Desktop DSO for hardware and firmware development of MCU systems

[tautech]

And now setup with a LAN connection to PC at the bench a webserver capture and remote screenshot of how the Serial Test parameter values are adjusted.

[tautech]

Effective ADC bit (ENOB) is about 1.5 to 2 less than the specification?  8 bit ADC has 6 bit real accuracy and resolve 64 steps of digital SPI signal.  Would that be enough for task below?

How much sampling rate do I need to see the signal, without risk of missing some short time glitch?  SDS2104 is 1GSa/s when all 4 ch activated, right?  For 20MHz SPI clock, 50 dots per cycle. Enough?

On speed alone, Rigol MSO5074 is exceptional high 8GSa/s for 700 UK pounds.

Divided down by the # of channels in use, as is the memory depth.
I don't remember by how much but a search should reveal the facts.

OTOH SDS2000X Plus uses 2 ADC's, each 2 GSa/s and with 200 Mpts memory each, therefore 1 GSa/s and 100 Mpts memory depth with all channels active.

[Wilson__]

oh!  14 pages long.

[Wilson__]

It will be 100 dots per cycle in 2 ch mode but need alternate trigger than directly decode the 4 actual signal wires of SPI.

[Wilson__]

Many thanks for the info.  That is very nice and neat. 

World moved on.  Would take lot of human  to move cursors (since time of CRT analogue scope with digital marker/cursor) and copy the timing number. now is auto

[tautech]

Further decode and zoom examples.....
All I have with SPI signal is STB-3

[Wilson__]

If you have the budget to go higher than the SDS2000X Plus, then compare it with the SDS2000X HD. Either one would be a great scope for your needs

Many thanks for info.  My use is design verification.  Use DSO to inspect signal lines to confirm that there is no abnormal glitch nor invalid signal. Are below correct?

1. The MCU and external chips works at 10MHz SPI clock.  Likely these silicon are tens of nm node and will not response to glitch that is many times narrower than the normal 10MHz wanted-signal.

2. SDS2000X Plus is 1 or 2GSa/s in 4 or 2 channel mode.  So, I got 100 to 200 dots for the 10MHz signal.  Spec. also says 1ns peak detection.  Presumably, refers to best case signal at full swing voltage (3.3 volts).  Presumably, the scope will detect lower-voltage glitch at, says, 1 volt, of a few ns long. 

3. SDS2000X Plus will do the job, right?  If the scope does not see any glitch, the glitch energy, (voltage multipy time) should be too weak to cause the chip to response, right?

[2N3055]

If you have the budget to go higher than the SDS2000X Plus, then compare it with the SDS2000X HD. Either one would be a great scope for your needs

Many thanks for info.  My use is design verification.  Use DSO to inspect signal lines to confirm that there is no abnormal glitch nor invalid signal. Are below correct?

1. The MCU and external chips works at 10MHz SPI clock.  Likely these silicon are tens of nm node and will not response to glitch that is many times narrower than the normal 10MHz wanted-signal.

2. SDS2000X Plus is 1 or 2GSa/s in 4 or 2 channel mode.  So, I got 100 to 200 dots for the 10MHz signal.  Spec. also says 1ns peak detection.  Presumably, refers to best case signal at full swing voltage (3.3 volts).  Presumably, the scope will detect lower-voltage glitch at, says, 1 volt, of a few ns long. 

3. SDS2000X Plus will do the job, right?  If the scope does not see any glitch, the glitch energy, (voltage multipy time) should be too weak to cause the chip to response, right?

You are doing too much of presumptions.

1ns Peak detection means it is in peak detect mode.
In which it will detect a 1ns pulse.

How wide pulse it will detect in normal sampling mode depends on rise/fall times of the pulse.
If pulse has sub ns edges, it will pretty much detect a 3ns pulse with full amplitude.
To cut the story short, SDS2000X+ ( with full BW) is going to be more than fast enough to look into SPI. 

But we are suddenly talk about low digit ns. To deal with that you need 1GHz + scope ...
And some good probing.

Even to achieve full BW of SDS2000X+ your biggest problem will be probing and will be most limiting factor.

[Wilson__]

To cut the story short, SDS2000X+ ( with full BW) is going to be more than fast enough to look into SPI.

Even to achieve full BW of SDS2000X+ your biggest problem will be probing and will be most limiting factor.

Sorry, coming from thru-hole age.  How to probe QFN chip SPI signals?  'Test version' PCB with test point?

Googled says, https://www.edn.com/oscilloscope-probe-accessories-its-the-little-things-that-matter/

[tggzzz]

1. The MCU and external chips works at 10MHz SPI clock.  Likely these silicon are tens of nm node and will not response to glitch that is many times narrower than the normal 10MHz wanted-signal.

