Author Topic: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected  (Read 2350 times)

0 Members and 1 Guest are viewing this topic.

Offline TheWaterbugTopic starter

  • Contributor
  • Posts: 15
  • Country: us
10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« on: January 23, 2024, 01:42:25 am »
Long question, but I'm stumped. I'm experimenting with creating a 40 MHz TTL clock signal from a 10 MHz TTL source, using a set of filters and frequency doubler modules from Mini-Circuits, along with a sinewave-to-squarewave converter (basically a comparator), as follows:



The result is 4 output pulses (blue) per input pulse (green), with an average frequency of 40 MHz, shown here when my CSA8200 is triggered on a copy of the 10 MHz signal:



but there's a distinct repeating pattern of A, B, C, and D output pulse per input pulse:



The width of A1 is slightly different from the width of B1, C1, and D1, but the same as the width of A2 (following D1).

The period from A1↑ to B1↑ is slightly different from the period from B1↑ to C1↑, but the same as from A2↑ to B2↑.

This is completely as expected from this setup, as the frequency multiplication modules are passive. The filters are to convert the square waves to sine wave before going into the multipliers, and to clean up the doubled frequencies so the converter can create closer to a 50% duty cycle when converting back to square wave. The middle stage of square wave conversion is required to get enough energy for the 2nd frequency doubler.

Since each set of edges from pulses A-D, rising or falling, is generated by a single green clock cycle, the jitter measured on A-D when triggered on 10 MHz is pretty good, typically < 10 ps RMS, regardless of which 40 MHz edge I measure or which 10 MHz edge I trigger on:



What I don't understand is what I see when I reverse the trigger relationship, and trigger on a copy of the 40 MHz signal, as follows:



If I display the 10 MHz and 40 MHz signals with a waveform database I see a sort of eye, but whereas I would have expected to see a lot of jitter on those 10 MHz edges, due to the edge jitter on the 40 MHz trigger, instead I see a clean 10 MHz edge:



The 40 MHz signal has 4 distinct sets of edges:



Why don't the rising edges line up if I'm triggering on the rising edge of its duplicate, or why don't the falling edges line up if I'm triggering on the falling edge of its duplicate? When I zoom in on that 10 MHz edge, I get <11 ps RMS jitter:



Am I missing something really obvious? Why aren't my 10 MHz edges jittering with respect to my 40 MHz trigger?

The trigger setup in all cases is External Direct with no prescalar, set to 50% level.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #1 on: January 23, 2024, 02:37:13 am »
Adjust the trigger holdoff.

What may be happening is that while you are applying the 40 MHz signal for triggering, the trigger itself may be only using every 4th edge, or every multiple of every 4th edge, because of how long it takes to rearm.  When this happens, you only see every 4th low frequency edge so no jitter is present.

Altering the rearm time by adjusting the holdoff can make sure that the trigger is asynchronous to the trigger signal and every edge will be represented.
 
The following users thanked this post: TheWaterbug

Offline dobsonr741

  • Frequent Contributor
  • **
  • Posts: 706
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #2 on: January 23, 2024, 03:56:23 am »
You are expecting a perfect time domain signal after spectral manipulation with filters.

How good are the filters? If the spectral domain is not perfect, the time domain signal will be imperfect, too.

Trying to make common sense, someone with a more persuasive math could possibly support it. I’m no longer good at Laplace, Z and Fourier trasnformations.
« Last Edit: January 23, 2024, 02:39:30 pm by dobsonr741 »
 
The following users thanked this post: TheWaterbug

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3516
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #3 on: January 23, 2024, 02:01:26 pm »
I think both points made are very important.

I have an 11801 with SD-22,24,26 & 32 heads.  My favorite toy.  It cannot trigger more often than  200 kHz when sampling a single signal, less in sequential mode sampling multiple inputs.  The CSA 8200 has the same limitation.  It's an 11801 with a fancier GUI.  Look at p. 8 of the CSA8200  datasheet.  At 10MHz you are triggering on every 40th edge.  At 40 MHz it's every 160th edge.  Therein lies the explanation. 

