EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: Mat on February 18, 2016, 10:39:14 am
-
Hi guys,
I'm trying to find out how much time an oscilloscope set in AUTO mode waits for a valid trigger condition before forcing a sweep.
The closest to the answer I found was in this Tektronix manual on page 12 (page 3 of the pdf) : http://clifton.mech.northwestern.edu/~me224/Documents_oscilloscope_20tutorial.PDF (http://clifton.mech.northwestern.edu/~me224/Documents_oscilloscope_20tutorial.PDF)
It states : "If no trigger condition occurs while the oscilloscope waits for a specific period (as determined by the time-base setting), it will force itself to trigger."
How much is "a specific period". Is it the duration of one division or the entire width of the displayed data (10 divisions) or something else?
I know it can be different between models, but a general commonly used value in the industry would be interesting to have.
Thanks,
Mat
-
I think it's "as long as it takes the internal memory to fill up".
ie. Your memory depth divided by your sample rate.
-
On the Rigol digital scopes, it is about 0.5 seconds. It varies between models, and analog vs digital...
-
I recall seeing that Tektronix' AUTO mode always looks terribly slow, because they have a particularly long timeout around 100ms. But as soon as a trigger pops up, there's your million-odd waveforms.
Back in the analog days, it might've been sweep dependent (say, 10 sweeps worth?), or fixed (100ms?). I'm not sure. It doesn't seem to be sweep dependent on the lowest ranges anyway, as AUTO is useless below 100ms/div or so.
Tim
-
On the Rigol digital scopes, it is about 0.5 seconds. It varies between models, and analog vs digital...
no its not... btw...
https://www.youtube.com/watch?v=ta096oBzSac (https://www.youtube.com/watch?v=ta096oBzSac)
-
hmm, the AUTO mode wait time isn't the same thing as trigger hold-off.
In trigger hold-off, the trigger is not armed and there is no way a sweep can occur.
After the hold-off period, the trigger is armed and the scope waits for a trigger event to happen.
However, in AUTO mode, if nothing happens, it timeouts eventually and still fires a sweep.
-
The easiest way to check this is to send the trigger out signal to a frequency counter or another scope. A lot of this down time will depend on the scope's architecture and settings. Longer memory or more analysis (decoding, measurements, stats, etc.) can potentially impact this time.
On my Keysight 2000 X-Series I get up to about 55 kHz trigger rate.
On my Keysight 4000 X-Series, I'm topping out at about 1.3 MHz.
Essentially, in auto mode we're running right around the spec'd waveform update rate.
Hold-off is different. It's a way to specify how long the scope should wait between valid triggers (at least on Keysight scopes). If I set the holdoff to 10s and hit run, it waits 10s until the next trigger but then goes back into the auto mode. So, setting a hold-off will only be effective when getting a valid trigger.
-
I'm also realizing now that this rate will probably be different depending on when valid triggers are found. I'm not sure specifically what the algorithm is after a valid trigger.
It looks like both the 2000 X-Series and 4000 X-Series wait about 50ms after a valid trigger.
-
I recall seeing that Tektronix' AUTO mode always looks terribly slow, because they have a particularly long timeout around 100ms. But as soon as a trigger pops up, there's your million-odd waveforms.
Back in the analog days, it might've been sweep dependent (say, 10 sweeps worth?), or fixed (100ms?). I'm not sure. It doesn't seem to be sweep dependent on the lowest ranges anyway, as AUTO is useless below 100ms/div or so.
Tim
As far as I can determine,"AUTO" on an analog "free runs" slightly faster than its rate when triggered,at least in the case of my Tektronix 7613.
The 7613 makes both Gate & Sawtooth waveforms available on the back panel.
I monitored the sawtooth,using a little Digitech single channel analog 'scope.
With the 7B53A set to "AUTO",& no vertical input:-
Varying the time/cm over its range caused the monitored waveform frequency to change in step with the settings.
With a 500kHz signal input to Channel A of the 7A18N,the 7B53A set to 0.2 us/div,was triggered for a stable display of two cycles:-
The waveform displayed on the Digitech was 4.1 us in duration,overall,with the ramp component 2.6us.
The 7B53A was now de-triggered,& the 7613’s horizontal sawtooth was again monitored:-
The waveform was now 3.2us in duration,with the ramp portion 2.7us.
The time/div on the Digitech may be inaccurate,but it shows the difference between triggered & untriggered in AUTO.
-
Further to the above,it makes sense to have a longer delay in DSOs,----as Fungus put it:- "as long as it takes the internal memory to fill up".
There is nothing really to show until then,anyway.
It also made sense to revert the "AUTO" timebase to a lower frequency in the earlier Tek DSOs.
Dave has a video on that as well.