Don't bet on it. Or rather it depends on "how many".

If it is important, measure it, taking account of PSU voltage, temperature, phase of the moon, chip manufacturer, IC datestamp etc.

In digital circuits the frequency is unimportant, but the timing is important. If you violate, say the t_hold, then it doesn't matter if the signal is 1MHz or 1Hz (and unlike t_setup it won't be ameliorated by reducing the clock frequency)

[nctnico]

To cut the story short, SDS2000X+ ( with full BW) is going to be more than fast enough to look into SPI.

Even to achieve full BW of SDS2000X+ your biggest problem will be probing and will be most limiting factor.

Sorry, coming from thru-hole age.  How to probe QFN chip SPI signals?  'Test version' PCB with test point?

Solder thin (enamel) wires to the best possible location. Typically you'll want to use series resistors on SPI interfaces to dampen reflections so these are good points. Otherwise you'll need to solder really thin (enamel) wires to QFN pads and bring them to a 2.54mm header or something like that. Not impossible to do but still a nuisance.

[2N3055]

To cut the story short, SDS2000X+ ( with full BW) is going to be more than fast enough to look into SPI.

Even to achieve full BW of SDS2000X+ your biggest problem will be probing and will be most limiting factor.

Sorry, coming from thru-hole age.  How to probe QFN chip SPI signals?  'Test version' PCB with test point?

Googled says, https://www.edn.com/oscilloscope-probe-accessories-its-the-little-things-that-matter/

One of the often forgotten things is "design for test". Provide test points for critical signals. Not only on prototype.
If you change layout from prototype to production, how do you know you didn't influence things?
You can solder in short extensions to probe tip (making sure you have good ground), use spring ground or similar.

Really high speed active probes are mostly solder in..

For some probing, you can use 10x LoZ probe together with 50Ω inputs..

Trick is to minimise capacity (to minimise circuit load) and inductance as seen by probe.. And to stay with high enough DC resistance to minimize DC load. Art of compromises.

Ideally you could use an active probe, but we if are talking about budget options, maybe something homemade or by one of the members here that occasionally spin some prototypes..

[tautech]

If you have the budget to go higher than the SDS2000X Plus, then compare it with the SDS2000X HD. Either one would be a great scope for your needs

Many thanks for info.  My use is design verification.  Use DSO to inspect signal lines to confirm that there is no abnormal glitch nor invalid signal. Are below correct?

1. The MCU and external chips works at 10MHz SPI clock.  Likely these silicon are tens of nm node and will not response to glitch that is many times narrower than the normal 10MHz wanted-signal.

2. SDS2000X Plus is 1 or 2GSa/s in 4 or 2 channel mode.  So, I got 100 to 200 dots for the 10MHz signal.  Spec. also says 1ns peak detection.  Presumably, refers to best case signal at full swing voltage (3.3 volts).  Presumably, the scope will detect lower-voltage glitch at, says, 1 volt, of a few ns long. 

3. SDS2000X Plus will do the job, right?  If the scope does not see any glitch, the glitch energy, (voltage multipy time) should be too weak to cause the chip to response, right?

The modern DSO is pretty powerful at finding stuff you might not even think is present.....it's all about using the features available to see that you might have a problem then applying the toolset to capture them.



This ^^^ screenshot is a good example where some Persistence shows it's present then we can narrow in and seen if it's a one off or repetitive.
In this post I did a similar exercise with the older SDS1104X-E which gives some idea of using a few of the scopes features:
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg1370717/#msg1370717

[Wilson__]

If you have the budget to go higher than the SDS2000X Plus, then compare it with the SDS2000X HD. Either one would be a great scope for your needs

Many thanks for info.  My use is design verification.  Use DSO to inspect signal lines to confirm that there is no abnormal glitch nor invalid signal. Are below correct?

1. The MCU and external chips works at 10MHz SPI clock.  Likely these silicon are tens of nm node and will not response to glitch that is many times narrower than the normal 10MHz wanted-signal.

2. SDS2000X Plus is 1 or 2GSa/s in 4 or 2 channel mode.  So, I got 100 to 200 dots for the 10MHz signal.  Spec. also says 1ns peak detection.  Presumably, refers to best case signal at full swing voltage (3.3 volts).  Presumably, the scope will detect lower-voltage glitch at, says, 1 volt, of a few ns long. 

3. SDS2000X Plus will do the job, right?  If the scope does not see any glitch, the glitch energy, (voltage multipy time) should be too weak to cause the chip to response, right?