Any harmonics that get by the LPFs will eat your lunch.  I suspect that a 4395A will find the culprit(s).

Have Fun!
Reg
 
The following users thanked this post: TheWaterbug

Offline tomato

  • Regular Contributor
  • *
  • Posts: 206
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #4 on: January 23, 2024, 05:21:08 pm »
Am I missing something really obvious? Why aren't my 10 MHz edges jittering with respect to my 40 MHz trigger?

The trigger setup in all cases is External Direct with no prescalar, set to 50% level.
Do you see the same results with the scope triggered on both positive and negative edges of the 40 MHz signal?
 
The following users thanked this post: TheWaterbug

Offline TheWaterbugTopic starter

  • Contributor
  • Posts: 15
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #5 on: January 23, 2024, 06:22:22 pm »
Adjust the trigger holdoff.

What may be happening is that while you are applying the 40 MHz signal for triggering, the trigger itself may be only using every 4th edge, or every multiple of every 4th edge, because of how long it takes to rearm.  When this happens, you only see every 4th low frequency edge so no jitter is present.

Altering the rearm time by adjusting the holdoff can make sure that the trigger is asynchronous to the trigger signal and every edge will be represented.

Thanks! I just tried this, and I don't see any change. I entered holdoff values from 5.000 µs (minimum) to 5.100 µs in 0.010 µs (10 ns) increments, which appears to be the resolution* of the holdoff setting, and the display always was the same, e.g. 4 distinct edges on the rising edge of the 40 MHz signal and a solid single edge on the 10 MHz signal.

Given that my two periods are 25 ns and 100 ns, I would think that stepping the holdoff in 10 ns increments over a 100 ns interval would give me all possible trigger points, but the results were always the same.  :-//

* I originally wasn't sure of the resolution, or that my holdoff setting was truly being applied, since the waveform didn't "refresh" when I clicked Ok, so I toggled the horizontal scale in/out after every increment to force a refresh. And I'm pretty sure my settings were valid, because I also closed and reopened the trigger dialog many times, and the holdoff setting was populated with what I'd typed in before I closed it (which it did _not_ do when I tried increments of 0.001 µs / 1 ns).
 

Offline TheWaterbugTopic starter

  • Contributor
  • Posts: 15
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #6 on: January 23, 2024, 06:30:22 pm »
You are expecting a perfect time domain signal after spectral manipulation with filters.

How good are the filters? If the spectral domain is not perfect, the time domain signal will be imperfect, too.

Trying to make common sense, someone with a more persuasive math could possibly support it. I’m no longer good at Laplace, Z and Fourier trasnformations.

I think the time domain signal is pretty darn good, as evidenced by the jitter measurements on each of the 40 MHz edges when triggered on the 10 MHz source. No matter which of the 4 pulses (A, B, C, or D) I measure, rising or falling edge, I always get something on the order of 10 - 12 ps, RMS.

Here are the filters I'm using from minicircuits:

I'm not a filter guy, but I think they're pretty good. The yellow trace in my first capture is the 15 MHz LPFed copy of my original 10 MHz TTL square wave.
 

Offline TheWaterbugTopic starter

  • Contributor
  • Posts: 15
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #7 on: January 23, 2024, 06:33:15 pm »
Am I missing something really obvious? Why aren't my 10 MHz edges jittering with respect to my 40 MHz trigger?

The trigger setup in all cases is External Direct with no prescalar, set to 50% level.
Do you see the same results with the scope triggered on both positive and negative edges of the 40 MHz signal?

Yes, the results are very, very similar. If I switch 40 MHz trigger edges the jitter on the 10 MHz edge changes by maybe 2 ps, which I would consider to be within sampling error.

If I switch 40 MHz trigger edges the "bunching" of the 40 MHz edges on the lower waveform changes slightly, so the opposite edges get a little more spread out, but I always see 4 distinct traces on both edges of the 40 MHz signal, no matter which 40 MHz edge I trigger on.
 

