EEVblog #617 - Tektronix Oscilloscope Anomaly

[EEVblog]

Dave shows how Tektronix digital oscilloscopes work fundamentally different in their Auto trigger mode from other brands.
And also some random ad-hock comparisons between the Tektronix MDO3000, Agilent MSOX3000, Rigol DS2000, and GW Instek GDS-2000A oscilloscopes.

[w2aew]

As you correctly stated, the Tek waveform rate is always dependent on what it gets from the trigger system.  When in auto trigger mode, and no valid trigger is found, the update rate is driven by the auto-trigger timer (about 50ms).  This is the way auto-trigger works on Tek scope, even back to the analog scope days.  The idea is to give you enough of a view of what the waveform looks like so that you can properly setup a trigger.  You can see if the waveform is sinusoidal. pulses, noise-like, etc., and that may help you make a better decision about what trigger settings to use.

I've seen instances where other scopes have trouble triggering on very infrequent events (like <1/sec) when in auto-trigger mode.  I presume that this is due to whatever mechanism they use to switch between fast-freerun and triggered mode.  I have never seen something missed with the Tek under same conditions, presumably due to the fact that the acquisition rate is always driven from the trigger circuits.

Note that on the Tek, hitting Run/Stop will pause the scope in it's tracks - thus retaining the display and all the detail in it's persistence memory.  If you want a single trace, you would use the Single button.

[retrolefty]

Digital scopes, blah, where some software geeks get to determine how your scope works. 

[taemun]

Note that on the Tek, hitting Run/Stop will pause the scope in it's tracks - thus retaining the display and all the detail in it's persistence memory.  If you want a single trace, you would use the Single button.

Will the Single button do that on an untriggered auto (free-run) signal? Won't it wait for a trigger event?

FWIW, a Tek in Auto will trigger as intended on a 300us wide 10MHz burst that happens a few times per second, whereas an Agilent (MSO6104A, but similar with the newer beasties) will free-run inbetween the bursts (you need to select Normal mode).

@Dave This is something else you could test if you wanted - how long it takes for the scope to go back to auto-run after a trigger event. On the Agilents, it varies with timebase.

Given the number of settings in modern scopes, rather than making decisions for us, scope manufacturers could just give us the option - how fast do you want auto-run to fire?

[David Hess]

I prefer how the Tektronix operates because it is apparent from the untriggered AUTO display what type of waveform is present.  If I wanted to change that behavior, I could connect the external trigger to some high frequency asynchronous trigger source.

I tested my Tektronix 2440 and it slows its waverform acquisition rate to about 10 Hz when untriggered but the 2232 and 2230 appear to operate at the same speed whether triggered or untriggered.  While they are not quite as fast making the potential difference in speed smaller, the real reason is probably related to how their analog trigger and sweep logic interact with their DSO mode. Tektronix analog oscilloscopes run their sweeps as fast as possible when in AUTO mode and untriggered.

As to the reason how the modern Tektronix DSOs behave the way they do, we can make a pretty good guess.  The AUTO mode timeout triggers a single acquisition followed by updating the display record which is that 873 microsecond dead time.  Maybe the holdoff time is extended as well.  When the other oscilloscopes timeout in AUTO mode, they fill their segmented acquisition memory and then update the display record with the multiple acquisitions stored there.

[bxs]

For me the Tek is the best because of the slow update (when no triggering) you can always see the waveforms, with the other you only see garbage and need to press stop to look at the have.

About the GW Instek GDS-2000A, it just seems slow... maybe the fast update only happens with a magic combination of configs 

[ejeffrey]

I definitely like the tek auto-update mode over the others.  I feel it gives me a better picture of the signal I am looking for and it is visually distinct from a bad timebase setting.

Does anyone have a lecroy to test?  I haven't used one in years, but I seem to recall that they work more like the tektronix.

[Towger]

Does anyone have a lecroy to test?

Dave's Lecroy is still work in progress ;-)
I bet it is still taking up space in the lab. Time to progress it into the garbage room. There is only so much stuff you can store, 'just in case'..

BTW +1 to the Tek method.

[EEVblog]

About the GW Instek GDS-2000A, it just seems slow... maybe the fast update only happens with a magic combination of configs 

Yes, I'm quite confused about this scope. Either I'm doing something real dumb (quite possible!) or the spec is BS and doesn't mean what people expect it to mean. More testing needed.

[EEVblog]

Dave's Lecroy is still work in progress ;-)
I bet it is still taking up space in the lab. Time to progress it into the garbage room. There is only so much stuff you can store, 'just in case'..

Yes, it's taking up a filing cabinet draw...
Lecroy did approach me a few years back to suss me out if I was worthy of reviewing their scopes. They ended up not liking the way I did things 

[mikeselectricstuff]

Semi-related question - why don't scopes allow you to set the auto trigger timeout... would be trivial to do, and particularly at low frequencies would be useful.

[W7NGA]

AUTO mode provides a slow trace in the absence of triggering that affords the user a trace when probing DC signals or simply assessing signal activity.

That is to say, it provides the starting point for signal evaluation and triggering. Nothing more.

Evaluating a scopes trigger-out rate in the absence of a trigger is non-sensical and serves no real purpose.

In forty years of engineering, I've rarely had to deal with waveform-update rates being at issue, preferring to concentrate on triggering techniques to identify problems.

Admittedly, I was not in my lab probing evaluation boards.

[Rutger]

Dave, please review the 'Rigol DS1074Z 70 MHz Digital Oscilloscope with 4 channels' for $ 585.00 (in the USA). I am dying to buy one, but would like to see a good review first.

