GW Intek GDS-2000E VPO and other problems

[Krisztián]

GDS-1000B has a very good 256 level gradient while GDS-2000E barely displays 16 gradient and even it's more expensive.

1000B series:
https://www.youtube.com/watch?v=71GWJ-A-tsQ&feature=youtu.be&t=718

My 2000E, sorry for quality (FW 1.36):
I've deleted my first uploaded video since it does not show the problem well enough. Please see the videos below. These videos are still of very poor quality. Due to the automatic dimming of my phone, the real problem is hardly apparent.
Soon, I'll replace the videos with better quality.

Xilinx Zynq is very powerful and therefore can compute much more complex visualization.

• 120k waveforms per second
  'the engineering team was able to improve the bandwidth range to 10 to 30 Msps'
  Cited from that pdf: https://www.xilinx.com/publications/prod_mktg/zynq7000/goodwill-casestudy.pdf
  
  30 Msps divided by 1000 sample is 30ksps.
  I not really care about that but why are 120ksps in the datasheet written then?
  
  
• 1mV/div is not real.
  Export waveform data to fast csv and you'll see only even numbers at 1mV/div vertical setting.
  Average helps. By turning on it will have odd numbers in the data.
  
  
• Large offset with long settling time. More than 1mV in 1 hour. Attached a screenshot.
  After waiting 1 hour and doing the SPC some channel still has 400uV offset right after SPC. Yes, I did vertical calibration already. My 30 years old Hitachi VC-6145 has at most 400uV without calibration.
  

It's very good thing that the backlight brightness can be adjusted.

[nctnico]

I think you should read the user manual more carefully. You'll see the offset change on every digital oscilloscope no matter what brand or price. That is why you need to wait for 30minutes or more before doing a self calibration (which among other things calibrates the offsets). Waveform update rate is always a function of memory depth and samplerate. For very deep memory you won't get high update rates. Don't get hung up on high update rates. It is just marketing BS. For example: when you enable roll-mode you get zero waveform updates per second because the waveform never ends. Does this make roll-mode useless??

VPO is a function of update rate (memory depth) and persistence setting. However GW Instek did implement it in a way so traces don't become invisible like on an analog oscilloscope. Maybe you don't like that but the whole purpose of an oscilloscope is to show signals.

[Krisztián]

Dear @nctnico, this is my third oscilloscope.

VPO is a function of update rate (memory depth) and persistence setting. However GW Instek did implement it in a way so traces don't become invisible like on an analog oscilloscope. Maybe you don't like that but the whole purpose of an oscilloscope is to show signals.

I've been testing this oscilloscope since Friday and this IS a problem. And for this reason, this screen will almost always have to be adjusted depending on the frequency of the signal. This is very confusing.

In all four videos 1.36 firmware, Persistence is off !!!, Vector mode, 50% Waveform Brightness!!!

VIDEO1
Display should not depend on memory size. But the GW-INSTEK 2000E depends on it. This is just a poorly written display algorithm. It needs to be fixed in the next firmware. [132mV pk-pk 50kHz sine AM modulated with 21.76Hz sine.]


VIDEO2
It is possible that only my oscilloscope is bad, but in Run mode it is badly calculated and the brightness is maximized everywhere. Look at VIDEO2 where I switch between Run and Stop mode. Only in Stop mode is the display good! [132mV pk-pk 76kHz sine AM modulated with 18.5Hz sine.]


VIDEO3
After changing Holdoff or timebase, the display is good for a moment and then it's bad again. [132mV pk-pk 76kHz sine AM modulated with 18.5Hz sine.]


VIDEO4
This time I only change timebase.
[132mV pk-pk 50kHz sine AM modulated with 21.76Hz sine.]



Anyway, the oscilloscope is very responsive, I like it. I hope they will fix visual problems.

[Krisztián]

VIDEO5
In the beginning, the oscilloscope is in Stop mode. At this point, the intensity graded display works. Then I press the Run/Stop button. [240ms Persistence, 50% waveform brightness. Vector mode]
Now, the current signal is always at maximum brightness. Then I change Persistence to 120ms. Persistence will then be asymmetrical. How is this possible? At the end of the video, I switch back to Stop mode and again the intensity graded display is working well.
https://youtu.be/o_HNlUL0BDA

[nctnico]

I see nothing wrong with it. What exactly are you expecting?