Offline TheWaterbugTopic starter

  • Contributor
  • Posts: 15
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #8 on: January 23, 2024, 06:36:23 pm »
Also, I gotta give a big Thanks!! overall to this forum. I posted the same question over on the Tektronix scope forum, and there are zero replies.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3516
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #9 on: January 24, 2024, 01:47:31 am »
The CSA8200 datasheet specifies 150 kSa/s  for two traces (i.e.  sequential sampling).  Your 40 MHz signal is being sampled every 266.66666 edges.  The 10 MHz signal is being sampled every 66.66666 edges.  I'll let you work out the details of the consequences.

It *may* not be sampling at 150 kHz.  That's just the datasheet specified maximum rate.

Have Fun!
Reg
« Last Edit: January 24, 2024, 01:50:11 am by rhb »
 
The following users thanked this post: TheWaterbug

Offline TheWaterbugTopic starter

  • Contributor
  • Posts: 15
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #10 on: January 24, 2024, 11:28:13 pm »
Hmmm. I swapped out the CSA8200 sampling scope for a TDS5104 digital phosphor scope. Triggering on the 40 MHz edge with 2 seconds of display persistence:



Now the 40 MHz signal looks more like what I'd expected--if I trigger on the 40 MHz rising edge, I get a single 40 MHz rising edge, and the 40 MHz falling edge shows my width jitter:



(hard to see on the scope capture, but there are 4 distinct falling edges in there).

If I trigger on the 40 MHz falling edge, I get a single 40 MHz falling edge, and the 40 MHz rising edge now shows 4 distinct traces:



And in all cases, the 10 MHz edge refuses to jitter :-), and looks too clean to be true. There's no jitter measurement available on this scope, but it looks as clean as my trigger edge:



If I were to guess, I'd say that the CSA8200 trigger circuit was trying to find some inherent periodicity in the somewhat unstable 40 MHz signal, and somehow locked onto the 10 MHz, since that's the original source of the 40 MHz pulses.

Whereas the TDS5104 is just dutifully doing what it's told, and triggering on the rising (or falling) edges.

But none of that explains why the 10 MHz is so clean relative to a 40 MHz trigger that has both period and width jitter. I don't believe it.  ;D

p.s. anyone know how to un-invert the colors on TekVisa Toolbar screen captures? I've tried File: Page Setup: Ink Saver Off and On, and it's always the same over TekVISA.
« Last Edit: January 25, 2024, 02:01:20 am by TheWaterbug »
 

Offline 2N3055

  • Super Contributor
  • ***
  • Posts: 7466
  • Country: hr
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #11 on: January 25, 2024, 12:05:57 am »
It is simple.

You are always seeing only one edge from 10MHz synchronized with first edge from 40MHz signal.
Other 3 trigger events from 40Mhz signal will not show any edges from 10 MHz one, but will show low or high (hence the solid line on top and bottom).
40Mhz signal is actually 4 periodic events repeated at 10Mhz period in perfect synch with original 10Mhz.
It is not jittery 40Mhz but asymmetrical pattern repeated at 10 Mhz.

This can be simply demonstrated on a scope with segmented memory, where all 4 consecutive trigger events are captured separately and can be observed sequentially.
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: points2, TheWaterbug

Offline TheWaterbugTopic starter

  • Contributor
  • Posts: 15
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #12 on: January 25, 2024, 12:20:26 am »
It is simple.

You are always seeing only one edge from 10MHz synchronized with first edge from 40MHz signal.
Other 3 trigger events from 40Mhz signal will not show any edges from 10 MHz one, but will show low or high (hence the solid line on top and bottom).
40Mhz signal is actually 4 periodic events repeated at 10Mhz period in perfect synch with original 10Mhz.
It is not jittery 40Mhz but asymmetrical pattern repeated at 10 Mhz.

This can be simply demonstrated on a scope with segmented memory, where all 4 consecutive trigger events are captured separately and can be observed sequentially.