[Hydrawerk]

About the GW Instek GDS-2000A, it just seems slow... maybe the fast update only happens with a magic combination of configs 

These GW Instek scopes are rather strange and the Trig Out is too low level and weird. The DPO feature is not well implemented.


Official video made by GW guys:

http://www.ittsb.eu/GDS-2102A%20Wfms%20measurments.html

c) The Trigger-Out has an voltage range not higher than 200 mV AC, which varies significantly depending the time base of the scope and usually ranges from 120 mV to 5 mV and even less than that.

Old threads here:
https://www.eevblog.com/forum/testgear/gds-2000a-new-economic-oscilloscope-by-gw-instek-comes-to-market/
https://www.eevblog.com/forum/blog/eevblog-474-gw-instek-gds-2000a-series-oscilloscope-unboxing-fi/

[EEVblog]

These GW Instek scopes are rather strange and the Trig Out is too low level and weird.

Ok, I'm able to get the 80Kwfs, but only at 500ns/div as per the video, 200ns/div and it drops to 30K, 20ns/div drops to 1K
And I can also confirm it returns to 80K when no input signal and auto triggering, just like the Agilent.
No idea why it works now, I hit the autoset button 

[Someone]

No idea why it works now, I hit the autoset button 

Delayed trigger off the screen to the left?

[Hydrawerk]

In the official video they are getting 80 kHz and only 150mV peak-peak at the Trig Out . That is bad. At my DSOX2002A it is 5V peak peak into 1MegOhm, so it can be used to synchronise other test equipment. https://www.eevblog.com/forum/reviews/my-new-toy-)-agilent-dsox2002a-sex-on-a-stick!/45/
At Rigol DS2000 the Trig Out level is still OK, as measured by marmad.

Just measured the Rigol and it appears to have a Trigger Out voltage level of ~3.5V DC into 1M; ~1.5V into 50 Ohm.

[German_EE]

Dave, you have a workshop to die for and two of the finest scopes on the planet, relax and be happy!! That video was a prime example of paralysis by analysis

[EEVblog]

Dave, you have a workshop to die for and two of the finest scopes on the planet, relax and be happy!! That video was a prime example of paralysis by analysis

I'm not complaining, it was an observation and an example of some differences in how different brand oscilloscopes work. I thought it might be useful to some people.

[dr.diesel]

That video was a prime example of paralysis by analysis

Dave's review is a prime example of completeness/thoughness, taking the comparison to the next level for those that wish to understand the differences between manufactures and operation.  TM equipment of this caliber has "character", thanks to Dave we get to experience it without ownership.   

[rf-design]

Digital scopes, blah, where some software geeks get to determine how your scope works. 

Software will define in the future what and how to trigger.

The trigger is the mean to differentiate between the changing and non-changing part of a waveform and display them on the screen. Old fashion triggering is based on on level, change, logic and timeouts but modern algorithmic methods could much easier discriminate between a regular part and a derivation from that. Also considering a scope as free programming enviroment similar to matlab where the workspace is a cycle buffer of samples and your trigger is a sequence of operations. Software is a matter of expression how functionality is actual executed. Your schematic of the trigger circuit is nothing better. Do not bash engineers for doing something different.

[Dinsdale]

Interesting video, Dave. I think the Tek gives you some idea of what kind of signal you have while in its untriggered state. What can be the point of a huge update rate on a signal you can't see, resolve, or measure?

Hydrawerk, great catch on the trigger signal levels. There's a parameter I have probably never thought about but sure could be a game changer.

And, Dave, don't worry about "rambling". Unexpected insight can come from rambling. You just don't need to comment on "rambling".

[marmad]

These GW Instek scopes are rather strange and the Trig Out is too low level and weird.

Ok, I'm able to get the 80Kwfs, but only at 500ns/div as per the video, 200ns/div and it drops to 30K, 20ns/div drops to 1K
And I can also confirm it returns to 80K when no input signal and auto triggering, just like the Agilent.
No idea why it works now, I hit the autoset button 

From published tests and videos posted by others, the most complete chart of the GDS-2000A series wfrm/s rates I was able to piece together previously was:

Time base  Sa/s       Wfrm/s
2ns        2GSa/s     850
5ns        2GSa/s     930
10ns       2GSa/s     940
20ns       2GSa/s     920
50ns       2GSa/s     55,936
100ns      2GSa/s     55,860
200ns      2GSa/s     55,600
500ns      2GSa/s     81,330
1us        1GSa/s     42,159
2us        500MSa/s   25,704
5us        200MSa/s   13,284
10us       100MSa/s   7,953
20us       50MSa/s    4,138
50us       20MSa/s    1,640
100us      10MSa/s    696
200us      5MSa/s     313

But this is single-channel with the <= 1k sample length setting. In other words, in terms of total samples captured per second, it's not very fast compared to other competing, similarly-priced brands.

[marmad]

Semi-related question - why don't scopes allow you to set the auto trigger timeout... would be trivial to do, and particularly at low frequencies would be useful.

Nice: one of those useful and simple-to-implement ideas. Let's hope someone with (DSO) design clout is paying attention.

[David Hess]

That video was a prime example of paralysis by analysis

Dave's review is a prime example of completeness/thoughness, taking the comparison to the next level for those that wish to understand the differences between manufactures and operation.  TM equipment of this caliber has "character", thanks to Dave we get to experience it without ownership.   

I like to think of it as a series of steps starting with, "Hmm, that is odd."