The modern DSO is pretty powerful at finding stuff you might not even think is present.....it's all about using the features available to see that you might have a problem then applying the toolset to capture them.



This ^^^ screenshot is a good example where some Persistence shows it's present then we can narrow in and seen if it's a one off or repetitive.
In this post I did a similar exercise with the older SDS1104X-E which gives some idea of using a few of the scopes features:
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg1370717/#msg1370717

Comparing SDS1104X-E and SDS2104X-Plus.  Does the cheaper unit have same toos/function (trigger, search, measure) for the purpose of design verification, to hunt for abnormal signals in 10MHz SPI signal between MCU and external chips?

Is there more differences than what I found so far from scanning manual:

500MSa/s   vs   1000MSa/s in 4 channels mode (MISO, MOSI, Clock, nChipSelect), 500M is 50 samples per 10MHz signal pulse.  Enough to see abnormal glitch, runt, overshoot, undershoot???

7 Mpts/CH  vs  100Mpts, 7M captures 140,000 pulses at 50 samples per pulse. 

No histogram, still has StdDev to qualtify signal jitter.

7 inches and non-touch screen.   Needs a few more human seconds to use knobs to active a function, right?

[tautech]

If you have the budget to go higher than the SDS2000X Plus, then compare it with the SDS2000X HD. Either one would be a great scope for your needs

Many thanks for info.  My use is design verification.  Use DSO to inspect signal lines to confirm that there is no abnormal glitch nor invalid signal. Are below correct?

1. The MCU and external chips works at 10MHz SPI clock.  Likely these silicon are tens of nm node and will not response to glitch that is many times narrower than the normal 10MHz wanted-signal.

2. SDS2000X Plus is 1 or 2GSa/s in 4 or 2 channel mode.  So, I got 100 to 200 dots for the 10MHz signal.  Spec. also says 1ns peak detection.  Presumably, refers to best case signal at full swing voltage (3.3 volts).  Presumably, the scope will detect lower-voltage glitch at, says, 1 volt, of a few ns long. 

3. SDS2000X Plus will do the job, right?  If the scope does not see any glitch, the glitch energy, (voltage multipy time) should be too weak to cause the chip to response, right?

The modern DSO is pretty powerful at finding stuff you might not even think is present.....it's all about using the features available to see that you might have a problem then applying the toolset to capture them.



This ^^^ screenshot is a good example where some Persistence shows it's present then we can narrow in and seen if it's a one off or repetitive.
In this post I did a similar exercise with the older SDS1104X-E which gives some idea of using a few of the scopes features:
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg1370717/#msg1370717

Comparing SDS1104X-E and SDS2104X-Plus.  Does the cheaper unit have same toos/function (trigger, search, measure) for the purpose of design verification, to hunt for abnormal signals in 10MHz SPI signal between MCU and external chips?

Is there more differences than what I found so far from scanning manual:

500MSa/s   vs   1000MSa/s in 4 channels mode (MISO, MOSI, Clock, nChipSelect), 500M is 50 samples per 10MHz signal pulse.  Enough to see abnormal glitch, runt, overshoot, undershoot???

7 Mpts/CH  vs  100Mpts, 7M captures 140,000 pulses at 50 samples per pulse. 

No histogram, still has StdDev to qualtify signal jitter.

7 inches and non-touch screen.   Needs a few more human seconds to use knobs to active a function, right?

TBH, today if wanting a lower cost solution and happy with a 7" display, SDS814X HD is where it's at.

[KungFuJosh]

SDS1104X-E and SDS2104X-Plus

My first scope was a Rigol piece of trash, that spawned my general hate for Rigol.

My second scope was the SDS1104X-E. It was a great scope, and generally met my needs at the time. But there is simply no comparison when jumping up to the SDS2104X Plus (my 3rd scope). The screen quality, general speed, additional features/triggers etc.

A very important difference for me is the far superior web console on the 2000 level scopes. I imagine you will likely be using the web console once your tests are setup. For that alone I wouldn't consider a lesser scope.

I sold my SDS2104X Plus because I had an opportunity to get an SDS2104X HD at a great price. I feel like the level between these two scopes is also a leap forward, but I wouldn't be sad at all if I was still using the SDS2104X+.

All that said, if budget is your biggest concern, get an SDS800X HD scope. If performance is your concern, start at the 2000 level, whether it's the Plus or HD, you'll be happy.

[Wilson__]

If you have the budget to go higher than the SDS2000X Plus, then compare it with the SDS2000X HD. Either one would be a great scope for your needs

Many thanks for info.  My use is design verification.  Use DSO to inspect signal lines to confirm that there is no abnormal glitch nor invalid signal. Are below correct?