There is no advantage in a long delay or lower free run frequency in analogs,so it isn't done.
-
Further to the above,it makes sense to have a longer delay in DSOs,----as Fungus put it:- "as long as it takes the internal memory to fill up".
There is nothing really to show until then,anyway.
Let's do a quick test.
On my DS1054Z I set the time base to 5ms. The 'scope chose a sample rate 125MSa/s and a memory size of 7.5Mpts.
Looking at the 'scope; I can just about count the screen refreshes visually. I counted 50 refreshes and my Casio Mudman says that took 10.8 seconds.
Math
Each refresh is 0.216 seconds (approx).
At 125MSa/s the memory will fill up in 7.5/125 = 0.06 seconds.
Result
It's about a quarter of the speed it should be. My theory is busted. :P
Edit: If you crank up the time base the screen refresh rate gets *much* faster.
-
On my Keysight 2000 X-Series I get up to about 55 kHz trigger rate.
On my Keysight 4000 X-Series, I'm topping out at about 1.3 MHz.
Essentially, in auto mode we're running right around the spec'd waveform update rate.
This is counter-intuitive to me. Shouldn't a scope in auto-trigger mode wait for "a while" whether it detects a valid trigger event, before displaying the next waveform? My understanding of auto-trigger mode is that the scope prefers to show triggered waveforms, and only falls back to displaying a free-running trace when no trigger is detected after some timeout?
-
On my Keysight 2000 X-Series I get up to about 55 kHz trigger rate.
On my Keysight 4000 X-Series, I'm topping out at about 1.3 MHz.
Essentially, in auto mode we're running right around the spec'd waveform update rate.
This is counter-intuitive to me. Shouldn't a scope in auto-trigger mode wait for "a while" whether it detects a valid trigger event, before displaying the next waveform? My understanding of auto-trigger mode is that the scope prefers to show triggered waveforms, and only falls back to displaying a free-running trace when no trigger is detected after some timeout?
Yes, I agree, the waveform update rate in AUTO mode with no triggered signal cannot possibly be the same than with a triggered signal, the difference being the timeout which I'm trying to find more info on.
-
This is counter-intuitive to me. Shouldn't a scope in auto-trigger mode wait for "a while" whether it detects a valid trigger event, before displaying the next waveform? My understanding of auto-trigger mode is that the scope prefers to show triggered waveforms, and only falls back to displaying a free-running trace when no trigger is detected after some timeout?
Yes, if there's no trigger state variable. But one could very easily add a flip-flop that says "hold on a moment, we're triggered, wait for another event" versus "I got nothin, just go and free-run that thing". Which I guess is what they do, so the trace can be relatively bright during untriggered operation.
Which implies that, when a trigger event is actually received, it could occur during an AUTO-triggered sweep, or has to propagate through an extra logic state (first trigger flips the auto-run flip-flop and temporarily puts it in NORM mode), in either case missing at least the first trigger. But this is not unwelcome for most uses of AUTO, because you just want to see if a repeating signal is present or not. If you are looking for single events, you use NORM in the first place.
I would imagine pretty much all DSOs do not exhibit this sequence, and simply display the memory around the trigger event, for every event (within the limitations of how much data can be displayed at once, which might've been a single series for the earliest DSOs, but is pretty much the full record, as a histogram, nowadays). After all, you can't get pretrigger data /at all/ unless you're always sampling at full tilt: one of the most obvious benefits of a DSO.
Tim
-
... one could very easily add a flip-flop that says "hold on a moment, we're triggered, wait for another event" versus "I got nothin, just go and free-run that thing"
This is basically it. When the scope gets a valid trigger, it says "hold on a moment." If the moment passes without another valid trigger, it says "I got nothin, just go and free-run until I get another valid trigger."
Rinse and repeat for each valid trigger.
-
More slightly "off-topic" meanderings:-
A very handy use of a "free-running" 'scope display is to show the change in value of a dc voltage which is being adjusted,or to quickly check dc supply rails whilst trouble shooting.
The only early DSOs I had a reasonable amount of "hands-on" time with,didn't seem to be able to do this,& would always attempt to trigger on a dc change.
Because they reverted to a very slow timebase speed in "AUTO",it would show a "step" somewhere along the display (often at either end) which was fairly useless.
Modern DSOs appear not to do this,& will act like an analog instrument.
Referring back to the analog case,the only time there may be a delay in commencement of a sweep in "AUTO" is when the 'scope is first switched from "NORM".
The first sweep will be delayed by however long it takes the mechanical switch to operate,& perhaps for the sweep circuitry to reach correct operating conditions.
In "AUTO",if a trigger is received,an analog operates in a similar manner to when in "NORM",except that if there is no further trigger,it reverts to a flat line.
I assume a DSO works in a similar manner.