CORRECTION: The R&S "Find Level" isn't as clever as I thought it was. From the manual: "Sets the trigger level automatically to 0.5 * (MaxPeak - MinPeak)." It just so happened there was a glitch at that mid point and hence found it. Entirely coincidence in this case, not a clever feature.
It’s helpful to understand how oscilloscopes capture and display a signal, specifically with how the trigger system works. For today’s digital oscilloscopes, it would be handy if the scope’s ADC could capture the samples, evaluate our trigger criteria, and stop the capture long enough to display the result to the screen, then rearm the trigger and repeat. But this generally isn’t possible with oscilloscopes, as this would require processing data at insane speeds. 12 bit/sample * 1.25 Gsamples/second = 15 Gbit/sec. At higher bandwidths and sample rates (Keysight’s 256 GSa/s UXR 110 GHz Monster of an oscilloscope), this outpaces the real-time processing power of most FPGAs and ASICS available today. So what we typically get is either a simple analog level trigger, a downsampled and decimated copy of the incoming signal, or an untriggered capture and an offline search within the capture buffer to find the signal. The zone trigger is computationally very expensive and generally can’t be done at the full sample rate of the oscilloscopes in real time-and is therefore either limited to lower sample rates or is performed as an offline search. It’s helpful to describe a trigger as either online or offline. Only an online trigger can guarantee that the oscilloscope will capture the very next trigger event after the trigger system is rearmed. An offline trigger (unsynchronized capture and offline search) cannot guarantee that it will capture the next event after being rearmed. It’s also helpful to understand that the capture system for today’s digital oscilloscopes goes offline while it’s processing and displaying the results. This is the blind time that the MXO reported. For catching rare signals that you haven’t figured out how to trigger on, having a minimal blind time increases the chances you’ll catch a signal in infinite persistence mode for a given amount of time. The holy grail for an oscilloscope is one that has no blind time (it can rearm a trigger immediately, processing and display of an event can happen in parallel with another acquisition), and a full-bandwidth online trigger that operates on the full signal, and supports multi-channel trigger logic.
I have a Rigol MSO5074 with all the add-ons. I reckon I don't have enough years left in me to discover (and use) all of its capabilities, BUT at least this is something I've now mastered and will put to some use. Great video and well worth a watch. Be great to see more oscilloscope tutorials like this.
Are we now needing AI in these scopes to analyse our waveforms and tell us if there are any glitch/runt pulses for us, instead of manually looking ourselves?
That's the thing that always gets me... Test gear is getting more and more feature-packed, as if it was designed to think for engineers. It's like the old science fiction "computer this! computer that! computer tie my friggin' shoelaces cause I'm too lazy to bend over!" only with scopes and meters.
I’m gonna have to get myself a modern scope. I’m still using a tektronix 40mhz crt jobbie that I cab barely lift. I can see a Rigol 800 series coming my way in the not too distant.
If you are unlucky your scope trigger rate, when all is said and done, may end up "in sync" with a glitch or event and it may never happen to capture at the right moment, always missing it. This is why the adjustment of triggering options is the better way to go. You can make sure the scope is ready and starts to acquire data and then it triggers on the glitch when it happens, regardless of how long it had to wait.
Keep ‘em coming, Dave. Even my HP 54645D from the 90s has pretty sophisticated glitch and pattern triggering options. For pattern, the operator can enter in a few points of what the scope should try and find. You’ll also find very sophisticated triggering options on LeCroy HDO series, if you’re into that universe of “different” user interfaces. I wonder if any current scopes still offer NTSC video line triggering…?
I have found ONE thing tektronix does better. You can setup a simple trigger and capture some waveform data. Then setup a different trigger to search through that data and jump to each found piece. This lets you capture the data once, and slowly refine your search trigger until you get what you want. This is kind of nice when generating the problem signal is an arduous task, like rebooting the chip and and sending a transmit command. It only takes 4 seconds but if you have to do it 50 times it gets old.