[Krisztián]

I see nothing wrong with it. What exactly are you expecting?

Which video do you write this? I uploaded at least five.

I expect intensity graded display in Run mode. Intensity grading does not equal persistence. They are two separate things, but they can be displayed simultaneously. I upload two pictures of this.

I think you should read my posts and watch my videos more carefully.
Please read my comments again. Watch my videos on a full HD screen.
I will make better quality videos if I have the opportunity. I've recorded the videos so far with my phone.

[nctnico]

You have to adjust your triggering. Your trigger level isn't even withing the signal amplitude so it is triggering after the auto-trigger time-out has expired. My GDS2204E  shows the bottom picture in run mode.

[Krisztián]

Please try again at 10 Mpts. My oscilloscope is set to 10M pts.
Whether it is triggered or not, Run mode does not have an intensity graded display at 10 Mpts.
But in stop mode it has.
Press the Run / Stop button at 10M. You'll see what I'm talking about.

I think Zynq should easily compute that visualization task even at 10M pts. Even the weaker FPGAs can calculate this in real time.

[Krisztián]

My GDS2204E  shows the bottom picture in run mode.

Are you sure your oscilloscope shows the bottom picture at 10M pts? In this case, my oscilloscope is defective.

[wraper]

My GDS2204E  shows the bottom picture in run mode.

Are you sure your oscilloscope shows the bottom picture at 10M pts? In this case, my oscilloscope is defective.

Does it even matter if you don't even set trigger properly? 

[Krisztián]

My GDS2204E  shows the bottom picture in run mode.

Are you sure your oscilloscope shows the bottom picture at 10M pts? In this case, my oscilloscope is defective.

Does it even matter if you don't even set trigger properly? 

Did you read these sentences?

Whether it is triggered or not, Run mode does not have an intensity graded display at 10 Mpts.
But in stop mode it has.

Do you think I did not try it out?
I saved that screenshot just to show persistence. 

Let's go to sleep. That was enough. 

[rf-loop]

My GDS2204E  shows the bottom picture in run mode.

Are you sure your oscilloscope shows the bottom picture at 10M pts? In this case, my oscilloscope is defective.

Does it even matter if you don't even set trigger properly? 

Trigger have nothing to do in this case. If you think it matter then please explain exactly how.

(if we think this case where acquisition length is 10M sample and t/div is 10ms/div and samplerate is 100Msa/s. There is 10M sample what need all be mapped to display (lot of samples overlaid for on position in display map and how many, it produce intensity. (I do not know if GoodWill maps every sample in run mode to display map and scale gradation depending how many samples in same position in map? It is clear it do it after stop as seen in these images-videos).
Of course with faster timebases where one TFT frame have several acquisitions overlaid it is other case and there of course trigger also affect how much sequential sample points is mapped to same position in display map for produce intensity. But this is other case. In this case here one acquisition time is 0.1s and really one image map do not include several acquisitions overlaid.

In this case intensity gradation need work without any trigger, single shot and continuous run

-- as it works in normal DPO scopes. Also here in these example images. 76kHz AM mod 18.5Hz and depth 90%.
(do not care freq counter, true carrier is 76000.00Hz, it is trigger event counter and part of carrier is not counted due to under level and time.)

1.Run
2.Stop after Run
3.One individual Single shot
4.Run but without any trig. Scope in Auto trig mode. Persistence 5s

[tautech]

rf-loop has new and unbranded DSO that looks much like a Siglent. 

Werewolf mode ?

[nctnico]

Just look at the signal update rate at long memory settings. There is nothing to intensity grade! GW Instek has choosen for an algorithm which doesn't hide signals so the intensity grading is minimal to begin with.

[rf-loop]

Just look at the signal update rate at long memory settings. There is nothing to intensity grade! GW Instek has choosen for an algorithm which doesn't hide signals so the intensity grading is minimal to begin with.

Also good manufacturer do not hide any information. Every single sample is mapped to display and displayed, independent of update rate, memory, sampling speed and memory settings. And if want even more visibility user can choose color grading instead of intensity grading or just turn trace intensity higher (what also reduce amount of intensity range in gradation.)
Intensity gradation give more information about things in signal and I can not find any reason (except these salesmen's arguments) why GoodWill select algorithm what hide this information, perhaps just because it is simplest method (and better need bit more processing).

