Tggzzz said so 

Both a sampling oscilloscope and a 40 MS/s digital storage oscilloscope would capture a runt pulse eventually, but neither could decode the data.  Actually that is not quite accurate; some older DSOs supported synchronous sampling so a 40 MS/s DSO could decode 40 MHz SPI if provided with the clock.

But it does come down to using the right tool for the job.  The subject of sampling oscilloscopes came up in connection with what could be done with Arduino class hardware.  Such would be useful to very accurately characterize the communications channel of an SPI interface, but not for decoding it.  Most often I use my sampling oscilloscope to verify the transient response of other oscilloscopes.

[rob77]

your sampling scope would not even catch the runt pulse, and your 1GHz bandwidth 40MS/s scope you love so much would not even notice half of the data... so what to use ? Mr. Tggzzz with a sharp pencil and a paper roll will do the trick i guess... while everyone else is happily using the 1GS/s scopes which are in fact not usable because the almighty omnipresent and omnipotent Mr. Tggzzz said so 

Both a sampling oscilloscope and a 40 MS/s digital storage oscilloscope would capture a runt pulse eventually, but neither could decode the data.  Actually that is not quite accurate; some older DSOs supported synchronous sampling so a 40 MS/s DSO could decode 40 MHz SPI if provided with the clock.

But it does come down to using the right tool for the job.  The subject of sampling oscilloscopes came up in connection with what could be done with Arduino class hardware.  Such would be useful to very accurately characterize the communications channel of an SPI interface, but not for decoding it.  Most often I use my sampling oscilloscope to verify the transient response of other oscilloscopes.

synchronous sampling ... isn't that called logic analyzer ? 

but anyways... sampling scope or a 40MS/s scope (be it whatever GHz bandwidth) is not the right tool in that case, we can agree on that.

and if we steer back to the topic of the thread , then neither of those is a good candidate for a DIY build as as learning experience.

[Mechatrommer]

The subject of sampling oscilloscopes came up in connection with what could be done with Arduino class hardware.

no, it came up much earlier when someone broke the box proposing that we dont need nyquist minimum sampling criterion (for real-time sampling) to build an dso. since there is no objection from OP so far, so lets discuss and expand it with current reinvention into tiny and cheaper unit linked earlier. known guy using sampling (equivalent) scope here is Leo Bodnar to verify his 40ps rise time pulser device which i bought one.

[tggzzz]

The subject of sampling oscilloscopes came up in connection with what could be done with Arduino class hardware.

no, it came up much earlier when someone broke the box proposing that we dont need nyquist minimum sampling criterion (for real-time sampling) to build an dso.

I haven't seen any claim to the effect that Nyquist can be circumvented.

I have seen some people fail to understand the definition of bandwidth of a signal in the context of the Nyquist criterion. And then either go on to either deny certain types of equipment work successfully, or change their statement without acknowledging the change.

[Mechatrommer]

because generally when a person want to build a diy dso implicitly means the real-time one. a person will come with building a "sampling" scope in the topic when he knows what he is talking about. and nobody will recommend newbies asking about buying a new scope to buy a "sampling" scope.

[tggzzz]

because generally when a person want to build a diy dso implicitly means the real-time one. a person will come with building a "sampling" scope in the topic when he knows what he is talking about. and nobody will recommend newbies asking about buying a new scope to buy a "sampling" scope.

Er, newbies are routinely recommended to buy a sampling scope.

You, at least I think it was you, previously ignored the point that all modern scopes (e.g. the Rigol DS1054) tale samples of the input signal. You can even look and see the individual samples on the display, presuming you can navigate the menu system to find the right config parameter.

[rob77]

because generally when a person want to build a diy dso implicitly means the real-time one. a person will come with building a "sampling" scope in the topic when he knows what he is talking about. and nobody will recommend newbies asking about buying a new scope to buy a "sampling" scope.

Er, newbies are routinely recommended to buy a sampling scope.

You, at least I think it was you, previously ignored the point that all modern scopes (e.g. the Rigol DS1054) tale samples of the input signal. You can even look and see the individual samples on the display, presuming you can navigate the menu system to find the right config parameter.

just stop it , seriously... twisting everyone's words and spreading bullcrap....

now you suggest that every DSO is a sampling scope because they take samples ? 

it's you who spreads half truth statements and derails the discussion in this thread.

it was you who stated sample rate and bandwidth are independent without context , which is half true

it was you who stated that cheap sdr dongles are sampling 2GHz... that's a misinformation... those have have a tuner outputting a much lower IF and that IF is sampled...

it was you who suggested to delete a FPGA and do discrete control for ADC and let the computer take data on it's convenience...when i asked you how exactly you would implement your "idea".. you just mumbled some generic stuff and steered away...

you might be a knowledgeable expert but that doesn't mean you can't be an trolling a-hole at the same time 

[tggzzz]

because generally when a person want to build a diy dso implicitly means the real-time one. a person will come with building a "sampling" scope in the topic when he knows what he is talking about. and nobody will recommend newbies asking about buying a new scope to buy a "sampling" scope.