Heh, this video comes just a few weeks after I finished evaluating a bunch of scopes, and I spent a lot of time testing triggering on this type of glitch. (I compared the R&S MXO 4, the Keysight 3000G, LeCroy WaveSurfer 3000z and 4000HD, and Tek MSO 24 and MDO 34.) I had a loaner Tek demo board, and it has two glitch capture signals (rare anomaly and frequent anomaly) that were very informative. FWIW I ended up going with the MXO 4, but the Keysight 3000G, LeCroy 4000HD, and Tek MDO34 were all nice, too. The Keysight is the absolute best at capturing glitches, with the MXO4 being almost as good. (The MXO4 slows down with measurements, as do all the others except the Keysight.) I think I like LeCroy’s UI design the best. If Keysight updated the 3000 series with higher resolution displays and deeper memory, I probably would have gone with that.
My mistake. Did not see the chapter at the very end of the video. Zone trigger can be used to capture this effectively using the other modes such as "do not intersect zone". I use it to capture runts and ringing events on power MOSFET hard switching. In this case, you can set rising level trigger to go off on the glitch, but then use do not intersect to filter out anything that is not a glitch (effectively becoming a trigger on maximum pulse width). Of course, the more surgical trigger methods would be better, but I have found zone trigger is more intuitive if trying to capture something completely unexpected when viewing signals in free-running mode or if the event to be captured has very specific behavior, hence zone 1 and zone 2 whose filters can be set independently. For example, it's really useful for capturing power MOSFET miller plateau and diode reverse recovery, as in those cases I am using the scope to find out how long those events are.
I came across pulse trigger when I was debugging timing in an ESP32 program doing interrupt based sampling. Wanted to compare sampling times compared to time to process the samples.
13:50 This is very interesting case. Just a few months ago I was debugging my code for encoder and couldn't understand why it shows twice as large value as normal period from time to time. And it appeared that encoder is noisy and sometimes it gives a short pulse just before normal one like here. And I was able to capture it by using this kind of scope's triggering menu.
So shen you capture runt pulses in the waveform, do you get to identify any specific design issue that is the source of the spurious signal or is it just something that has to be accepted, such as pulse ringing in step functions?
@@davidwillmore Pretty much every benchtop scope from about $5000 up (and many below that) support mice and keyboards. As for good web interfaces, even $500 scopes commonly have them now. Older models (like the venerable Rigol DS1054Z, but also the Keysight 1000/2000/3000/4000 series, which are kinda long in the tooth) have web interfaces but with slow (Keysight) or nonexistent (Rigol) screen streaming. Newer Rigol models have robust web interfaces. What surprised me is that even at $10K (or $50k!), LeCroy scopes don’t have screen streaming. You get a 1 frame per second slideshow at best, and only within their desktop app. :/
"turn up the intensity" is so cathode ray tube era.... A phosphor-ish intensity grading is frankly almost useless. (Especially since NO single digital scope I have yet seen does it properly. Fading to background color =/= fading to transparent. This matters when overlaying one trace on top of another.) But main reason intensity grading is almost useless is due to most measurements on a scope not strictly caring much about the average signal. The rare oddities are usually way more interesting. Ie, inverse intensity grading is the wonderful feature worth having. It removes all the uninteresting stuff that otherwise just clutters up the trace and makes life harder. (and some scopes have this feature.) Zone triggering is likewise a nice feature. However, zone triggering is just a sequence of triggers that all needs to happen for the final trigger to be considered true. Sequencing trigger conditions is a fairly powerful feature as far as finding specific things in a signal. But yes, triggering of multiple uncorrelated events is not ideal here as far as time delays as a triggering condition is concerned. Sequencing triggers can get even more flamboyant if one also does it across multiple channels. As in if channel 1 needs to see some condition and then channel 3 needs to see its own condition and that this then leads to capturing a waveform.
@@EEVblog Yes, it does. Same for infinite persistence. But an inverse intensity is far better than trying to imitate a CRT display for most glitch/fault hunting adventures.
