EEVblog #685 - What Is Oscilloscope AC Trigger Coupling?

[Odysseus]

As explained by Dave, the default on my analog scope is AC trigger coupling.  However, suppose you are observing a square wave of widely varying duty cycle.  At the extremes, if left on AC trigger coupling, you will loose stable triggering since the DC component of the signal is so close to the edge of the waveform. Here, it would be beneficial to just fix a DC trigger level at 50% that won't change with the signal. 

[rf-loop]

Trigger coupling AC and duty cycle of signal - problems? No.
In analog scopes it is highly related to scope quality.


This digital oscilloscope have true digitized side trigger engine.


Here 3V p-p pulse. Pulse cycle perod set for  100 000 000 ns (100ms)   and pulse width set for 25ns and set for rise time 10ns and for falling time in this case same 10ns.

Input channel DC coupled.
Trigger mode Normal and Rising edge.

First Trigger DC coupled.
Next Trigger AC coupled.  (and of course in AC coupling can also adjust trigger level)

Persistence was on for collect jitter sum. (generator itself jitter/drift + oscilloscope delay jitter/drift + trigger jitter)
(This "square wave" Duty cycle  =  ~0.000025%)

Visible is collected total jitter peak to peak over ~30s time and it looks stay around inside 1ns.

Around same with AC and DC trigger coupling.  Trigger AC coupling do not make any difficulty for rock solid trig.
Total peak to peak time jitter over 100ms time period is 1ns (0.01 ppm) independent of Trigger coupling AC or DC.


Of course time when take picture, current last trace is in random place inside this jitter range. And (repeat) all timing jitter is really not trig jitter alone. (it is sum of jitters and drifts during 30s period in generator and also oscilloscope delay counting during 100ms delay time + trig jitter)
(more tests about this oscilloscope basic triger perfomance  can find later in other place - not relevant in this thread)

[rs20]

Around same with AC and DC trigger coupling.  Trigger AC coupling do not make any difficulty for rock solid trig.

You're confusing the issue by using an AC-coupled trigger with a non-zero trigger level (don't think I didn't see the "L 1.42V" on your screenshot). If you put a 99.999% duty cycle signal in, instead of a 0.001% duty cycle signal in, your scope would fail to trigger on it until you fiddled with the level. Not so with DC-coupled trigger.

This is the point people previously posting in this thread are making. In your example, if you set DC coupling with level set to 0V, and leave the settings like that, the scope will trigger on 0.00001% and 99.99999% duty cycles flawlessly. However, if you set AC coupling with the level set to 0V, it'll only work from 2% to 98% (OK, I completely made these numbers up, but you get the idea). Sure, you can cheat and "fix" the problem by manipulating the trigger level, but that's really annoying to mess around with. Hence, DC trigger coupling is more reliable and easier.

Trigger coupling AC and duty cycle of signal - problems? No.
In analog scopes it is highly related to scope quality.

I imagine the range of duty cycles supported by the scope when set on AC triggering, level = 0 will vary from scope to scope, but there isn't a scope on the market where DC triggering isn't a simpler and more reliable solution to the problem.

[rf-loop]

This was not just test for this named case.
This was only picked up example from testings what I have made for this scope trigger system for find its goods and bads and that AC trigger can easy adjust also and it do not have any issue for trig. Álso in this case point where is trigger event detected is far out from sampling buffer in this time when it capture signal after this delay and it show that delay counting do not add any bad jitter.


(Small "bug"
as there can see in pictures there is one guite small problem related to how it show trigger level. (in picture there can see only this triggeer level mark in right side. Level number is qquite ok but level mark is not.  When adjust level it also show trigger level dashed line over display.  Related to signal image trigger level indicator is in wrong vertical position. Yes number is right.  But trig mark distance for AC "zero" is wrong. It may be confusing because need adjust visual trigger line nearly signal top or even over it for trig (if want trig is around 50% p-p).  AC zero is very near this signal bottom. But it display trig marker related to DC zero. Trigger level number is related to this AC "zero" and it is right.  Trigger level mark and marker line when adjust is not right level.  This is not right when there is selected trigger coupling AC and signal coupling DC.
Small thing but "not so nice".)

[rs20]

(Small "bug"
as there can see in pictures there is one guite small problem related to how it show trigger level. (in picture there can see only this triggeer level mark in right side. Level number is qquite ok but level mark is not.  When adjust level it also show trigger level dashed line over display.  Related to signal image trigger level indicator is in wrong vertical position. Yes number is right.  But trig mark distance for AC "zero" is wrong. It may be confusing because need adjust visual trigger line nearly signal top or even over it for trig (if want trig is around 50% p-p).  AC zero is very near this signal bottom. But it display trig marker related to DC zero. Trigger level number is related to this AC "zero" and it is right.  Trigger level mark and marker line when adjust is not right level.  This is not right when there is selected trigger coupling AC and signal coupling DC.
Small thing but "not so nice".)