Er, newbies are routinely recommended to buy a sampling scope.

You, at least I think it was you, previously ignored the point that all modern scopes (e.g. the Rigol DS1054) tale samples of the input signal. You can even look and see the individual samples on the display, presuming you can navigate the menu system to find the right config parameter.

just stop it , seriously... twisting everyone's words and spreading bullcrap....

now you suggest that every DSO is a sampling scope because they take samples ? 

You do realise how stupid that sentence is, don't you?

Could you please point us towards a DSO that doesn't sample the inputs?

it's you who spreads half truth statements and derails the discussion in this thread.

it was you who stated sample rate and bandwidth are independent without context , which is half true

They are independent, in theory and in practical instruments and products that you can go out and buy. I've given examples.

it was you who stated that cheap sdr dongles are sampling 2GHz... that's a misinformation... those have have a tuner outputting a much lower IF and that IF is sampled...

The "tuner" (as you call it) is a mixer and that is doing the sampling. That's precisely how the Tayloe/PolyPhase mixer in SDR dongles work!.

The output of the Tayloe/PolyPhase mixer is then sampled (and digitised) at a much lower rate, usually a few MHz.

You appear to be confusing the sampling in the mixer with the digitisation sampling rate.

FFI you can enlighten yourself with the info in this thread: https://eevblog.com/forum/rf-microwave/polyphase-or-n-path-mixer/ Be warned that it isn't easy reading, and many people find it is counter-intuitive. Alternatively the "ARRL Handbook for radio communications 2014" has a brief description and a practical circuit in section 10. I imagine it is also in other ARRL Handbooks.

it was you who suggested to delete a FPGA and do discrete control for ADC and let the computer take data on it's convenience...when i asked you how exactly you would implement your "idea".. you just mumbled some generic stuff and steered away...

Well, I certainly haven't got time to (poorly) explain the basic theory here; a textbook will do a much better job.

As for designing an implementation, I doubt the OP wants that. If he did then I would discuss the commercial terms.

you might be a knowledgeable expert but that doesn't mean you can't be an trolling a-hole at the same time 

Can I suggest your being a little less sure that you know and understand the theory and practice.

[Mechatrommer]

just stop it , seriously... twisting everyone's words and spreading bullcrap....

i know the problem now. he is a literalist and want to stick with terminology when it was first appeared some 80 years ago. let the old man give his point of view, we only need to know his dictionary to come to an agreement.

[David Hess]

synchronous sampling ... isn't that called logic analyzer ? 

Logic analyzers can operate synchronously or asynchronously.

because generally when a person want to build a diy dso implicitly means the real-time one. a person will come with building a "sampling" scope in the topic when he knows what he is talking about. and nobody will recommend newbies asking about buying a new scope to buy a "sampling" scope.

Er, newbies are routinely recommended to buy a sampling scope.

You, at least I think it was you, previously ignored the point that all modern scopes (e.g. the Rigol DS1054) tale samples of the input signal. You can even look and see the individual samples on the display, presuming you can navigate the menu system to find the right config parameter.

DSO stands for digital storage oscilloscope distinguishing it from digital sampling oscilloscope.  The later term was almost never used, but does refer to a specific type of instrument which is not a DSO.

Nothing requires a digital storage oscilloscope, or any ADC or digitizer, to use sampling, and when they do use sampling, it is *not* the same type of sampling that a sampling oscilloscope uses.  Datasheets tend to no longer distinguish them, but ADCs are available in sampling and non-sampling varieties.  Going back to the beginning of DSO evolution, the Tektronix 468, 2230, and 2232 all use non-sampling ADCs, however the 2230 included a discrete sampling unit before its 20 MHz flash ADC, and the 2232 included integrated samplers before each of its 100 MS/s flash ADCs.  None of these worked the same way as the sampler does inside of a sampling oscilloscope. These were digital storage oscilloscopes.