1. The MCU and external chips works at 10MHz SPI clock.  Likely these silicon are tens of nm node and will not response to glitch that is many times narrower than the normal 10MHz wanted-signal.

2. SDS2000X Plus is 1 or 2GSa/s in 4 or 2 channel mode.  So, I got 100 to 200 dots for the 10MHz signal.  Spec. also says 1ns peak detection.  Presumably, refers to best case signal at full swing voltage (3.3 volts).  Presumably, the scope will detect lower-voltage glitch at, says, 1 volt, of a few ns long. 

3. SDS2000X Plus will do the job, right?  If the scope does not see any glitch, the glitch energy, (voltage multipy time) should be too weak to cause the chip to response, right?

The modern DSO is pretty powerful at finding stuff you might not even think is present.....it's all about using the features available to see that you might have a problem then applying the toolset to capture them.



This ^^^ screenshot is a good example where some Persistence shows it's present then we can narrow in and seen if it's a one off or repetitive.
In this post I did a similar exercise with the older SDS1104X-E which gives some idea of using a few of the scopes features:
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg1370717/#msg1370717

Comparing SDS1104X-E and SDS2104X-Plus.  Does the cheaper unit have same toos/function (trigger, search, measure) for the purpose of design verification, to hunt for abnormal signals in 10MHz SPI signal between MCU and external chips?

Is there more differences than what I found so far from scanning manual:

500MSa/s   vs   1000MSa/s in 4 channels mode (MISO, MOSI, Clock, nChipSelect), 500M is 50 samples per 10MHz signal pulse.  Enough to see abnormal glitch, runt, overshoot, undershoot???

7 Mpts/CH  vs  100Mpts, 7M captures 140,000 pulses at 50 samples per pulse. 

No histogram, still has StdDev to qualtify signal jitter.

7 inches and non-touch screen.   Needs a few more human seconds to use knobs to active a function, right?

TBH, today if wanting a lower cost solution and happy with a 7" display, SDS814X HD is where it's at.

Many thanks.  From SDS814X HD user manual, it has a new trigger mode, among manys, that seem to fit my speific SPI signal hunting.

Being new to modern MSO, please kindly advise if below is correct understanding:

1. Is trigger setup ">, <, in-range and out-of-range" for a numberic value for the setup time and holdtime respectively?

2. If I set trigger "setup time < data sheet value", and let the McU run overnight.  if the scope did not trigger, there was no glitch nor rare abnormal events leading to non-compliance of "setup time should be longer than datasheet value".

3. If I capture a random frame of data to full memory of the scope (tens of ms of MCU real run time at 1GSa/s), I can use SEARCH with "setup_time of >1ns", I should get 100% event hit. 

Slowly increase, to says, 100ns, as data sheet, if still got 100% event hit, this VERIFY that the data frame meet the data sheet value.

Further increase time to get 50% event hit, this will be the medium value of the circuit's setup time, right?

[Fungus]

2. If I set trigger "setup time < data sheet value", and let the McU run overnight.  if the scope did not trigger, there was no glitch nor rare abnormal events leading to non-compliance of "setup time should be longer than datasheet value".

Even better: Modern DSOs can configure segmented memory and record all the events separately, along with a count of how many there were.

(assuming there's not too many of them to fit in the 'scopes memory ... )

[tatel]

I you want a MSO option that works fine, you should have a look at this user report linked below. I don't think you'll be happy with SDS1104X-E for that, you'd better get at least Rigol MSO5000 or Siglent SDS2000, or perhaps the new siglents 800 HD, and even then, I would ask for some real user-experience-based report about the specific model you're looking at, before pulling the trigger

https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg3528422/#msg3528422

[Fungus]

or perhaps the new siglents 800 HD

The Siglent 800DH logic option is a weird USB thing. I'd take a good look at reviews before deciding on that.

[KungFuJosh]

The Siglent 800HD logic option is a weird USB thing. I'd take a good look at reviews before deciding on that.

It's not that weird being an external analyzer/adapter, but the consensus is that the 2000 level stuff is obviously better. The SPL2016 logic probe used with the nicer scopes with built-in analyzers is also nicer...but my generic SPL2016 copy works well enough for my needs for ~10% of the price. 😉

[tautech]

If you have the budget to go higher than the SDS2000X Plus, then compare it with the SDS2000X HD. Either one would be a great scope for your needs

Many thanks for info.  My use is design verification.  Use DSO to inspect signal lines to confirm that there is no abnormal glitch nor invalid signal. Are below correct?