This is exactly why most (?) scopes don't display the AC trigger level visually on screen at all. In theory you could have the oscilloscope measure the average DC value and actually calculate & display the trigger level on-screen, but I think this doesn't work either because AC-coupling only couples frequencies over a certain value, so it's not even that simple.

In other words, the bug is that it even tries to show you. That triangle should just be hidden for AC trigger coupling.

[ciasteczkowypotwor]

This is not a good schematic block diagram of an oscilloscope triggering, analog or digital.

[artag]

But what about the case where there is more than one zero-crossing per cycle ? Imagine something like a heartbeat with two pulses, one bigger than the other, and zero-crossings between both.

This is similar to the problem with extreme duty cycles, except that there will be a large region where AC trigger will jitter between the two zero crossings. You really want to trigger on a crossing that only occurs on the bigger peak.

You can get around that with DC triggering (but at the cost of being unable to reject a DC offset) or with trigger holdoff (at the cost of being sensitive to frequency). I'd prefer to be able to see the DC-blocked trigger waveform with the actual trigger level overlaid on it.

[David Hess]

The only thing AC trigger coupling does is remove the DC component of the trigger signal.  If you had a waveform with multiple levels which could be triggered on, then setting the trigger level when AC trigger coupled triggering used works the same was as when DC coupled triggering is used.  Set it high enough to pick up only the largest pulse edge.  This assumes however that the frequencies of interest are not filtered out as AC trigger coupling has a low pass cutoff just like AC input coupling.

The thing I find interesting about how AC trigger coupling fails on the Rigol oscilloscopes is that low frequency reject coupling is just AC trigger coupling with a higher cutoff frequency.  How can one work but not the other?

[Smoking]

Some people have requested a video of the trigger Siglent SDS2000.

Here is a youtube video showing there is no Jitter problem :

[Teneyes]

If one wished to Auto search the input for a signal which mode of these 4  would one desire most?
1.  DC input coupling.     With.     DC.  Trigger coupling
2.  DC input coupling.     With.     AC.  Trigger coupling

3.  AC input coupling.     With.     DC.  Trigger coupling
4.  AC input coupling.     With.     AC.  Trigger coupling

What does your DSO. Do?

[rs20]

If one wished to Auto search the input for a signal which mode of these 4  would one desire most?
1.  DC input coupling.     With.     DC.  Trigger coupling
2.  DC input coupling.     With.     AC.  Trigger coupling

3.  AC input coupling.     With.     DC.  Trigger coupling
4.  AC input coupling.     With.     AC.  Trigger coupling

What does your DSO. Do?

Are you referring to the "Auto" button that automatically sets up the scope's parameters based on the signal present at the scope's inputs? DC input coupling and AC input coupling are among the settings reset by the "Auto" button, so I claim your question makes no sense. The state of those settings before you press the "Auto" button has no impact on what happens whatsoever, they just get wiped out. We could speculate as to what the internal implementation of the "Auto" button actually does while it's going through its Auto routine, but I don't think that's particularly useful. In short, please clarify your question.

However, I will also mention that contrary to the video, most DSOs have absolutely no difference in behavior between your #3 and #4 cases, since the trigger signal is branched off after the input coupling; if the signal is already AC coupled at the input, AC coupling it again at the trigger does virtually nothing. Hence why most DSOs default to "DC trigger coupling", which could be more accurately named as "triggered off the signal you see on the screen".

[Teneyes]

We could speculate as to what the internal implementation of the "Auto" button actually does while it's going through its Auto routine, but I don't think that's particularly useful. In short, please clarify your question.

To explain:
In a Rigol Beta FW a surprise is that the Auto button is somewhat configurable.
See New Menu. pix below

However, I will also mention that contrary to the video, most DSOs have absolutely no difference in behavior between your #3 and #4 cases, since the trigger signal is branched off after the input coupling; if the signal is already AC coupled at the input, AC coupling it again at the trigger does virtually nothing. Hence why most DSOs default to "DC trigger coupling", which could be more accurately named as "triggered off the signal you see on the screen".

Yes , I agree #3 & #4 are the same .
For the Rigol DS2000 the trigger level line is shown in AC input w/DC Trigger mode
                    But the trigger level line is NOT shown in AC input w/AC Trigger mode
see pic. for DC Trigger, a way to indicate Trigger Coupling Mode


             

[David Hess]

However, I will also mention that contrary to the video, most DSOs have absolutely no difference in behavior between your #3 and #4 cases, since the trigger signal is branched off after the input coupling; if the signal is already AC coupled at the input, AC coupling it again at the trigger does virtually nothing. Hence why most DSOs default to "DC trigger coupling", which could be more accurately named as "triggered off the signal you see on the screen".

There is more to it because if they were triggering off of the literal signal seen on the screen with DC trigger coupling, then changing the vertical position might be expected to change the trigger position; oscilloscopes where this happens support AC triggering or something more sophisticated to avoid this behavior which otherwise looks rather peculiar.

Modern DSOs use digital triggering so the vertical position adjustment can easily take place after the point where the trigger is processed however DSOs which support offset adjustments have this problem; the DC offset will change the trigger point unless AC coupled triggering is used.

Some oscilloscopes combine the vertical position and offset adjustments into one control since the later can do things that the former cannot.  Rigol apparently lacks an offset control despite their misleading documentation implying that they do have one since adjusting the vertical position has no effect on the trigger point when DC coupled triggering is used.

[rs20]

However, I will also mention that contrary to the video, most DSOs have absolutely no difference in behavior between your #3 and #4 cases, since the trigger signal is branched off after the input coupling; if the signal is already AC coupled at the input, AC coupling it again at the trigger does virtually nothing. Hence why most DSOs default to "DC trigger coupling", which could be more accurately named as "triggered off the signal you see on the screen".

There is more to it because if they were triggering off of the literal signal seen on the screen with DC trigger coupling, then changing the vertical position might be expected to change the trigger position; oscilloscopes where this happens support AC triggering or something more sophisticated to avoid this behavior which otherwise looks rather peculiar.

Modern DSOs use digital triggering so the vertical position adjustment can easily take place after the point where the trigger is processed however DSOs which support offset adjustments have this problem; the DC offset will change the trigger point unless AC coupled triggering is used.

Some oscilloscopes combine the vertical position and offset adjustments into one control since the later can do things that the former cannot.  Rigol apparently lacks an offset control despite their misleading documentation implying that they do have one since adjusting the vertical position has no effect on the trigger point when DC coupled triggering is used.

This is hard to word well, so bear with me. Isn't it possible that the software just looks at the vertical position control, commands the offset circuitry accordingly (this must happen surely, or else you'd need more than 8 bits of ADC to get 8 bits of output), and then it calculates which digital value corresponds to the chosen trigger level and sends that to the digital trigger circuitry? That way, a change in vertical position will change the both the digital trigger level, and the values coming in off the ADC in exactly the same way, so that it all cancels out and you end up triggering at the same points in time.

FWIW, I was just playing with my DS2202 and moving the vertical position causes the trigger level indicator to follow on-screen, staying on the same voltage-at-the-input, as you would hope. However, if you thusly move it far enough off-screen, it clamps and you can see the trigger level voltage (also shown numerically on screen) changing as it reaches values outside the range of the ADC. Specifically, at 20mV/div, and with the vertical control centered (such that the 8 vertical divs on screen cover the +/- 80mV range), the trigger may be set anywhere from -100 mV to +100mV, as if the ADC range actually covers 10 divs. With the vertical on-screen range extending from -120mV to 40mV, the available trigger settings are -140mV to 60mV. Always an extra div available at the top and bottom of the range. So that seems to confirm that the trigger level selection is converted by the software into a corresponding digital (or at least, post attenuation+offset in the analog domain) value, and used accordingly. This seems like ideal behaviour to me, modulo the limitations in the range of available trigger level voltages inherent in doing triggering post-ADC.

BTW, you're right that my wording was perhaps deceiving -- all I meant by "what you see on the screen" is just referring to the idea that the trigger point can be rendered meaningfully as a horizontal line on-screen w.r.t. to the trace.

[Bud]

This is not a good schematic block diagram of an oscilloscope triggering, analog or digital.

In which case can you explain how  you can have a AC Trigger (beside external) on a DC coupled channel.

[rs20]

In which case can you explain how  you can have a AC Trigger (beside external) on a DC coupled channel.

Hehe, good catch. But the answer basically is, "in that 'trigger amp and LF/HF filter' section". I do agree with you, in that a more conceptually useful model would have the AC coupling cap switch after the trigger selection switch; this would represent how the scope behaves . However, in many digital scopes with digital trigger, it's well-and-truly digital by this stage, so it's just another option on that DSP block (basically it's a certain kind of digital highpass filter, after all).

[David Hess]

However, I will also mention that contrary to the video, most DSOs have absolutely no difference in behavior between your #3 and #4 cases, since the trigger signal is branched off after the input coupling; if the signal is already AC coupled at the input, AC coupling it again at the trigger does virtually nothing. Hence why most DSOs default to "DC trigger coupling", which could be more accurately named as "triggered off the signal you see on the screen".

There is more to it because if they were triggering off of the literal signal seen on the screen with DC trigger coupling, then changing the vertical position might be expected to change the trigger position; oscilloscopes where this happens support AC triggering or something more sophisticated to avoid this behavior which otherwise looks rather peculiar.

Modern DSOs use digital triggering so the vertical position adjustment can easily take place after the point where the trigger is processed however DSOs which support offset adjustments have this problem; the DC offset will change the trigger point unless AC coupled triggering is used.

Some oscilloscopes combine the vertical position and offset adjustments into one control since the later can do things that the former cannot.  Rigol apparently lacks an offset control despite their misleading documentation implying that they do have one since adjusting the vertical position has no effect on the trigger point when DC coupled triggering is used.

This is hard to word well, so bear with me. Isn't it possible that the software just looks at the vertical position control, commands the offset circuitry accordingly (this must happen surely, or else you'd need more than 8 bits of ADC to get 8 bits of output), and then it calculates which digital value corresponds to the chosen trigger level and sends that to the digital trigger circuitry? That way, a change in vertical position will change the both the digital trigger level, and the values coming in off the ADC in exactly the same way, so that it all cancels out and you end up triggering at the same points in time.

This is certainly possible and even likely.

FWIW, I was just playing with my DS2202 and moving the vertical position causes the trigger level indicator to follow on-screen, staying on the same voltage-at-the-input, as you would hope. However, if you thusly move it far enough off-screen, it clamps and you can see the trigger level voltage (also shown numerically on screen) changing as it reaches values outside the range of the ADC. Specifically, at 20mV/div, and with the vertical control centered (such that the 8 vertical divs on screen cover the +/- 80mV range), the trigger may be set anywhere from -100 mV to +100mV, as if the ADC range actually covers 10 divs. With the vertical on-screen range extending from -120mV to 40mV, the available trigger settings are -140mV to 60mV. Always an extra div available at the top and bottom of the range. So that seems to confirm that the trigger level selection is converted by the software into a corresponding digital (or at least, post attenuation+offset in the analog domain) value, and used accordingly. This seems like ideal behaviour to me, modulo the limitations in the range of available trigger level voltages inherent in doing triggering post-ADC.

When DSOs used analog triggers it was always possible to do the trigger pickoff before any offset.  With digital triggers that is not possible so if the position control acts as an offset at the input to the digitizer which is a desirable feature, then the trigger level will have to be made to track to prevent it from changing with position changes.

BTW, you're right that my wording was perhaps deceiving -- all I meant by "what you see on the screen" is just referring to the idea that the trigger point can be rendered meaningfully as a horizontal line on-screen w.r.t. to the trace.

I only brought it up because I have run into this before on oscilloscopes.  Where it gets complicated is if the DSO has separate position and offset controls.  The former should not affect the trigger level but the later might.  None of my current DSOs have both and I do not remember how the ones I have used which did handled it.  I think they compensated for both so the trigger level did not change.

On my older high performance analog oscilloscopes, the vertical position controls either act by generating an offset or they also have dedicated offset controls.  In both cases they affect the trigger level but by default on these oscilloscopes, the trigger is either AC coupled or uses peak-to-peak triggering so any DC offset if removed.  If DC trigger coupling is used, then the trigger level becomes a fixed point on the display.

[David Hess]

In which case can you explain how  you can have a AC Trigger (beside external) on a DC coupled channel.

Hehe, good catch. But the answer basically is, "in that 'trigger amp and LF/HF filter' section". I do agree with you, in that a more conceptually useful model would have the AC coupling cap switch after the trigger selection switch; this would represent how the scope behaves . However, in many digital scopes with digital trigger, it's well-and-truly digital by this stage, so it's just another option on that DSP block (basically it's a certain kind of digital highpass filter, after all).

The way I like to think of it is AC trigger coupling is same same as LF reject coupling but with a lower cutoff frequency; LF reject coupling is a type of AC coupling.  At least in all of the oscilloscopes I have used, AC trigger coupling extends below 60 Hz so it may be used with AC line sources while LF reject coupling rejects AC line sources.