The sampler in a sampling oscilloscope operates with a strobe which is *shorter*, often a lot shorter, than the acquisition time needed to charge the storage element.  This means that either the storage element must be reset between samples or the output is based on both the old and new value, which is where sampling gain come into play.  By requiring only a fraction of the charge to be transferred, the "sampler" has gain which can be applied to increase its bandwidth.  (1) (2)

The distinguishing feature of a sampling oscilloscope is that sampling occurs first in the signal chain.  Gain is applied after sampling in the analog domain.  This means that bandwidth is determined entirely by the the sampling strobe width, which also produces a non-linear frequency response, so the 0.35 rule does not apply.  One advantage of this besides predictable frequency response, is instant overload recovery.  Overload never gets past the sampler so nothing needs to recover.

(1) Nobody is saying how modern integrated ADCs do their sampling for high bandwidth.  An internal 50 ohm termination results in a 25 ohm source impedance to charge the sampling storage element which sure doesn't seem like enough for the bandwidth they achieve.

(2) I am describing the sampler needed for random sampling which is the more general case.  The sampler in a sequential sampling oscilloscope can operate with a gain of 1 by repetitively sampling the same point of a waveform.

[David Hess]

Could you please point us towards a DSO that doesn't sample the inputs?

The Tektronix 468 does not use sampling.  Not all high speed ADCs use sampling.

[pcprogrammer]

So the distinction basically is that sampling means charging a capacitor to the analog voltage at the input and not the conversion to digital to make a digital sample.

So some of the scopes discussed here do not turn it into a digital value, and just use the analog voltage stored in capacitors to show things on the display. Where as modern digital scopes hold the analog sample while the analog to digital converter turns it into a digital value which is referred to as a sample.

Hence the assumption that a DSO uses sampling.

Very confusing unless you know the history of the technology. 

And now the question is how this is done in very high speed ADC's because the charging of such a capacitor for taking the sample is possibly to slow to do this. For this I don't know the answer.

Is this a correct recap?

[tautech]

So the distinction basically is that sampling means charging a capacitor to the analog voltage at the input and not the conversion to digital to make a digital sample.

So some of the scopes discussed here do not turn it into a digital value, and just use the analog voltage stored in capacitors to show things on the display. Where as modern digital scopes hold the analog sample while the analog to digital converter turns it into a digital value which is referred to as a sample.

Hence the assumption that a DSO uses sampling.

Very confusing unless you know the history of the technology. 

And now the question is how this is done in very high speed ADC's because the charging of such a capacitor for taking the sample is possibly to slow to do this. For this I don't know the answer.

Is this a correct recap?

Of course they do. Why else would their sampling capability be listed in every datasheet in GSa/s ?

[pcprogrammer]

Of course they do. Why else would their sampling capability be listed in every datasheet in GSa/s ?

Read David his post.

The way I read it, is that according to him sampling is charging a capacitor with an analog value, not the turning it into a digital value. It is semantics.

[tautech]

Of course they do. Why else would their sampling capability be listed in every datasheet in GSa/s ?

Read David his post.

The way I read it, is that according to him sampling is charging a capacitor with an analog value, not the turning it into a digital value. It is semantics.

I did also Mech's, rob77's and tggzzz's earlier all with some amusement.

But you just can't get away from facts, a DSO samples.....fast !

[pcprogrammer]

But how? Like David wrote the charging of the capacitor might be to slow when it comes to GHz signals.

Looking at analog to digital conversion it depends on the used technology if there needs to be a sample and hold circuit before the analog to digital converter. With successive-approximation the input needs to be stable during the conversion. There are several types and the question is which does not need a sample and hold circuit. https://dewesoft.com/daq/types-of-adc-converters

[tautech]

But how?

Do you think the sampling dots in a DSO in Dot mode are just invented ?

[pcprogrammer]

But how?

Do you think the sampling dots in a DSO in Dot mode are just invented ?

That is not the point. Yes a DSO uses digital values that represent the signal. And we call these samples. But what part of semantics don't you get?

David states that sampling is the act of capturing an analog value in a capacitor and not the converting it into a digital value.

And what he, and me too, wonders about how this works for a high speed ADC. Does it use a sample and hold circuit or not.

[gf]

And what he, and me too, wonders about how this works for a high speed ADC. Does it use a sample and hold circuit or not.

Virtualy any ADC does. Continuous-time sigma-delta ADCs are the only exception i'm aware of.

[tggzzz]

synchronous sampling ... isn't that called logic analyzer ? 

Logic analyzers can operate synchronously or asynchronously.

because generally when a person want to build a diy dso implicitly means the real-time one. a person will come with building a "sampling" scope in the topic when he knows what he is talking about. and nobody will recommend newbies asking about buying a new scope to buy a "sampling" scope.

Er, newbies are routinely recommended to buy a sampling scope.

