EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: onesixright on June 15, 2016, 09:45:59 pm
-
Probably a silly question :-[, but what is actually the definition of a "Waveform" (Capture Rate) / wfms ?
For the sake of the argument, lets say a scope has 1 wfms/s, would that indicate it can capture 1 "graph" (that what you see on the screen) per second? or would that be 1 data point (voltage / time) capture per second?
Thanks.
-
Probably a silly question :-[, but what is actually the definition of a "Waveform" (Capture Rate) / wfms ?
For the sake of the argument, lets say a scope has 1 wfms/s, would that indicate it can capture 1 "graph" (that what you see on the screen) per second? or would that be 1 data point (voltage / time) capture per second?
Thanks.
Neither.
Think of it as the # of snapshots of the entire waveform /s sent to the buffer memory depth available, one of which is then displayed.
DSO's that have History functionality allow the user to then scroll through those snapshots after waveform acquisition is stopped.
-
Think of it as the # of snapshots of the entire waveform /s sent to the buffer memory depth available, one of which is then displayed.
Ah, so you could say that 1 waveform contains a "full" set of datapoints? Number of datapoints depends of-course on the timebase, sampling, etc.
-
For the sake of the argument, lets say a scope has 1 wfms/s, would that indicate it can capture 1 "graph" (that what you see on the screen) per second?
This. The "waveform" in this case is considered to be one capture & plot. If you have a waveform update rate that is faster than the screen's update rate, the scope will plot multiple captures (or "waveforms") at once.
I made a short video on this a few weeks back (yay blogspam :)) :
https://www.youtube.com/watch?v=CPDIrKSDrZk (https://www.youtube.com/watch?v=CPDIrKSDrZk)
-
Think of it as the # of snapshots of the entire waveform /s sent to the buffer memory depth available, one of which is then displayed.
Ah, so you could say that 1 waveform contains a "full" set of datapoints? Number of datapoints depends of-course on the timebase, sampling, etc.
Yeah. Full meaning the memory depth specified by the scope. Number of data points generally depends on capture memory depth = sample rate x timebase. There can be exceptions in certain modes, and you can often tell the scope to export less points.
-
Also check the other thread with some interesting considerations:
https://www.eevblog.com/forum/testgear/some-important-points-to-remember-when-evaluating-waveform-update-rates/ (https://www.eevblog.com/forum/testgear/some-important-points-to-remember-when-evaluating-waveform-update-rates/)
-
Think of it as the # of snapshots of the entire waveform /s sent to the buffer memory depth available, one of which is then displayed.
Ah, so you could say that 1 waveform contains a "full" set of datapoints? Number of datapoints depends of-course on the timebase, sampling, etc.
Yep, but it's not any simple relationship between any of the HW and can/does change with FW versions.
Only those DSO that have been fully inspected for performance on each timebase settings will give you the full picture of what's going on and at what settings the manufacturer has chosen to optimise the wfm/s rate.
See rf_loop's table:
https://www.eevblog.com/forum/testgear/siglent-sds1000x-series-oscilloscopes/msg823523/#msg823523 (https://www.eevblog.com/forum/testgear/siglent-sds1000x-series-oscilloscopes/msg823523/#msg823523)
-
Still it is very important to remember that to capture a glitch with 100% certainty you'll need infinite waveforms/s. Since that is not going to happen you'll have to realise that:
a) having many waveforms/s makes it likely you spot a problem by chance but doesn't guarantee it
b) you have to setup a dedicated measurement using a trigger conditions to make 100% sure a signal doesn't misbehave (*)
Now some oscilloscope manufacturers went wild (race to the bottom) over having lots of waveforms/s and even skimped on more usefull things like memory depth to keep pricing competitive and/or keep the waveforms/s rates high. Longer memory takes more time to fill so the waveforms/s rate drops with longer memory.
* A trigger condition is like putting a sliding window over a signal while testing every piece of it and therefore captures an anomaly with 100% certainty.