Author Topic: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...  (Read 76054 times)

0 Members and 1 Guest are viewing this topic.

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #150 on: June 06, 2013, 01:11:55 pm »
Elsewhere in the manual it explains that they have a pre-trigger buffer which is FIFO and which they fill before looking for the trigger event. Then they have the post trigger buffer. They always fill at the maximum sample rate and reduce it by decimation for slower sampling rates.

So for example, on an Agilent 3000X, given 1M wfrm/s @ 10ns/div, if I move my trigger position to -10 divs, does my wfrm/s drop to 750k - or 500k - or stay the same?

Quote
For segmented memory they treat each segment in the same way (pre-trigger and post-trigger buffer). It lists the reset time after each segment as 1microsecond for the 3000X series so the update time will be 2MB at 2GS/s or 1microsec plus an extra microsec which presumably reduces it the wfm/s rate from around 1M to 500K.

As soon as I spotted the word 'presumably', it made me wonder exactly how you define 'quite well'.  ;)
 

Online EEVblog

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #151 on: June 06, 2013, 01:35:12 pm »
In some ways, I would say yes. But aside from the fact that (if an Agilent owner) I would just prefer to be able to turn this feature on and off

Why?
Please cite a case were you would need to sacrifice update rate for deep memory in run mode.

Quote
The problem with 'automatic' features in complex technology is that if they're not extremely well-documented (in terms of the ramifications on all related sub-systems), the trade-offs are often not clear.

Agreed. But in this case I am claiming that there is no downside to what Agilent have done here.

Quote
For example, what does the ASIC do exactly when I want my trigger position 5 divisions to the left of the screen edge?

Depends on the type of trigger. In some cases the trigger can be entirely independent of same rate etc.

Quote
Or what does it do exactly with segments: does it capture them at the fastest speed possible while cutting down the sample length - or does it maintain the sample length while reducing it's update rate?

It must capture them at the maximum memory depth, because these are effectively multiple "stop" captures, not simply run mode screen updating.
Agilent know you want to capture and investigate these captured segmented data, so they use all the memory they can. Update rate might be dependent upon trigger type.
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #152 on: June 06, 2013, 02:00:01 pm »
Why?
Please cite a case were you would need to sacrifice update rate for deep memory in run mode.
Never - assuming the technology worked 100% perfectly 100% of the time. And I know you don't think this kind of result is important, but it is to me. Again, it's fine if you know that this is possible given the Agilent technology and certain settings - and can shut it off. Otherwise it's problematic.

Quote
Agreed. But in this case I am claiming that there is no downside to what Agilent have done here.
I think the automatic adjustment of sample size to maximize wfrm/s is a great idea - but considering that it can occasionally produce, IMO, unwanted artifacts - coupled with the fact that it imposes some conditions on you (such as constant interpolation) - means I'd just like to be able to turn it on and off if desired.

Quote
It must capture them at the maximum memory depth, because these are effectively multiple "stop" captures, not simply run mode screen updating.
Agilent know you want to capture and investigate these captured segmented data, so they use all the memory they can. Update rate might be dependent upon trigger type.
Got it - thanks for the clarifications.
 

Online EEVblog

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #153 on: June 06, 2013, 02:09:10 pm »
Never - assuming the technology worked 100% perfectly 100% of the time. And I know you don't think this kind of result is important, but it is to me. Again, it's fine if you know that this is possible given the Agilent technology and certain settings - and can shut it off. Otherwise it's problematic.

I was talking about apart from the dot/vector & interpolation thing. I do not disagree with you here.

Quote
coupled with the fact that it imposes some conditions on you (such as constant interpolation)

The interpolation is not really related to the automatic memory depth/update rate thing we are talking about, it's simply a separate decision agilent made.
Just because they both can't be changed by the user does not mean they are related.
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #154 on: June 06, 2013, 02:18:55 pm »
The interpolation is not really related to the automatic memory depth/update rate thing we are talking about, it's simply a separate decision agilent made.
Just because they both can't be changed by the user does not mean they are related.