You, at least I think it was you, previously ignored the point that all modern scopes (e.g. the Rigol DS1054) tale samples of the input signal. You can even look and see the individual samples on the display, presuming you can navigate the menu system to find the right config parameter.

DSO stands for digital storage oscilloscope distinguishing it from digital sampling oscilloscope.  The later term was almost never used, but does refer to a specific type of instrument which is not a DSO.

Nothing requires a digital storage oscilloscope, or any ADC or digitizer, to use sampling, and when they do use sampling, it is *not* the same type of sampling that a sampling oscilloscope uses.  Datasheets tend to no longer distinguish them, but ADCs are available in sampling and non-sampling varieties.  Going back to the beginning of DSO evolution, the Tektronix 468, 2230, and 2232 all use non-sampling ADCs, however the 2230 included a discrete sampling unit before its 20 MHz flash ADC, and the 2232 included integrated samplers before each of its 100 MS/s flash ADCs.  None of these worked the same way as the sampler does inside of a sampling oscilloscope. These were digital storage oscilloscopes.

The sampler in a sampling oscilloscope operates with a strobe which is *shorter*, often a lot shorter, than the acquisition time needed to charge the storage element.  This means that either the storage element must be reset between samples or the output is based on both the old and new value, which is where sampling gain come into play.  By requiring only a fraction of the charge to be transferred, the "sampler" has gain which can be applied to increase its bandwidth.  (1) (2)

The distinguishing feature of a sampling oscilloscope is that sampling occurs first in the signal chain.  Gain is applied after sampling in the analog domain.  This means that bandwidth is determined entirely by the the sampling strobe width, which also produces a non-linear frequency response, so the 0.35 rule does not apply.  One advantage of this besides predictable frequency response, is instant overload recovery.  Overload never gets past the sampler so nothing needs to recover.

(1) Nobody is saying how modern integrated ADCs do their sampling for high bandwidth.  An internal 50 ohm termination results in a 25 ohm source impedance to charge the sampling storage element which sure doesn't seem like enough for the bandwidth they achieve.

(2) I am describing the sampler needed for random sampling which is the more general case.  The sampler in a sequential sampling oscilloscope can operate with a gain of 1 by repetitively sampling the same point of a waveform.

The key phrase there is "it is *not* the same type of sampling".

I entirely agree that - to put it overly simply - one type of sampling captures the average voltage during a period, and another captures the average voltage at an instant in a period. The two types' very different aperture times lead to differing characteristics, advantages and disadvantages.

Nonetheless they do both provide representations of the input voltage at discrete time intervals - and that is the key distinguishing feature of sampling vs analogue processing.

Classic examples of radically different aperture times found in sampling scopes are a diode ring (aperture time << sampling interval) and capacitor in a CCD (aperture time = sampling interval).

[tggzzz]

Of course they do. Why else would their sampling capability be listed in every datasheet in GSa/s ?

Read David his post.

The way I read it, is that according to him sampling is charging a capacitor with an analog value, not the turning it into a digital value. It is semantics.

I did also Mech's, rob77's and tggzzz's earlier all with some amusement.

But you just can't get away from facts, a DSO samples.....fast !

Not the old CCD-based Tek scopes! If you had a 1s sweep and 1000 capacitors in the CCD, then each capacitor contained the average voltage in a 1ms interval. A 1MHz oscillation would be invisible - and that was a problem.

More modern ADCs have a constant sampling rate even at slow sweep speeds, and that is the reason they can have "peak detect" modes which can show the presence of a 1MHz oscillation as a "noise" band.

[tggzzz]

So the distinction basically is that sampling means charging a capacitor to the analog voltage at the input and not the conversion to digital to make a digital sample.

Sampling means representing a signal at discrete intervals of time.

Some scopes simply display those discrete representations.

Some scopes digitise the representations, then store and digitally process them before displaying them.

[pcprogrammer]

Do you think the sampling dots in a DSO in Dot mode are just invented ?

Actually the original FNIRSI-1013D does invent (calculates) them when the input is above 44MHz 

[tggzzz]

And what he, and me too, wonders about how this works for a high speed ADC. Does it use a sample and hold circuit or not.

Virtualy any ADC does. Continuous-time sigma-delta ADCs are the only exception i'm aware of.

And they also produce samples, of course.

I have an ADC that produces one sample every 50 seconds (a Solartron 7081). If the signal isn't constant within a sampling period then the output is the average of the input over that period.

[pcprogrammer]

Sampling means representing a signal at discrete intervals of time.

In this context yes. Sampling has many meanings. You can sample food or drinks, or take a sample of something you send to the printers to see if it matches the quality you require.