Ah, now this is finally making sense! It's exactly the inverse of the ABCD diagram, above, when triggering on the 10 MHz:



Each time I trigger on a 40 MHz edge it'll be associated with one of an A, B, C, or D edge from the 10 MHz, but they'll always be in those positions relative to my trigger edge, so the A edge never jitters, and the B edge never jitters, but the interval from A to B is different from the interval from B to C, etc.

Thank you!! This has been driving me crazy for a week.
 

Offline points2

  • Regular Contributor
  • *
  • Posts: 54
  • Country: fr
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #13 on: January 25, 2024, 03:17:07 am »
hi 2N3055,
This can be simply demonstrated on a scope with segmented memory, where all 4 consecutive trigger events are captured separately and can be observed sequentially.
Waouh ! Interesting !
Do you have in mind a scope that can do that ?

My "noob-level" comment (if I'm wrong, don't hesitate to tell me, I have some many thing to learn...  ;D ):
- I use my scope to get a rough view of the wave (eg a 10MHz osc, or even an Ethernet signal @ 125MHz). But I don't trust its figures.
- To get trusty results => I use a spectrum analyzer, to see/measure the spectral density around the frequency corresponding to the rising edge seen on the scope! As simple & efficient as that ! (I'm correct or... not ? comments from experts is requested  :-DD )

So, to me :
- an eye-diagram is a marketing stuff ! Because it is nice to watch... But at which cost, and to get sometimes untrusty/weird/false results ?! Waouh ! => time-domain => watch out ! A scope isn't a fine & precise device (hobbyist scopes).
- the spectral density (mirror of an eye-D in the frequency domain) using a SA is less fun to watch... of course.. => but these devices are way cheaper, fixtures/etc are way simplier, and you get trusty & very very precise results (eg : a 10MHz sinwave signal, using a basic SA, you can look at the spectral density with a RBW=1Hz ! Converted in second => crazy precision ! )

My conclusion :
- I have a 500MHz BW scope, good&quite unexpensive (SDS2104X+ converted into 2504X+) => cool => it's a keeper
- last weeks, I've been looking for a more advanced scope to see more clearly down to the ns & sub-ns level => no go ! Bad idea !
- to see clear & deep => I already have a SA (SSA3021X+) => it is a keeper.
- if I change a measurement device in the near future => I'll get a more advance SA => a RTSA (Aaronia is releasing a new RTSA for hobbyist with crazy low price/perf ratio vs any other RTSA for hobbyist, and way cheaper than high-end scope for hobbyist when the goal is to watch a eye-diag. ; but I wait for feedbacks from early adopters...)
 

Offline 2N3055

  • Super Contributor
  • ***
  • Posts: 7466
  • Country: hr
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #14 on: January 25, 2024, 06:21:49 am »
hi 2N3055,
This can be simply demonstrated on a scope with segmented memory, where all 4 consecutive trigger events are captured separately and can be observed sequentially.
Waouh ! Interesting !
Do you have in mind a scope that can do that ?

My "noob-level" comment (if I'm wrong, don't hesitate to tell me, I have some many thing to learn...  ;D ):
- I use my scope to get a rough view of the wave (eg a 10MHz osc, or even an Ethernet signal @ 125MHz). But I don't trust its figures.
- To get trusty results => I use a spectrum analyzer, to see/measure the spectral density around the frequency corresponding to the rising edge seen on the scope! As simple & efficient as that ! (I'm correct or... not ? comments from experts is requested  :-DD )

So, to me :
- an eye-diagram is a marketing stuff ! Because it is nice to watch... But at which cost, and to get sometimes untrusty/weird/false results ?! Waouh ! => time-domain => watch out ! A scope isn't a fine & precise device (hobbyist scopes).
- the spectral density (mirror of an eye-D in the frequency domain) using a SA is less fun to watch... of course.. => but these devices are way cheaper, fixtures/etc are way simplier, and you get trusty & very very precise results (eg : a 10MHz sinwave signal, using a basic SA, you can look at the spectral density with a RBW=1Hz ! Converted in second => crazy precision ! )