9:13 you say that it seems to be triggering on every second glitch. Is the 15 ms between glitches for both the top and bottom glitches? If it is 15 ms for any glitch and you only trigger on the top glitches then shouldn’t it be at half the rate?
Question EEVblog; when premium multimeters like Metrahit's 30M series claim extreme precision/accuracy, is it even legit compared to say, a Fluke 179? That is, the accuracy it's showing might simply be limited to the probes or other errors so it's just some arbitrary numbers you're seeing?
You did not give a demonstration of zone trigger. You did not take the five minutes to actually learn how to use this incredibly useful feature. It is right on keysight's website that the boxes are to be used in conjuction (logical AND) with traditional trigger settings. Rising glitches can be easily captured using a rising level trigger and a do not intersect zone. Same thing with falling glitches, except you use a falling level trigger
CORRECTION: The R&S "Find Level" isn't as clever as I thought it was. From the manual: "Sets the trigger level automatically to 0.5 * (MaxPeak - MinPeak)." It just so happened there was a glitch at that mid point and hence found it. Entirely coincidence in this case, not a clever feature.
I love your videos dave . I don't have a oscilloscope but im learning a lot from you🤍
Free Palestine 🇵🇸🇵🇸
Will I hope you get one soon 😊
Free Palestine!
Always nice to learn about the oscilloscope functions that are there but not always obvious how to best use.
Most interesting. Oscilloscopes are so much more than the old day TV scopes on a hobby budget. 🥳 Thanks.
Extremely informative Dave!! YES, keep these types of training videos coming!
More tutorials on using an oscilloscope please, mainly the more affordable mid range ones
It’s helpful to understand how oscilloscopes capture and display a signal, specifically with how the trigger system works.
For today’s digital oscilloscopes, it would be handy if the scope’s ADC could capture the samples, evaluate our trigger criteria, and stop the capture long enough to display the result to the screen, then rearm the trigger and repeat. But this generally isn’t possible with oscilloscopes, as this would require processing data at insane speeds. 12 bit/sample * 1.25 Gsamples/second = 15 Gbit/sec. At higher bandwidths and sample rates (Keysight’s 256 GSa/s UXR 110 GHz Monster of an oscilloscope), this outpaces the real-time processing power of most FPGAs and ASICS available today.
So what we typically get is either a simple analog level trigger, a downsampled and decimated copy of the incoming signal, or an untriggered capture and an offline search within the capture buffer to find the signal. The zone trigger is computationally very expensive and generally can’t be done at the full sample rate of the oscilloscopes in real time-and is therefore either limited to lower sample rates or is performed as an offline search.
It’s helpful to describe a trigger as either online or offline. Only an online trigger can guarantee that the oscilloscope will capture the very next trigger event after the trigger system is rearmed. An offline trigger (unsynchronized capture and offline search) cannot guarantee that it will capture the next event after being rearmed. It’s also helpful to understand that the capture system for today’s digital oscilloscopes goes offline while it’s processing and displaying the results. This is the blind time that the MXO reported. For catching rare signals that you haven’t figured out how to trigger on, having a minimal blind time increases the chances you’ll catch a signal in infinite persistence mode for a given amount of time.
The holy grail for an oscilloscope is one that has no blind time (it can rearm a trigger immediately, processing and display of an event can happen in parallel with another acquisition), and a full-bandwidth online trigger that operates on the full signal, and supports multi-channel trigger logic.
I've been saving up for one of those R&S scopes. Only $20,000 to go!
You missed Black Friday.
I have a Rigol MSO5074 with all the add-ons. I reckon I don't have enough years left in me to discover (and use) all of its capabilities, BUT at least this is something I've now mastered and will put to some use. Great video and well worth a watch. Be great to see more oscilloscope tutorials like this.
Are we now needing AI in these scopes to analyse our waveforms and tell us if there are any glitch/runt pulses for us, instead of manually looking ourselves?
Wait for next years model, probably inevitable.