[rf-loop]

rf-loop has new and unbranded DSO that looks much like a Siglent. 

Werewolf mode ?

No, brand do not mean here anything. This is just only tiny example how some modern even entry level (but not toy level) oscilloscopes works, independent of who have made it or who's name is printed on front panel or displayed in TFT. Name in TFT top left corner, 1 cent to "lot" depending brand value and front panel sticker perhaps 20 jiao or more depending volumes and brand value.

[Krisztián]

Intensity gradation give more information about things in signal and I can not find any reason (except these salesmen's arguments) why GoodWill select algorithm what hide this information, perhaps just because it is simplest method (and better need bit more processing).

@rf-loop

Thank you very much for your informative comments!   
Although I knew all this, but English is not my native language and it would be difficult to describe it.

Seems that some of us only look at digital signals on the oscilloscope...

So this is a played match? Does GW-Instek no longer change the display algorithm? I wrote to them, but they have not responded yet.
Tomorrow is the last day I can return the oscilloscope to the seller.

Chance for a 3rd party custom firmware near future ?

Thanks again!

[rhb]

Chance for a 3rd party custom firmware near future ?

Thanks again!

Not near future, but I bought a GDS-2072E solely for that purpose along with a couple of Zynq dev boards.  While the DSP is trivial for me, the FPGA is just as difficult as for anyone else starting out with FPGAs.  And I am easily distracted.

My LeCroy DDA-125 which is an LC684DLX with some enhancements is certainly not an entry level product and it only does a color graded display with persistence turned on.

The Instek can be annoyingly buggy, but that seems to be the case with all DSOs these days.  After experience with an MSOX3104T and an RTM3104 I've become more tolerant of the annoyances of the Instek.  And GW has been pretty good about fixing bugs I've reported to them.

[Krisztián]

I've become more tolerant of the annoyances of the Instek.  And GW has been pretty good about fixing bugs I've reported to them.

Thank you for your opinion! 

Have you already mentioned errors in displaying them? Did they answer something about that?

[rhb]

What FW version are you at?  I think that 1.35 is the most recent, but GW does not notify users of updates, so you have to go online to check.  They may on their Facebook page, but I don't use Facebook.

FYI Color graded persistence is a *difficult* problem unless persistence is infinite.  It's one of the few DSO operations that I consider seriously challenging to implement.  So, I think your expectations may be a bit skewed from reality.  Intensity grading or color grading are showing you probability density distributions over the period of time specified by the persistence. To implement that you have to constantly be adding and subtracting points to the counts for each pixel.  Adding the current point is easy.  But to remove a point added N sweeps ago you have to have stored it somewhere.  So to do that smoothly you need a *lot* of memory available.

I ran into problems with trace math which is partially fixed in 1.35.  It still won't let me make hysteresis plots by using C1 and Integral(C2), but Keysight and Rohde & Schwarz can't either.  That's a large part of why I bought a LeCroy DDA-125/LC684DLX.  It does.

The sad fact is *any* DSO you buy is going to be buggy and lots of the bugs will never get fixed.  Whether your spend $300 or $300,000.

I also have a Rigol DS1102E which is pretty tolerable, but the FFT is completely useless.  The Instek FFT is actually quite good even if a pain to use.  The MSOX3K and RTM3K  FFTs are completely unusable for the same reason as the Rigol.  Too short an FFT length.

FWIW I got *so* frustrated by buggy DSOs that I bought a refurbished Tek 485 just to have something that did *not* depend on software for its operation.

It's worth noting that I also have an Instek MSO-2204EA as my primary scope.   One other note of possible interest,  I was able to convert the GDS-2072E to 200 MHz BW using the instructions in a thread on the forum.  Don't bother with the 300 MHz option as the Nyquist sampling at 500 MSa/S is only 250 MHz.

I suggest you ask the seller for a 7-14 day extension of the return period while you try to resolve the VPO issue. Tell them you have contacted GW and are waiting for them to respond and that you'd like to give them more time rather than return the scope now.  I feel pretty confident they will agree.

Who are you trying to contact at GW?

[nctnico]

FYI Color graded persistence is a *difficult* problem unless persistence is infinite.  It's one of the few DSO operations that I consider seriously challenging to implement.  So, I think your expectations may be a bit skewed from reality.  Intensity grading or color grading are showing you probability density distributions over the period of time specified by the persistence.

True but on the other hand DSOs have an advantage that analog scopes don't have: they can display a stray trace which would remain hidden on an analog oscilloscope. For that reason I don't get why people get so hung up on emulating the part which is the worst of all on an analog oscilloscope. Some degree of intensity grading is usefull but beyond some point you don't get any extra information from a signal. I agree it would make much more sense to have color grading or inverse brightness to discriminate between often and rarely occuring signals.

But let's go back to the GDS2204E. I did some more testing and I think the problem is in the triggering combined with VPO since there is an intensity graded image in stop mode. It is very hard to get a stable trigger on an AM modulated signal so I think the signal is shifting left/right due to a different trigger point in run mode preventing an intensity graded image in some situations. I'm quite sure there is nothing to fix here on the GDS2200E.

[Krisztián]

It is very hard to get a stable trigger on an AM modulated signal so I think the signal is shifting left/right due to a different trigger point in run mode preventing an intensity graded image in some situations. I'm quite sure there is nothing to fix here on the GDS2200E.

Thank you for coming back.

No, it's not hard at all, just press the trigger menu button and set the holdoff time. First, press the gray round button to the right of '4.00ns'. A 10-based logarithmic scale comes up. Use the variable button to set the roughly right value. Then press the select button to refine it.

I think the signal is shifting left/right due to a different trigger point in run mode preventing an intensity graded image in some situations. I'm quite sure there is nothing to fix here on the GDS2200E.

rf-loop has already refuted this with his attached images. Siglent can do that at 14Mpts. Intensity grading must work independently of triggering.

They are hiding the vertical distribution of the signal with full brightness in some situations.
In these situations, the oscilloscope screen is virtually only good for determining peak to peak.

Otherwise persistence is permanently turned off because I do not see any benefit.
I want to see a fast, dynamic high contrast display, not a blurry patch.

Please look at rf-loop's response and screenshots again.

[rhb]

I've put my 40  PS rise time 10 MHz square wave on my MSO-2204EA with  FW 1.32.    Despite 5 minutes of fiddle, I am unable to get anything resembling a useful gradient display at any sample length.  Which is not to say it can't be done.  Just I can't find a way to do it.  I don't find any option except to set the persistence, but it never does a graded intensity at all.

I'm using an edge trigger, so it's very stable.  The only part of the signal with any variation is the ripple after the step function.

I'm beginning to think I should devote a day or two to documenting a long list of bugs and then raising the issues with GW. For example, the digital filter app doesn't work at all.  There is no way to turn it on.

Looking at the videos posted and based on my experience, I don't think the GDS-2000E does any sort of intensity graded display with 1.32.  As it appears that 1.36 is available I'll give it a try.

[Krisztián]

What FW version are you at?  I think that 1.35 is the most recent, but GW does not notify users of updates, so you have to go online to check.  They may on their Facebook page, but I don't use Facebook.

The current firmware is 1.36 for GDS-2000E. This is currently on the oscilloscope with BW extension to 200MHz.

You can download from here, but only after logging in:
https://www.gwinstek.com/en-global/download/index?cate=55&subcate=92&ser=118&down=86&key=

Who are you trying to contact at GW?

I do not know them, but here I wrote:
technical@goodwill.com.tw,
services@goodwill.com.tw

[Krisztián]

Despite 5 minutes of fiddle, I am unable to get anything resembling a useful gradient display at any sample length.  Which is not to say it can't be done.  Just I can't find a way to do it.  I don't find any option except to set the persistence, but it never does a graded intensity at all.

This is exactly the situation I wrote earlier to nctnico here:

They are hiding the vertical distribution of the signal with full brightness in some situations.
In these situations, the oscilloscope screen is virtually only good for determining peak to peak.

Otherwise persistence is permanently turned off because I do not see any benefit.
I want to see a fast, dynamic high contrast display, not a blurry patch.