My conclusion :
- I have a 500MHz BW scope, good&quite unexpensive (SDS2104X+ converted into 2504X+) => cool => it's a keeper
- last weeks, I've been looking for a more advanced scope to see more clearly down to the ns & sub-ns level => no go ! Bad idea !
- to see clear & deep => I already have a SA (SSA3021X+) => it is a keeper.
- if I change a measurement device in the near future => I'll get a more advance SA => a RTSA (Aaronia is releasing a new RTSA for hobbyist with crazy low price/perf ratio vs any other RTSA for hobbyist, and way cheaper than high-end scope for hobbyist when the goal is to watch a eye-diag. ; but I wait for feedbacks from early adopters...)

Fast scopes are expensive..
Eye diagram is a diagram. It is useful for certain type of signals and usage.

Many scopes have segmented mode. If you have SDS2000X+ then you already have one that has it.
Problem is that retrigger time on scopes is much slower that <25ns period you have here..
So it won't be able to catch every consecutive period as trigger sequentially.  But by simply using history mode you can "walk" and see what they are..
« Last Edit: January 26, 2024, 08:33:23 am by 2N3055 »
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: TheWaterbug

Offline points2

  • Regular Contributor
  • *
  • Posts: 54
  • Country: fr
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #15 on: January 26, 2024, 03:57:36 am »
hi 2N3055
Fast scopes are expansive..
well... thanks to confirm something that everyone knows simply by watching at price tags whatever the seller website...
Eye diagram is a diagram. It is useful for certain type of signals and usage.
well... what else ?...
Many scopes have segmented mode. If you have SDS2000X+ then you already have one that has it.
Problem is that retrigger time on scopes is much slower that <25ns period you have here..
So it won't be able to catch every consecutive period as trigger sequentially.  But by simply using history mode you can "walk" and see what they are..
so what ?
you're just saying what I said above => a DSO is crap when you have to deal with fast signal.
Honestly, I never used the segment mode of my scope... because no-trust  8) ; but as you confirm it is not trusty... I'll stay away from it  :-DD

ok,
I stop complaining about your post, let's be efficient & measurement-driven :
what's your experience using a SA vs a DSO, in such an example like TheWaterbug above ? Your post is about DSO, not about the perf of SA vs a the perf of a DSO in such a case => why ? I find this no-answer "weird"...

@ TheWaterbug : only 1 basic remark vs your setup :
let's be honest, what kind of guy has adivised you to run such precise measurements on such "quite" fast signal, with SO MANY stuff between the DUT (DUT = where you "probe") and the scope !?
I do this remark based upon a farily basic lesson & calculations any pupil learn at +/- 15-17years-old in chemistry (sorry, I've only that example in mind  :-DD ) : what's the OVERALL uncertainty of my experiment !? Each step has its own incertainty, thus as I need x step to run my expriment => uncertaintly builds up. But if the overall uncertaintly remains negligeable vs the results => that's ok.
Here, honestly, no EE can stand your hypothesis that the overall uncertainty of your setup is negligeable vs the final result ! And espcially given the huge uncertainty given by the scope vs the overall uncertainty of all these boxes between the scope & the DUT !

my 2 cents.
If I'm wrong, anyone is welcome to teach me I'm wrong of course ! (my story : I got a degree in process&chem engineering some time ago. I've leanrt only ONE thing : "you know nothing, but you can learn"  :-DD ; as any science guy, after years, I still know nothing... and keep learning  :-DD  ; by the way, thanks to Dave to enable us to share knowledge, just to be aware about how "I still know nothing"  |O WTF ! :-DD )
 

Offline 2N3055

  • Super Contributor
  • ***
  • Posts: 7466
  • Country: hr
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #16 on: January 26, 2024, 09:08:23 am »
hi 2N3055
Fast scopes are expansive..
well... thanks to confirm something that everyone knows simply by watching at price tags whatever the seller website...
Eye diagram is a diagram. It is useful for certain type of signals and usage.
well... what else ?...
Many scopes have segmented mode. If you have SDS2000X+ then you already have one that has it.
Problem is that retrigger time on scopes is much slower that <25ns period you have here..
So it won't be able to catch every consecutive period as trigger sequentially.  But by simply using history mode you can "walk" and see what they are..
so what ?
you're just saying what I said above => a DSO is crap when you have to deal with fast signal.
Honestly, I never used the segment mode of my scope... because no-trust  8) ; but as you confirm it is not trusty... I'll stay away from it  :-DD

ok,
I stop complaining about your post, let's be efficient & measurement-driven :
what's your experience using a SA vs a DSO, in such an example like TheWaterbug above ? Your post is about DSO, not about the perf of SA vs a the perf of a DSO in such a case => why ? I find this no-answer "weird"...

@ TheWaterbug : only 1 basic remark vs your setup :
let's be honest, what kind of guy has adivised you to run such precise measurements on such "quite" fast signal, with SO MANY stuff between the DUT (DUT = where you "probe") and the scope !?
I do this remark based upon a farily basic lesson & calculations any pupil learn at +/- 15-17years-old in chemistry (sorry, I've only that example in mind  :-DD ) : what's the OVERALL uncertainty of my experiment !? Each step has its own incertainty, thus as I need x step to run my expriment => uncertaintly builds up. But if the overall uncertaintly remains negligeable vs the results => that's ok.
Here, honestly, no EE can stand your hypothesis that the overall uncertainty of your setup is negligeable vs the final result ! And espcially given the huge uncertainty given by the scope vs the overall uncertainty of all these boxes between the scope & the DUT !

my 2 cents.
If I'm wrong, anyone is welcome to teach me I'm wrong of course ! (my story : I got a degree in process&chem engineering some time ago. I've leanrt only ONE thing : "you know nothing, but you can learn"  :-DD ; as any science guy, after years, I still know nothing... and keep learning  :-DD  ; by the way, thanks to Dave to enable us to share knowledge, just to be aware about how "I still know nothing"  |O WTF ! :-DD )

Scopes are simply a tool.. User provides the thinking.
Every tool will have limitations. Very small screwdriver is needed to work on watches, but it will break if you try to use it on something big.

Understanding how tool exactly works and what are the limitations is the key to using it to maximum effect.
It is user's responsibility to know how to use it to good result...

As a science guy you know that, i suspect...

Best,

Siniša
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline TheWaterbugTopic starter

  • Contributor
  • Posts: 15
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #17 on: January 26, 2024, 05:47:29 pm »
It is simple.

You are always seeing only one edge from 10MHz synchronized with first edge from 40MHz signal.
Other 3 trigger events from 40Mhz signal will not show any edges from 10 MHz one, but will show low or high (hence the solid line on top and bottom).
40Mhz signal is actually 4 periodic events repeated at 10Mhz period in perfect synch with original 10Mhz.
It is not jittery 40Mhz but asymmetrical pattern repeated at 10 Mhz.

This can be simply demonstrated on a scope with segmented memory, where all 4 consecutive trigger events are captured separately and can be observed sequentially.

Ah, now this is finally making sense! It's exactly the inverse of the ABCD diagram, above, when triggering on the 10 MHz:



Each time I trigger on a 40 MHz edge it'll be associated with one of an A, B, C, or D edge from the 10 MHz, but they'll always be in those positions relative to my trigger edge, so the A edge never jitters, and the B edge never jitters, but the interval from A to B is different from the interval from B to C, etc.

Thank you!! This has been driving me crazy for a week.

. . . and now I think I also understand why this trace from the sampling scope differs from this trace from the DPO.

The sampling scope sees the trigger and starts sampling, but doesn't sample (only) the edge that caused the trigger, whereas it looks like the DPO does do that.

Correct?
 