That's the thing that always gets me... Test gear is getting more and more feature-packed, as if it was designed to think for engineers. It's like the old science fiction "computer this! computer that! computer tie my friggin' shoelaces cause I'm too lazy to bend over!" only with scopes and meters.
You would never need more than 32k neurons per Ai.
Lol
@@hommadi2001 GLaDOS surely had a lot more! :)
AI requires compute power
I’m gonna have to get myself a modern scope. I’m still using a tektronix 40mhz crt jobbie that I cab barely lift. I can see a Rigol 800 series coming my way in the not too distant.
The Siglent SDS2000X HD scope has zone triggering as well. It's very useful if you want to capture a runt or glitch that isn't free running.
Yep, details like this are solid gold !...cheers.
I’ll watch any video with the MXO 4 😳😍
Now i have a question sir Dave, where did you get that glitch and how could we eliminate it?
If you are unlucky your scope trigger rate, when all is said and done, may end up "in sync" with a glitch or event and it may never happen to capture at the right moment, always missing it.
This is why the adjustment of triggering options is the better way to go. You can make sure the scope is ready and starts to acquire data and then it triggers on the glitch when it happens, regardless of how long it had to wait.
Keep ‘em coming, Dave. Even my HP 54645D from the 90s has pretty sophisticated glitch and pattern triggering options. For pattern, the operator can enter in a few points of what the scope should try and find. You’ll also find very sophisticated triggering options on LeCroy HDO series, if you’re into that universe of “different” user interfaces.
I wonder if any current scopes still offer NTSC video line triggering…?
Amazing video, thanks!
Nice video Dave - thanks!
I have found ONE thing tektronix does better. You can setup a simple trigger and capture some waveform data. Then setup a different trigger to search through that data and jump to each found piece. This lets you capture the data once, and slowly refine your search trigger until you get what you want. This is kind of nice when generating the problem signal is an arduous task, like rebooting the chip and and sending a transmit command. It only takes 4 seconds but if you have to do it 50 times it gets old.
Nice Demo Dave.🤓
Interesting stuff.
Heh, this video comes just a few weeks after I finished evaluating a bunch of scopes, and I spent a lot of time testing triggering on this type of glitch. (I compared the R&S MXO 4, the Keysight 3000G, LeCroy WaveSurfer 3000z and 4000HD, and Tek MSO 24 and MDO 34.) I had a loaner Tek demo board, and it has two glitch capture signals (rare anomaly and frequent anomaly) that were very informative.
FWIW I ended up going with the MXO 4, but the Keysight 3000G, LeCroy 4000HD, and Tek MDO34 were all nice, too. The Keysight is the absolute best at capturing glitches, with the MXO4 being almost as good. (The MXO4 slows down with measurements, as do all the others except the Keysight.) I think I like LeCroy’s UI design the best. If Keysight updated the 3000 series with higher resolution displays and deeper memory, I probably would have gone with that.
5:00. i was surprised, your davecad until now is still operational. will becoming vintage. :D
My mistake. Did not see the chapter at the very end of the video. Zone trigger can be used to capture this effectively using the other modes such as "do not intersect zone". I use it to capture runts and ringing events on power MOSFET hard switching. In this case, you can set rising level trigger to go off on the glitch, but then use do not intersect to filter out anything that is not a glitch (effectively becoming a trigger on maximum pulse width). Of course, the more surgical trigger methods would be better, but I have found zone trigger is more intuitive if trying to capture something completely unexpected when viewing signals in free-running mode or if the event to be captured has very specific behavior, hence zone 1 and zone 2 whose filters can be set independently. For example, it's really useful for capturing power MOSFET miller plateau and diode reverse recovery, as in those cases I am using the scope to find out how long those events are.
I came across pulse trigger when I was debugging timing in an ESP32 program doing interrupt based sampling. Wanted to compare sampling times compared to time to process the samples.
Hello David . Maybe a Micsig's NEW Differential Probe ? - MDP701 Review Ehh? Please?