Technically true, but the scopes perform the sin(x)/x interpolation in DSP hardware, which is part of the scope’s MegaZoom ASIC - helping to achieve the speeds they get. So I don't know how much it would affect the functioning - or alter the wfrm/s rates - if the processes were separated (although I wish they were).

BTW, I threw together a spreadsheet based on Agilent's published wfrm/s rates for the X2000 and X3000 - which estimates likely sample sizes and blind times for the different time bases. It should be interesting for X-Series owners. Note: the X2000 is faster than the X3000 at 7 time base settings.

« Last Edit: June 06, 2013, 04:17:49 pm by marmad »
 

Offline jpb

  • Super Contributor
  • ***
  • Posts: 1771
  • Country: gb
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #155 on: June 06, 2013, 02:48:40 pm »
So for example, on an Agilent 3000X, given 1M wfrm/s @ 10ns/div, if I move my trigger position to -10 divs, does my wfrm/s drop to 750k - or 500k - or stay the same?
It always captures the same amount of memory so the trigger position shouldn't make any difference. It might cause it to vary the relative sizes of the pre-trigger and post-trigger buffers.

The pan and zoom options are only available on the last captured trace and presumably this is captured after you press stop and uses the full memory depth (i.e. not split into two buffers).

I think the idea that it optimizes its memory use for rapid capture is misleading - it just splits the memory in two when in run mode - there is no optimization for different triggers or time bases or anything like that.

The only issue with it might be aliasing at slower timebases because you've only got half the memory but if the capture that is displayed when you press stop uses the full memory (the last trace) you will see the correct waveform at that point anyway.

I think this is a rather different issue to whether or not different interpolations can be turned off.
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #156 on: June 06, 2013, 03:08:49 pm »
It always captures the same amount of memory so the trigger position shouldn't make any difference. It might cause it to vary the relative sizes of the pre-trigger and post-trigger buffers.

The pan and zoom options are only available on the last captured trace and presumably this is captured after you press stop and uses the full memory depth (i.e. not split into two buffers).

I think the idea that it optimizes its memory use for rapid capture is misleading - it just splits the memory in two when in run mode - there is no optimization for different triggers or time bases or anything like that.

Sorry, jpb, but I get the impression you haven't read the other posts in the thread - and perhaps that's why you think that Agilent 'define this quite well in their manuals.'  ;)  Of course it optimizes the memory - it's not using the full size of the sample length split in half - this is obvious by simple math. Please read the previous post(s) describing the impossibility of capturing the full half-length given the wfrm/s - and look at the spreadsheet above. Then we can continue the discussion if you like.
« Last Edit: June 06, 2013, 03:16:26 pm by marmad »
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #157 on: June 06, 2013, 03:42:14 pm »
Never - assuming the technology worked 100% perfectly 100% of the time. And I know you don't think this kind of result is important, but it is to me. Again, it's fine if you know that this is possible given the Agilent technology and certain settings - and can shut it off. Otherwise it's problematic.
I was talking about apart from the dot/vector & interpolation thing. I do not disagree with you here.

I think this is a rather different issue to whether or not different interpolations can be turned off.
What I'm referring to above is not the interpolation issue (even though, yes, I think it would be good to be able to switch it off), it's about the locations of the actual sampled points. In the attached image, the Agilent is clearly doing linear interpolation, and I've marked where the sample points must be - but many of them are nowhere near the correct positions for a perfectly stable, real-time sampled sine wave - regardless of under-sampling. Compare it against another 5MHz sine wave captured with the exact same sampling rate on the Rigol. Perhaps the Agilent is also limiting based on sample rate?

Anyway, this has drifted way off-topic -- so sorry about that.
« Last Edit: June 06, 2013, 04:30:37 pm by marmad »
 

Offline jpb

  • Super Contributor
  • ***
  • Posts: 1771
  • Country: gb
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #158 on: June 06, 2013, 05:34:59 pm »
It always captures the same amount of memory so the trigger position shouldn't make any difference. It might cause it to vary the relative sizes of the pre-trigger and post-trigger buffers.

The pan and zoom options are only available on the last captured trace and presumably this is captured after you press stop and uses the full memory depth (i.e. not split into two buffers).

