Simple Technique to measure Waveform Update Rates: DSOs w/either Edge Triggering

[marmad]

I mean the trigger sweep mode=normal (not auto or single).

4 Hz is stable.

Yes, I 'doh'ed that shortly after I wrote it    then erased the comment 

Yes, Trigger level at 0V

Well, either the DS1000 has different logic depending on the rise time speed - or the rise time speed is affecting the normal logic - but it seems as if it's working with the sine wave, so just run your tests (only using sine) and see what you get. 

[alank2]

1ms timebase

7-12 Hz - stable
13-20 Hz - unstable
21-24 Hz - stable
25-33 Hz - unstable
34-35 Hz - stable

So does this mean 20 waveform updates per second???

[marmad]

1ms timebase

7-12 Hz - stable
13-20 Hz - unstable
21-24 Hz - stable
25-33 Hz - unstable
34-35 Hz - stable

So does this mean 20 waveform updates per second???

Hey - did you watch the video?    Seriously, though - the maximum waveform per second rate is always the frequency right before (or just as) you exit the FIRST stable period. In this case 12-13Hz - which matches exactly with Harv's measurement of the Trigger Out. No need to keep scanning after you've had the first blind time (only a single-edge visible - or as you call it, stable).

[alank2]

Hey - did you watch the video?    Seriously, though - the maximum waveform per second rate is always the frequency right before (or just as) you exit the FIRST stable period. In this case 12-13Hz - which matches exactly with Harv's measurement of the Trigger Out.

    I did, but I had distractions while watching it, or at least that is my excuse!

So odd that it doesn't work properly with square waves...

[marmad]

  I did, but I had distractions while watching it, or at least that is my excuse!

   I'm guilty of the same often myself.

[marmad]

So odd that it doesn't work properly with square waves...

It might be, as I mentioned before, some kind of decision by Rigol's engineers to assume a user might want to see both edges when there is a fast rise/fall time - and so they use 'alternating' Edge logic then. Actually, it's not such a bad idea if it's true: you would get the fastest wfrm/s rate with slow rise times ('either') - but the benefit of seeing both edges with fast ('alternating').

EDIT: In any case, I've re-edited the first post to 'sine wave' instead of 'square wave', in case other Rigol owners try the test.

[marmad]

Can anyone confirm Marmad's method doesn't work on a DS1000 series scope with a 00.04.00 firmware?

@Ivan - have you run the test using a sine wave? It appears square waves don't work with the Rigol DS1000 series - but sine waves do.

At 2ms/div, the wfrm/s rate of the Rigol is 5.62 - which means the acquisition cycle time is ~178ms - with a blind time of ~154ms. That means if you give the DSO a sine wave of 4Hz (250ms period = 125ms edge to edge), it shouldn't trigger on the opposite edge (it will definitely fall in the blind time) - so if you see both edges, it's doing 'alternating' Edge triggering - if it remains locked on the rising (or falling) Edge, it's doing 'either'.

Test it with these settings:

2ms/div
Normal mem depth
Edge trigger -> both slopes
Normal sweep
4Hz sine wave

[Teneyes]

4 Hz sinewave - rising only (stable)
4 Hz triangle - rising only (stable)
4 Hz square - alternating rising and falling (unstable)

  With Trigger set at 0 Volts, could the Square wave be  0-> x Volts,  and Not +  - ?

[marmad]

  With Trigger set at 0 Volts, could the Square wave be  0-> x Volts,  and Not +  - ?

Good point - I didn't think to ask. But as long as a sine wave works for him, it doesn't really matter.

[alank2]

Hi,

  With Trigger set at 0 Volts, could the Square wave be  0-> x Volts,  and Not +  - ?

I tried it both ways.  I have a TTL output on my AWG (it is a cheap one off ebay) and I set the trigger for the middle, and it also has an output that produces a balanced square wave and I trigger it again in the middle (0V).

[darrylp]

Hi,

  With Trigger set at 0 Volts, could the Square wave be  0-> x Volts,  and Not +  - ?

I tried it both ways.  I have a TTL output on my AWG (it is a cheap one off ebay) and I set the trigger for the middle, and it also has an output that produces a balanced square wave and I trigger it again in the middle (0V).

you have the setup on the trigger on 0.1 div, and minumum holdoff of 500ns (  these are the lowest values that my 2.4sp1 firmware allow. 

its working for me...  I get slightly lower figures in these 5us to 10ms range than Harv, but its similar.  mine is hardware 58 ( from memory when I was last inside when I replaced the cooling fan with larger ( quieter ) one )

[ivan747]

I got Marmad's original technique to work on my oscilloscope now. Thanks for the hint Marmad! (via PM)

I was expecting a much faster update rate on the oscilloscope. I started on a 10ms timebase. Apparently not even 1Hz is slow enough for that. I got it to work on 4Hz at the 2ms timebase. But I can't still find anything else on the scope, so:

Maybe the trigger is not the same as the DS2000 series or I'm doing something wrong.

I was doing something wrong. Sorry for the confusion.

I'll grab the wfrm/s table someone posted earlier and try to confirm it is the same in this firmware. Also I want to try from scratch using frequency sweeps.

[ivan747]

Still, can someone try this method on the "dodgy" LeCroy and Atten oscilloscopes that skip edges on their triggering?

It's the following:
*Get 2 frequencies, one called F1, the other F2. Try square and sine waves, whatever works best.
*F2 is twice as fast as F1 and these are in phase.
*Set your trigger to the Ext. trigger channel, on the rising edge, normal trigger, not auto.
*Connect F2 to the Ext. trigger channel.
*Connect F1 to Ch1 (leave Ch2 off).
*Start looking for the same phenomenon described in Marmad's video, everything should work the same in that regards.

I tried this with the same settings as my last post, obviously with the exception of the points I just mentioned. It displays only a single edge, so it works more or less the same. Granted, if your blind time changes because of the external trigger, that can affect your measurements, but I don't see why the blind time could increase with external trigger.

[marmad]

I was expecting a much faster update rate on the oscilloscope. I started on a 10ms timebase. Apparently not even 1Hz is slow enough for that. I got it to work on 4Hz at the 2ms timebase. But I can't still find anything else on the scope, so:

Well, it was more my mistake for suggesting 10ms/div to you as a test point. I've now re-edited the first post to suggest people first test on 5ms/div with a sine wave frequency from 1 - 12Hz. I think this will work for most DSOs (that have 'either' Edge triggering). On the Rigol DS1000, it should produce a stable image at 1 or 2Hz.

[tinhead]

Still, can someone try this method on the "dodgy" LeCroy and Atten oscilloscopes that skip edges on their triggering?

It's the following:
*Get 2 frequencies, one called F1, the other F2. Try square and sine waves, whatever works best.
*F2 is twice as fast as F1 and these are in phase.
*Set your trigger to the Ext. trigger channel, on the rising edge, normal trigger, not auto.
*Connect F2 to the Ext. trigger channel.
*Connect F1 to Ch1 (leave Ch2 off).
*Start looking for the same phenomenon described in Marmad's video, everything should work the same in that regards.

this didn't make any sense, twice the frequency as trigger will always produce "expected" result.
If you don't believe apply 50MHz/100MHz and you will see it works as well, and i doubt your DSO is capable of actually what? 50M or 100M wfrm/s?

[ivan747]

You go from the lowest frequency to the highest. The first time you get the expected result is the waveform update rate. The same problem could happen on Mark's (marmad) method using alternate edge trigger. I don't know if we can call it aliasing, but it's sort of like that.

[Teneyes]

 Interesting to see that my Rigol DS2072 Update rate change from 46,000 wfrm/s  to 5K wfrm/s when changing
 the scan rate  from 20nsec/div to  20.1nsec/div. But understandable.

[tinhead]

Ok so I hit some luck.

Probing around in the internal headers of the DS1000E series has some very interesting signals on them.  For the purpose of this discussion though, I found the trigger output.

Harvs,

that pin could be sync to logic analyzer module as well, and not trigger out.

Another thing is, with long memory enabled you should got different values, a real trigger out should get affected by sample memory changes.

Anyway, you can do another one test (maybe you ran it already?), apply only single pulse, then two, then burst, etc and compare it with the "trigger out" you found,
if that's really trigger out then there must be no difference (on each pulse/burst you need to get signal there).

Ok so I hit some luck.

Probing around in the internal headers of the DS1000E series has some very interesting signals on them.  For the purpose of this discussion though, I found the trigger output.

Nice find!! I'm rather amazed no one has located this before. 

probably nobody was really looking for, but when you check attached picture you will see that Rigol is having some kind of
phase/trigger delay? circuit (i found it as well in older Rigol series, as well in ATTEN, Tekway A series, Hantek hw1005,
Siglent. Attached schematic of Hantek's version of that circuit as well, Rigol is almost the same).
That could be another place to look for trigger out.

Something else to watch (for those who checked with "edge disapears" way) is not to get fooled with display refresh sync,
Rigol DS1000 seems to have display refresh rate menu, when value changed the estimated wfms/s should not change when
the DSO-core is in FPGA only, when in DSP as well then it should change when DSP running single thread, but when more threads then
it should remains (almost, there can be some overhead) the same (as the display thread should have no influence on capture/postprocess thread).

[onlooker]

Again maybe someone already did this. Here is a better version of double pulse method using DDS3x25 with the attached waveform.

1). It work for single channel.
2). Easier to operate: It detects the flicking point by adjusting the oscilloscope's trigger level. 
3). Better at filtering out false triggering.

For details see the picture.

On the scope, move Trig level up and down to find the level where flickering starts showing up. At this point the green time interval defined the waveform update rate. The scope should be on DC coupling.

At 10Hz run of this waveform on FG, the change of scope trig level covers the update rate from about 50Hz to 1kHz. The range can be changed by changing FG running freq (e.g 5Hz run covers wfm upd rate from 25Hz to 500Hz).

The long flat part of the waveform is about 80% of total. It is the way to overcome/reduce the aliasing effect (false trig at wrong FG running freq) .


 

[marmad]

1). It work for single channel.
2). Easier to operate: It detects the flicking point by adjusting the oscilloscope's trigger level. 
3). Better at filtering out false triggering.

At 10Hz run of this waveform on FG, the change of scope trig level covers the update rate from about 50Hz to 1kHz. The range can be changed by changing FG running freq (e.g 5Hz run covers wfm upd rate from 25Hz to 500Hz).

What DSO did you test it on?

[onlooker]

I tried a few settings on Owon SDS7102 (bought after watching your videos and before you changed your opinion).

I am more interested in the discussion about methods, than detailed numbers. Besides, I remember you had already listed the numbers from trig-out a while back.

[marmad]

I tried a few settings on Owon SDS7102 (bought after watching your videos and before you changed your opinion).

I am more interested in the discussion about methods, than detailed numbers. Besides, I remember you had already listed the numbers from trig-out a while back.

I was just trying to understand whether you actually tested the claims that you made; e.g. "... the change of scope trig level covers the update rate from about 50Hz to 1kHz." But in any case, it's an interesting idea - unfortunately, many low cost DSO owners don't have AWGs - hell, some of them have had to dig around to come up with a method for sweeping a simple sine wave to test my method. 

Anyway, I'm off to bed, but I did some quick testing. It seems to be reasonably accurate with my DSO update rate until the 500us/div setting (~140 wfrm/s) - then starts losing accuracy more and more at each smaller time base setting. I only tested to 20us/div, at which point it was off by ~20% - although I'm sure it's probably possible to increase the accuracy with further tweaking of the sample period/frequency.

I used the double pulse method for checking wfrm/s with the DDS-3X25 quite some time ago - and the Hantek software just becomes excruciating to work with after awhile.

[Harvs]

that pin could be sync to logic analyzer module as well, and not trigger out.

Yeah I imagine that's it's actual purpose.  There's a few signals on that header that seem to correspond to certain events in the sampling process.

Another thing is, with long memory enabled you should got different values, a real trigger out should get affected by sample memory changes.

How should it get affected?  I don't understand, I thought a trigger out just has an edge that corresponds to "time zero" on the scope?  How does that change with long memory depth?

Anyway, you can do another one test (maybe you ran it already?), apply only single pulse, then two, then burst, etc and compare it with the "trigger out" you found, if that's really trigger out then there must be no difference (on each pulse/burst you need to get signal there).

Sorry I don't understand what a burst of pulses is likely to do?  In all the test previous I've feed in a 30MHz square wave and it only outputs when the scope triggers (this is in the screen shots earlier), is this what you mean?

[marmad]

How should it get affected?  I don't understand, I thought a trigger out just has an edge that corresponds to "time zero" on the scope?  How does that change with long memory depth?

The maximum number of triggers a scope can output at any given point is the inverse of the acquisition cycle time. The acquisition cycle time includes the acquisition time - which, as I wrote in my last post to you, is always the longer of either (time base * divs) OR (sample length / sample rate). With long memory depths, it's almost always the latter - so the waveforms per second (and the number of trigger out pulses) would then decrease.  Now please don't make me write this again 

Edit: BTW, almost EVERYTHING you can do on a DSO affects the wfrm/s rate (and thus the Trigger out): persistence, measurements, extra channels, having a menu out, etc. because they change the variable blind time (i.e. the DSO is doing more processing).

[marmad]

BTW, blind times for these rates are (approximately):

2ms	154ms
1ms	68ms
500us	38ms
200us	30ms
100us	24ms
50us	21ms
20us	18ms
10us	18ms
5us	17ms