Yes but where this glitches come? Are they a problem and why? How to remove them? Thanks
13:50 This is very interesting case. Just a few months ago I was debugging my code for encoder and couldn't understand why it shows twice as large value as normal period from time to time.
And it appeared that encoder is noisy and sometimes it gives a short pulse just before normal one like here.
And I was able to capture it by using this kind of scope's triggering menu.
So shen you capture runt pulses in the waveform, do you get to identify any specific design issue that is the source of the spurious signal or is it just something that has to be accepted, such as pulse ringing in step functions?
Can you plug a mouse into that R&S scope? Those menus look easier to navigate with one.
It supports USB keyboards and mice. You can also operate it nicely via the screen streaming via the internal web server over Ethernet.
@@tookitogo I guess for a $10K+ device that makes sense.
@@davidwillmore Pretty much every benchtop scope from about $5000 up (and many below that) support mice and keyboards.
As for good web interfaces, even $500 scopes commonly have them now. Older models (like the venerable Rigol DS1054Z, but also the Keysight 1000/2000/3000/4000 series, which are kinda long in the tooth) have web interfaces but with slow (Keysight) or nonexistent (Rigol) screen streaming. Newer Rigol models have robust web interfaces.
What surprised me is that even at $10K (or $50k!), LeCroy scopes don’t have screen streaming. You get a 1 frame per second slideshow at best, and only within their desktop app. :/
Is the R&S MXO 4 a buggy scope? At around @2:12 a glitch disappears and moves despite of persistent mode being on.
There's an edit cut. If you see the clock on the scope screen it jumps.
"turn up the intensity" is so cathode ray tube era....
A phosphor-ish intensity grading is frankly almost useless. (Especially since NO single digital scope I have yet seen does it properly. Fading to background color =/= fading to transparent. This matters when overlaying one trace on top of another.)
But main reason intensity grading is almost useless is due to most measurements on a scope not strictly caring much about the average signal. The rare oddities are usually way more interesting. Ie, inverse intensity grading is the wonderful feature worth having. It removes all the uninteresting stuff that otherwise just clutters up the trace and makes life harder. (and some scopes have this feature.)
Zone triggering is likewise a nice feature. However, zone triggering is just a sequence of triggers that all needs to happen for the final trigger to be considered true. Sequencing trigger conditions is a fairly powerful feature as far as finding specific things in a signal. But yes, triggering of multiple uncorrelated events is not ideal here as far as time delays as a triggering condition is concerned.
Sequencing triggers can get even more flamboyant if one also does it across multiple channels. As in if channel 1 needs to see some condition and then channel 3 needs to see its own condition and that this then leads to capturing a waveform.
Turning up the intensity does actually help.
@@EEVblog Yes, it does. Same for infinite persistence.
But an inverse intensity is far better than trying to imitate a CRT display for most glitch/fault hunting adventures.
9:13 you say that it seems to be triggering on every second glitch. Is the 15 ms between glitches for both the top and bottom glitches? If it is 15 ms for any glitch and you only trigger on the top glitches then shouldn’t it be at half the rate?
Now we got the glitch, and what is causing it?
A deliberate test signal in this case.
75 "tech points" awarded.
Question is why the glitches are there?!?!
yeah you found them now its off to ChatGPT to ask what they mean !
A deliberate test signal in this case.
@@EEVblognice, how ?
Interesting.
I wonder how many scopes Dave owns....
Question EEVblog; when premium multimeters like Metrahit's 30M series claim extreme precision/accuracy, is it even legit compared to say, a Fluke 179? That is, the accuracy it's showing might simply be limited to the probes or other errors so it's just some arbitrary numbers you're seeing?
You did not give a demonstration of zone trigger. You did not take the five minutes to actually learn how to use this incredibly useful feature.
It is right on keysight's website that the boxes are to be used in conjuction (logical AND) with traditional trigger settings.
Rising glitches can be easily captured using a rising level trigger and a do not intersect zone. Same thing with falling glitches, except you use a falling level trigger
Please feed this waveform to an analog oscilloscope