1. The MCU and external chips works at 10MHz SPI clock.  Likely these silicon are tens of nm node and will not response to glitch that is many times narrower than the normal 10MHz wanted-signal.

2. SDS2000X Plus is 1 or 2GSa/s in 4 or 2 channel mode.  So, I got 100 to 200 dots for the 10MHz signal.  Spec. also says 1ns peak detection.  Presumably, refers to best case signal at full swing voltage (3.3 volts).  Presumably, the scope will detect lower-voltage glitch at, says, 1 volt, of a few ns long. 

3. SDS2000X Plus will do the job, right?  If the scope does not see any glitch, the glitch energy, (voltage multipy time) should be too weak to cause the chip to response, right?

The modern DSO is pretty powerful at finding stuff you might not even think is present.....it's all about using the features available to see that you might have a problem then applying the toolset to capture them.



This ^^^ screenshot is a good example where some Persistence shows it's present then we can narrow in and seen if it's a one off or repetitive.
In this post I did a similar exercise with the older SDS1104X-E which gives some idea of using a few of the scopes features:
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg1370717/#msg1370717

Comparing SDS1104X-E and SDS2104X-Plus.  Does the cheaper unit have same toos/function (trigger, search, measure) for the purpose of design verification, to hunt for abnormal signals in 10MHz SPI signal between MCU and external chips?

Is there more differences than what I found so far from scanning manual:

500MSa/s   vs   1000MSa/s in 4 channels mode (MISO, MOSI, Clock, nChipSelect), 500M is 50 samples per 10MHz signal pulse.  Enough to see abnormal glitch, runt, overshoot, undershoot???

7 Mpts/CH  vs  100Mpts, 7M captures 140,000 pulses at 50 samples per pulse. 

No histogram, still has StdDev to qualtify signal jitter.

7 inches and non-touch screen.   Needs a few more human seconds to use knobs to active a function, right?

TBH, today if wanting a lower cost solution and happy with a 7" display, SDS814X HD is where it's at.

Many thanks.  From SDS814X HD user manual, it has a new trigger mode, among manys, that seem to fit my speific SPI signal hunting.

Being new to modern MSO, please kindly advise if below is correct understanding:

Yep, a significant but certainly not insurmountable learning curve awaits you.

However a DSO provides capability a CRO can't, DSO trigger capabilities are next level....and some and then add how you can inspect a captured signal or its decode.
All you have left is to consider your spend and the features a little more $ can provide.

We have already discussed the Serial Test feature the SDS2000X Plus models offer, not available in the lower cost models.
Take the time to revisit the previous replies and the screenshots .......

[Wilson__]

2. If I set trigger "setup time < data sheet value", and let the McU run overnight.  if the scope did not trigger, there was no glitch nor rare abnormal events leading to non-compliance of "setup time should be longer than datasheet value".

Even better: Modern DSOs can configure segmented memory and record all the events separately, along with a count of how many there were.

(assuming there's not too many of them to fit in the 'scopes memory ... )

After the scope has capture multiple copies of the trigger event plus one screen width of data around trigger point, I can SEARCH the captured data using different 'trigger condition' to find some abnormal events, right?  For example, pulse abnormally higher voltage;  some has RUNT

[Wilson__]

I you want a MSO option that works fine, you should have a look at this user report linked below. I don't think you'll be happy with SDS1104X-E for that, you'd better get at least Rigol MSO5000 or Siglent SDS2000, or perhaps the new siglents 800 HD, and even then, I would ask for some real user-experience-based report about the specific model you're looking at, before pulling the trigger

https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg3528422/#msg3528422

I scanned this forum.  Apparent lower model Siglent, LA is a separate sub-system that link to the scope over a cable.  This has limited speed.   Higher end model has the LA circuit inside the scope, run fastrer and better integration.

I shall limit my expection to 4 channel LA, using the scope channel. SPI only need 4 and that fits my task on hand.   I can get an higher model if next task requires and factor into projet cost.

[Performa01]

After the scope has capture multiple copies of the trigger event plus one screen width of data around trigger point, I can SEARCH the captured data using different 'trigger condition' to find some abnormal events, right?  For example, pulse abnormally higher voltage;  some has RUNT

Even better: the always active history stores the trigger events for you without the need for you to do anything for it. Up to 80000 trigger events (depending on timabase and memory depth) can be stored in the history and inspected to your hearts content afterwards. All the tools are available and a couple advanced triggers are supported for event search, which works across the history if so desired. Compared to Sequence mode, history is just not as fast and there is no guaranteed max. dead time (down to 2 µs!), which is still sufficient in many use cases.