Offline 2N3055

  • Super Contributor
  • ***
  • Posts: 7466
  • Country: hr
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #18 on: January 26, 2024, 06:01:24 pm »
It is simple.

You are always seeing only one edge from 10MHz synchronized with first edge from 40MHz signal.
Other 3 trigger events from 40Mhz signal will not show any edges from 10 MHz one, but will show low or high (hence the solid line on top and bottom).
40Mhz signal is actually 4 periodic events repeated at 10Mhz period in perfect synch with original 10Mhz.
It is not jittery 40Mhz but asymmetrical pattern repeated at 10 Mhz.

This can be simply demonstrated on a scope with segmented memory, where all 4 consecutive trigger events are captured separately and can be observed sequentially.

Ah, now this is finally making sense! It's exactly the inverse of the ABCD diagram, above, when triggering on the 10 MHz:



Each time I trigger on a 40 MHz edge it'll be associated with one of an A, B, C, or D edge from the 10 MHz, but they'll always be in those positions relative to my trigger edge, so the A edge never jitters, and the B edge never jitters, but the interval from A to B is different from the interval from B to C, etc.

Thank you!! This has been driving me crazy for a week.

. . . and now I think I also understand why this trace from the sampling scope differs from this trace from the DPO.

The sampling scope sees the trigger and starts sampling, but doesn't sample (only) the edge that caused the trigger, whereas it looks like the DPO does do that.

Correct?

Yes.
It will sometimes take only a single sample (dot) and take all 1000 others from 1000 consecutive separate triggers...

Realtime sampling scope takes whole screen in one go from the trigger.
Repetitive sampling scopes (ETS and RIS) "assemble" the screen worth of data from many, many separate trigger events.

Best,
Siniša
« Last Edit: January 26, 2024, 06:03:37 pm by 2N3055 »
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: TheWaterbug

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #19 on: January 27, 2024, 12:11:36 pm »
The sampling scope sees the trigger and starts sampling, but doesn't sample (only) the edge that caused the trigger, whereas it looks like the DPO does do that.

Modern sampling oscilloscopes work that way, but nothing prevents sampling before the trigger like some old Tektronix  models do.  Sampling before the trigger allows displaying the trigger edge and earlier without a delay line which would limit bandwidth.

In order to do this, the sampling oscilloscope phase locks to the trigger source allowing for a negative trigger delay to be applied to the sampling strobe.
 
The following users thanked this post: TheWaterbug

Offline TheWaterbugTopic starter

  • Contributor
  • Posts: 15
  • Country: us
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #20 on: January 27, 2024, 04:55:09 pm »
Modern sampling oscilloscopes work that way, but nothing prevents sampling before the trigger like some old Tektronix  models do.  Sampling before the trigger allows displaying the trigger edge and earlier without a delay line which would limit bandwidth.

In order to do this, the sampling oscilloscope phase locks to the trigger source allowing for a negative trigger delay to be applied to the sampling strobe.

Is the CSA8200 considered "old" 😁
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: 10 MHz on Sampling Scope w/40 MHz Trigger, Not As Expected
« Reply #21 on: January 27, 2024, 06:04:49 pm »
Modern sampling oscilloscopes work that way, but nothing prevents sampling before the trigger like some old Tektronix  models do.  Sampling before the trigger allows displaying the trigger edge and earlier without a delay line which would limit bandwidth.

In order to do this, the sampling oscilloscope phase locks to the trigger source allowing for a negative trigger delay to be applied to the sampling strobe.

Is the CSA8200 considered "old" 😁

No, the CSA8200 is modern on this scale, 2002?  The Tektronix 3T2 (1967) and 7T11 (1969 to 1990) supported random sampling.  Tektronix also made delay lines for sequential sampling oscilloscopes so the trigger edge could be viewed, but they severely limit bandwidth.

I think it came down to digital storage oscilloscopes becoming fast enough, and sampling oscilloscope applications not requiring viewing of the trigger edge, so random sampling was no longer required.
 
The following users thanked this post: TheWaterbug


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf