Keysight 4000 X & 6000 X 2000 X 3000 X

[lem_ix]

Was working on a pulse oximeter(same frequency), and ofc used roll off mode on my hameg to display the signal. It's not a bug, rather the way the tool works. I like that it shows stuff even when I'm not triggering so I know something is there. To each their own i guess?

[Howardlong]

I don't think this is a  bug either,  this seems to me to basic operator error.
Just because you turn on Auto trigger doesn't mean you don't have to setup trigger levels or holdoff if you want to display a signal clearly and repeatably. Trigger level and holdoff (along with more advanced trigger settings) settings work the same in Auto as they do in Normal.
Keysight scopes just seem to have very fast rearming after an auto trigger, that's a good thing if your hunting around for a unknown signal.

Thats not the point.  The point is, that the scope is NOT triggering at all in Auto mode with non-repetitive signals. Even when there is only one short signal in 10 seconds: the scope will NOT trigger. I know what to do with holdoff and of course trigger level etc. are set.
But here there is a scope in Auto-Trigger mode, waiting a second or longer for a signal. Then a signal appears and .... the scope will not trigger on it. So you miss it.

The Keysight _will_ trigger if the timebase is appropriately slow for a given signal.

But what you are describing is what Normal mode is for.

The Tek, by the way, does not trigger at all at 40ms/div and slower, it goes into untriggered Roll mode automatically on Auto. You have to switch the trigger to Normal to allow triggering (errr!). Here is a 6.3Hz 9.7ms pulse, the Keysight triggers, the Tek does not.

[D3f1ant]

Because its early and I had just enough coffee to be bothered I am trying to replicate. I have a sig gen putting a 1us pulse, 1Hz. I set scope to Auto trigger, scope displays the signal intermittently, so I can immediately see there is something there, I twirl the holdoff knob for stability and bingo I have a perfectly centered blip (or pretty trace) depending on where I have timebase. If I take the signal away the trace goes back to flat line. Where is the problem?
I'm not sure how your setting up your scope that you don't see the intermittent pulse at least somewhere on the screen. If your timebase is set extremely slow and you have your scope set to hi-res or some other averaging/filtering/bw limit turned on maybe it wouldn't display at least 'something' but thats still a setup problem. I would think you would/should normally have a fair idea of what a signals parameters should be and setup accordingly.

[Keysight DanielBogdanoff]

Hi everyone, I spent the afternoon getting down to the nitty gritty.  I love my job!

Essentially, it comes down to an ideology that we hold here at Keysight scopes.  That is: “Don’t compromise the waveform update rate!”

Keysight auto trigger mode works like this:

0. Assume a stable trigger was found
1. Wait for another valid trigger event
     a.wait time = maximum of (4 * total time on screen) or 50 µs
             i.note: on the 2000 and 3000A scopes, the wait time = maximum of (1/2 * total time on screen) or 50 µs.  This is what we changed in firmware.
     b. If a valid trigger event is found before (a) has passed, repeat (1)
     c. If no event found, go to (2)
2. Enter “autotrigger” (or free-run) mode: acquire and display signals as fast as possible
     a. The “as fast as possible” will be the fastest update rate available at that time/div setting
3. While doing (2), check trigger circuitry every 250ms to see if there was a valid trigger event in the past 250 ms
     a. If there was a valid event, go to (1)
     b. If there wasn’t a valid event, go to (2)

So here’s what is happening.  For the 1 Hz cardiac pulse, because the pulses are so far apart and only happening once, the scope essentially stays in auto trigger mode (2). E.g. if there was a pulse in the previous 250ms period then it will wait for another pulse, but will time out before the required 750ms has elapsed unless the time/div setting is high enough (see 1). For example, put the cardiac pulse on and set the time/div to 25ms.  Because the screen is then showing 250 ms of time, the wait period will be 4*250 ms = 1s and you will see the regular trigger. 

What Tek and some other vendors do is build in a wait time after every single capture.  I think Tek has an initial wait time of 4*time on screen, the switches to 2*time on screen after each subsequent trigger event. We spend the majority of the free-run time (2) acquiring and plotting instead of waiting.

The 3000T, 4000 X-Series, and 6000 X-Series scopes all follow the 1a guidelines, while the older 2000 X-Series and 3000A scopes keep the 1ai scheme.  Ultimately, I don’t see this as a bug. I’m not just saying that because I work for Keysight, but as I’ve looked at this closer it seems to be more of a difference in ideologies about how auto trigger mode should work and how auto trigger mode is implemented, especially in the free-run portion of the state machine. 

Another thought I had here is that our built-in DVM could run on the side and let you know immediately if the pulse stops because the frequency will change.  Our DVM will work on the same channel you are probing, too, so no extra probing is necessary.

I hope this makes sense, I’d be happy to discuss further or clarify if I wasn’t clear enough.

[lem_ix]

Great explanation! Not sure if a case of RTFM? Include such information if it's not in there already. Normally you would find a solution the moment you encounter such a signal but it can't hurt to point it out 

[Keysight DanielBogdanoff]

I don't think it's an RTFM (read the factory manual) situation, it's sort of a obscure spec/feature/characteristic.  I definitely see how it could be confusing or look like a bug.

[tautech]

I don't think it's an RTFM (read the factory manual) situation, it's sort of a obscure spec/feature/characteristic.  I definitely see how it could be confusing or look like a bug.

Nearly Daniel, it's read the f**king manual. 
Good to see the trouble you've gone to in order to check the OP's problems out. 

[Sbampato12]

It is very great to have a employee that use the same forum you use! 

Looking foward, sure it is more a characteristic than a bug. And in comparison as some users posted here, I think that this is the most interesting approach.

Anyway, this was good to learn a little bit more on triggers!

[Dubbie]

After upgrading from my Rigol to a Keysight 3000T, this one caught me by surprise. I couldn't figure out why I suddenly couldn't trigger at all on sparse (10ms) signals. On auto they were just jittering around all over the screen. The manual wasn't very helpful but this thread cleared things up for me. Putting a long holdoff made it settle down.

[LabSpokane]

And this is why I will continue to buy from Keysight. 

[taemun]

Isn't the real KS deficiency here that the scope can't detect a trigger mid-acquisition, and move the trigger-point to the appropriate place on the x-axis?

If it did that, wouldn't everybody be happy?


Disclaimer: happy MSOX3054A owner over here 

[Dubbie]

I would be happy if this was how it worked.


Sent from my phone using Tapatalk

[HighVoltage]

On slow signals I just switch from AUTO to MANUAL trigger mode and it works fine on all my Keysight scopes.

[taemun]

On slow signals I just switch from AUTO to MANUAL trigger mode and it works fine on all my Keysight scopes.

You're right, 'Normal' mode triggers only when it should. However, that doesn't really excuse the operation of the 'Auto' mode.

I'm wading into a religious debate, but I think that the auto mode could be improved to meet everyone's expectations if the scope could detect and act on a trigger mid-acquisition, rather than just blindly running over it.