I think the idea that it optimizes its memory use for rapid capture is misleading - it just splits the memory in two when in run mode - there is no optimization for different triggers or time bases or anything like that.

Sorry, jpb, but I get the impression you haven't read the other posts in the thread - and perhaps that's why you think that Agilent 'define this quite well in their manuals.'  ;)  Of course it optimizes the memory - it's not using the full size of the sample length split in half - this is obvious by simple math. Please read the previous post(s) describing the impossibility of capturing the full half-length given the wfrm/s - and look at the spreadsheet above. Then we can continue the discussion if you like.

Presumably it doesn't capture memory that is not visible. It is not really optimization any more than standard oscilloscopes. It splits the memory into two buffers to allow one to be filled whilst the other is processed. Each buffer is split into pre-trigger and post-trigger buffers and it fills the pre-trigger buffer until it gets to the point where it is looking for a trigger (normally half a screen worth), while looking for the trigger the pre-trigger buffer is continuously filled with oldest points removed if there is no space. Once the trigger happens the post-trigger buffer is filled to the point of end of the display.

This all seems reasonably straight forward and is stated in the manual. It is only the last capture, the one you look at, that fills the whole memory as one buffer to allow panning and zooming.

So in your example of 10nsec/div the pre-trigger buffer only fills up to 200 points (for 5 divisions at 4GS/s). So yes, I agree it is much less than 1M or even 250k, but I don't think this is optimization anymore than my WaveJet optimises its memory when it does a similar thing (but rather more slowly ;)).

The unusual feature is the last capture which does fill the memory but this doesn't affect the wfm/s.
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #159 on: June 06, 2013, 06:06:25 pm »
Presumably it doesn't capture memory that is not visible.

Presumed by who? If you presume ALL DSOs only capture sample memory that is visible on the display at all times you're terribly wrong. They don't all operate like your WaveJet.  ;)
 
And I know what's written in the manual and extra Agilent literature quite well - but I don't feel like arguing about semantics; I'm more interested in the reasons behind my previous post.
« Last Edit: June 06, 2013, 06:19:32 pm by marmad »
 

Offline jahonen

  • Super Contributor
  • ***
  • Posts: 1052
  • Country: fi
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #160 on: June 06, 2013, 06:26:24 pm »
Those sample points are probably altered by the anti-alias algorithm. I did a test with my Agilent MSO6034A, which has an option to turn anti-aliasing on and off. It seems that the sine becomes "noisy" when anti-alias is turned on. Too bad that they have removed that option to turn aa off in newer MegaZoom generation.

But I must say that I have never noticed that, or any measurement hasn't gone wrong in past 8 years or so that I have used that Agilent. Working anti-alias has been quite good trade-off.

Regards,
Janne
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #161 on: June 06, 2013, 06:29:03 pm »
Those sample points are probably altered by the anti-alias algorithm. I did a test with my Agilent MSO6034A, which has an option to turn anti-aliasing on and off. It seems that the sine becomes "noisy" when anti-alias is turned on. Too bad that they have removed that option to turn aa off in newer MegaZoom generation.

But I must say that I have never noticed that, or any measurement hasn't gone wrong in past 8 years or so that I have used that Agilent. Working anti-alias has been quite good trade-off.

Regards,
Janne

Thanks for the clarification, Janne - very informative! So it seems a pity they didn't leave anti-aliasing, as well as interpolation, as switchable options. I wonder precisely why that's happening? I couldn't replicate the error on my Rigol using anti-aliasing at those sample rates.
« Last Edit: June 06, 2013, 06:34:23 pm by marmad »
 

Offline jpb

  • Super Contributor
  • ***
  • Posts: 1771
  • Country: gb
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #162 on: June 06, 2013, 06:46:25 pm »
Presumably it doesn't capture memory that is not visible.

Presumed by who? If you presume ALL DSOs only capture sample memory that is visible on the display at all times you're terribly wrong. They don't all operate like your WaveJet.  ;)
 
And I know what's written in the manual and extra Agilent literature quite well - but I don't feel like arguing about semantics; I'm more interested in the reasons behind my previous post.
Fair enough, I too don't particularly want to get into an argument about a scope I don't even have. My presumption was only that since there is no option to look at points outside the visible window (you can only pan and zoom on the last capture, at least that is what the manual states) it would make no sense to capture such data. This is a different situation to say segmentation where you can look at the saved history.

Almost all the points of discussion so far seem to involve data that is stored (such as the interpolation used on points that are zoomed in on), whilst the high waveforms per second data is differently captured on the Agilent - only one set is captured in detail presumably when you press the stop button. So whilst it is running it is capturing the minimum of data at a fast rate, when you stop it to look at the results it does a completely different sort of capture which will be much slower and more detailed but as there is only one of it this it doesn't matter. (Assuming the time base is fast enough for such a capture to be well under a second.)
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #163 on: June 06, 2013, 07:08:44 pm »
Almost all the points of discussion so far seem to involve data that is stored (such as the interpolation used on points that are zoomed in on), whilst the high waveforms per second data is differently captured on the Agilent - only one set is captured in detail presumably when you press the stop button. So whilst it is running it is capturing the minimum of data at a fast rate, when you stop it to look at the results it does a completely different sort of capture which will be much slower and more detailed but as there is only one of it this it doesn't matter. (Assuming the time base is fast enough for such a capture to be well under a second.)

Right - that's what we've been getting at - although tinhead added that the first capture might fill the buffer as well. So it might look something like:
(RUN)FULL/DISPLAY/DISPLAY/DISPLAY...DISPLAY/DISPLAY/DISPLAY(STOP)/FULL

I think it's a clever scheme, but I'm not sure this is explicitly spelled-out in the literature - especially since some DSOs capture the full record length over and over again - even when it's larger than the display window.
 

Offline grego

  • Frequent Contributor
  • **
  • Posts: 330
  • Country: us
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #164 on: June 06, 2013, 07:12:57 pm »
BTW folks I recently spoke to tequipment.net and Instek and I am RMA'ing my GDS2204A and the LA module.  Unfortunately the errors in the printed specs affected my purchase decision so I made the decision to try to do a return.

Please note this does not mean I don't like the scope - I do - I just personally made a purchase decision based on erroneous data and didn't want to be held to that as it would cause me issues down the road.
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #165 on: June 06, 2013, 07:15:20 pm »
Greg, any final thoughts on the LA module and decode options? I think a number of people were curious to hear information/opinions about that.
« Last Edit: June 06, 2013, 07:21:56 pm by marmad »
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #166 on: June 06, 2013, 07:47:37 pm »
Speaking of tequipment.net, I just noticed that they've cut the prices on the entire GDS-2000A series (and all of the accessories and options) by 15%. With the 5OFF coupon code, that means 20% off total.
 

Offline grego

  • Frequent Contributor
  • **
  • Posts: 330
  • Country: us
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #167 on: June 06, 2013, 07:56:17 pm »
Greg, any final thoughts on the LA module and decode options? I think a number of people were curious to hear information/opinions about that.

Not really - it worked like any other MSO LA to be honest.  I didn't use it a ton as I was waiting for the firmware update to enable CAN decodes but it was fine - basically what I wanted spec wise and seemed to function as such.  It's a plus that it's there, it works well, and it includes all the decodes with the module purchase.  I'd recommend it for that.  I am just frustrated the published specs don't conform with the expected output based on the published documentation at the time of my purchase decision.
 

Offline Hydrawerk

  • Super Contributor
  • ***
  • Posts: 2599
  • Country: 00
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #168 on: June 06, 2013, 08:03:04 pm »
Well, I didn't expect that you will return your scope... Which scope are you going to buy now? Well, I would wait for the DS2000-S with function generator...
Amazing machines. https://www.youtube.com/user/denha (It is not me...)
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #169 on: June 06, 2013, 08:05:04 pm »
Not really - it worked like any other MSO LA to be honest.  I didn't use it a ton as I was waiting for the firmware update to enable CAN decodes but it was fine - basically what I wanted spec wise and seemed to function as such.  It's a plus that it's there, it works well, and it includes all the decodes with the module purchase.  I'd recommend it for that.  I am just frustrated the published specs don't conform with the expected output based on the published documentation at the time of my purchase decision.

Thanks for the update, Greg.

BTW, if any EU members are interested in the DSO, you should seriously consider having it shipped from Tequipment (although I can't attest to dealing internationally with them). Without VAT or shipping, the difference in price, for example, of the GDS-2074A 70MHz 4-channel is:
Tequipment.net = ~€ 913,00
EU Prices           = ~€ 1.165,00

€ 200 buys a lot of shipping - and there's no duty on DSOs. And, of course, you'd save even more if you add any 20%-off options on top of that.
« Last Edit: June 06, 2013, 08:07:07 pm by marmad »
 

Offline Hydrawerk

  • Super Contributor
  • ***
  • Posts: 2599
  • Country: 00
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #170 on: June 06, 2013, 08:06:22 pm »
Now it's time for Dave to make more experiments with GDS-2000A. At least the search and place mark function looks good. DS2000 or DSOX2000 have nothing like that!
Amazing machines. https://www.youtube.com/user/denha (It is not me...)
 

Offline grego

  • Frequent Contributor
  • **
  • Posts: 330
  • Country: us
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #171 on: June 06, 2013, 08:07:00 pm »
Well, I didn't expect that you will return your scope... Which scope are you going to buy now? Well, I would wait for the DS2000-S with function generator...

Probably the new 4000 with MSO option.  But that'll have to wait until I get this RMA processed.
 

Offline Hydrawerk

  • Super Contributor
  • ***
  • Posts: 2599
  • Country: 00
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #172 on: June 06, 2013, 08:10:37 pm »
Well, the DS4000 is a heavy duty scope with extra long memory  :-+ :-+ and large fan, but too expensive for me. Most hobbyists cannot afford one.  :(
« Last Edit: June 06, 2013, 08:13:14 pm by Hydrawerk »
Amazing machines. https://www.youtube.com/user/denha (It is not me...)
 

Offline grego

  • Frequent Contributor
  • **
  • Posts: 330
  • Country: us
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #173 on: June 06, 2013, 08:26:04 pm »
Main thing is I want 4 channels, intensity grading with an MSO option - so we'll see.

And no, I don't want to spend umpty-kajillion-dollars on an Agilent.  I may have more of a budget than the average hobbyist but I'm not THAT rich! :)
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
Re: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...
« Reply #174 on: June 06, 2013, 10:43:51 pm »
Main thing is I want 4 channels, intensity grading with an MSO option - so we'll see.
Some interesting stuff in the data sheet for the upcoming MSO4000 series (if you want to Google Translate Chinese):

Here are some specs for the Digital channels (lots o'memory - up to 28M per channel):

Digital channel sampling rate 1GSa/s, memory depth per channel up to 28Mpts standard
Digital channel waveform capture rate of 85,000 wfms/s
Support hardware real-time digital channel waveform recording, playback, recording a maximum of up to 64,000 frames
Supports analog and digital channels mixed triggering and decode
Convenient digital channel grouping and group operations
Support for multiple logic levels
Time-correlated analog and digital channels are displayed waveform

Real-time sampling rate: Digital channels:  1.0 GSa/s
Peak detection:               Digital channels:  1 ns
Minimum detectable
pulse width:                    Digital Channel: 5 ns
Memory Depth                 Digital channels: Max 28M Point per channel

Threshold:                      8-channel group 1 adjustable threshold
Threshold Selection:       TTL (1.4 V)
                                       5.0 V CMOS (+2.5 V), 3.3 V CMOS (+1.65 V)
                                       2.5 V CMOS (+1.25 V), 1.8 V CMOS (+0.9 V)
                                       ECL (-1.3 V)
                                       PECL (+3.7 V)
                                       LVDS (+1.2 V)
                                       0 V
                                       User-defined
Threshold range:            ± 20.0V, 10 mV steps
Threshold Accuracy:        ± (100 mV + 3% of threshold setting)
Dynamic Range:              ± 10 V + threshold
Minimum voltage swing:  500 mVpp
Vertical resolution:          1 bit
« Last Edit: June 06, 2013, 10:54:02 pm by marmad »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf