Author Topic: EEVblog #474 - GW Instek GDS-2000A Series Oscilloscope Unboxing & Fi...  (Read 76035 times)

0 Members and 1 Guest are viewing this topic.

Offline firewalkerTopic starter

  • Super Contributor
  • ***
  • Posts: 2450
  • Country: gr


Those fast alternating colors could easily give you a seizure...  :P :P :P

It was like to anime warriors fighting with light swords.

Alexander
Become a realist, stay a dreamer.

 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
@Dave - Perhaps some different combination of the persistence time and intensity level gives a better result for the waveform grading?

If it doesn't, that's certainly not well-implemented.
« Last Edit: May 26, 2013, 03:06:31 pm by marmad »
 

Offline tinhead

  • Super Contributor
  • ***
  • Posts: 1918
  • Country: 00
    • If you like my hacks, send me a donation
Dave,

you probably worked much more with Agilent than TEK scopes  :scared:, the color grading is beautiful thing.
In the last minutes of your video, you can see glitches coming with different color, i always liked this on TEK
when color grading enabled. On the other side, when using single color, there should be better intensity
implementation (right, like on Agilent/Rigol/Hameg).


The glitch you saw on moduleated signal, probably coming from DMM module enabled or the digital filter (i think it was still on)
« Last Edit: May 26, 2013, 02:27:31 pm by tinhead »
I don't want to be human! I want to see gamma rays, I want to hear X-rays, and I want to smell dark matter ...
I want to reach out with something other than these prehensile paws and feel the solar wind of a supernova flowing over me.
 

Offline firewalkerTopic starter

  • Super Contributor
  • ***
  • Posts: 2450
  • Country: gr
Ethernet and VGA out are modular? You have to remove the function or the logic analyzer?

Alexander.
Become a realist, stay a dreamer.

 

Offline madires

  • Super Contributor
  • ***
  • Posts: 7695
  • Country: de
  • A qualified hobbyist ;)
I got mixed feelings about the GDS-2000A. Some stuff is designed badly, but there are also some nice features. The firmware looks very beta, the function generator is a bad joke, the blue border on the screen stealing real estate, the cumbersome frequency setting and so on. If this would be the very first scope of new no-name company for €1299 I'd say it's ok and two or three firmware updates are going to fix most software and UI issues. But for US$ 2700, no thanks!

And for a more gerenal rant, I dislike that culture of selling the same hardware capable of x MHz for different prices while limiting the bandwidth by software to create some virtual price levels. That means that I pay for the fully fletched hardware anyway but have to pay for a byte in the firmware a lot of money extra. There's no difference in development or production costs, it's just a stupid idea of marketing! For me that's a rip-off, plain and simple.

BTW, a local company is selling the 2304A for € 3500. It's hard to believe that!
 

Offline Hydrawerk

  • Super Contributor
  • ***
  • Posts: 2599
  • Country: 00
https://www.eevblog.com/forum/testgear/gds-2000a-new-economic-oscilloscope-by-gw-instek-comes-to-market/315/
Also see this long thread.
Quote
The firmware looks very beta, the function generator is a bad joke
Especially after Rigol releases their DS2000-S with 200Ms/s 25MHz arbitrary function generator.
Amazing machines. https://www.youtube.com/user/denha (It is not me...)
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 8515
  • Country: us
    • SiliconValleyGarage
about the looks : the buttons on the rigol are too close together. the front panel is too 'busy'. i kinda like the spread-out look of the instek. Everyting nicely lines up on the instek (controls and bnc)

the only thing that looks a bit cartoonsih on the instek are the 'bubbles' . like the round dots around the vertical settings , the round buttons right of the display and the oversized klunky round power button.

not having pushable encoder ( alternate function) is a bit of a turndown. like you said ( autocentre , centre trigger point etc .. ) the cost between a pushable encoder and a non-pushable is negligible ( we're talking a few cents... ) so yeah , that's inexcusable.

no autoprobe is simply because the probes they supply don't have that capability.

Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline Hydrawerk

  • Super Contributor
  • ***
  • Posts: 2599
  • Country: 00
For some reason even Tektronix almost never uses pushable knobs. I wonder why?? I love pushable knobs.
Amazing machines. https://www.youtube.com/user/denha (It is not me...)
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
There's no difference in development or production costs, it's just a stupid idea of marketing! For me that's a rip-off, plain and simple.

Not to agree or disagree about the policy - but I think the point is that the company can recoup research and development costs for the whole series more quickly by spreading it out between many models. If they just sell a 300MHz BW DSO for $X, they're less likely (or it will take much longer) to start making profit than if they sell 300MHz for $3/4X, 200MHz for $1/2X, and 100MHz for $1/4X (or something like that).
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 8515
  • Country: us
    • SiliconValleyGarage
There's no difference in development or production costs,



you really do not understand economy of scale when it comes to integrated circuits.

you are making 4 scopes :
1 - 4 channel 200 mhz 1 million points  4gs/s
2 - 4 channel 200mhz 4 million points 4gs/s
3 - 4 channel 500mhz 1 million points 4gs/s
4 -  4 channel 500mhz 4 million points 4gs/s

50% of the buyers  go fo rmodel 1
25% go for model 2
20% go for model 3
5% go for model 4.

you assume there will be a total of 1 million scopes sold.
half a million model 1, quarter million model 2 etc ..

if you need to write off the development cost  of model 4 over the volume you will sell these will be very expensive machines.

The maskset for a typical scope asic is around 4 million $ ... these days. so you are burning 12 million $ to cover 50% of your targeted volume ! ( models 2,3 and 4 combined are 50% )

and you need to make 4 boards, etc ..

now, if you make the thing programmable. you pay for 1 maskset that covers 100% of the scopes.

The cost of the chip itself ( the cost of producing the silicon , assuming everything else is already paid for ) is purely dictated by square millimeter. having a chip with 1 megapoints go to 4 megapoints adds maybe a square millimeter ... which will cost you 1 $ ...

but, throw in the development and mask cost ( if you are only going to sell 50000 of those machines and that all of a sudden becomes a much more expensive chip !. )

if you make 1 million chips . the chip itslef costs you 5$ in silicon and 8$ in mask cost and add in 20$ development cost (10 million dollar to design it spread over half a million chips) . let's say 30$ total

for the chip for the hi end machine : it will stilll cost 7$ ( slightly larger area ), but now you tack on the 10 million dollar split over 50000 chips... which is is 200$ and the mask cost split over 50000 chips... which is 80$ ..

and the chip will cost roughly 300$ ... . add all that stuff up for your 4 designs and thats a lot of money !

do everything with 1 chip.  7$ in silicon (we're always using largest die ) , 4$ in mask ( 4 million spread over 1 million chips now ) and 10$ devcost ( 10 million spread over 1 million ) and that chip costs you 21$ ..

you can't beat that price. The most profitable solution is to mass produce the high end chip and simply disable functions.  the cost of the silicon is nothing compared to the design and the startup cost !

that is the reality of semiconductor production.
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline tinhead

  • Super Contributor
  • ***
  • Posts: 1918
  • Country: 00
    • If you like my hacks, send me a donation
that is the reality of semiconductor production.

and btw, this is the case for DSOs since 15+ years, e.g. you can take cheap ass TDS540B, change 8 resistors to make the scope to think "i have color dac"
(and then make use of external monitor or course), change another 3 resistors to make the scope thinking "i'm TDS784A", recalibrate it and now
you have 4GS/s max (1GS/s on all 4ch), 500kpoint memory and 1GHz scope instead of 2GS/s (500MS/s on all 4ch), 50kpoint memory, 500MHz bw scope.
No idea what was the list price of TDS540B (TDS744A was however 17k USD), but TDS784A was 34 USD (+3k for 1Mpoint memory). So already between
744A and 784A we have 17k USD diff!, for 3 resistors ( and sharing of development costs of course).

madires,

and yeah, nobody was complained about, you will find as well other manufacturers/models from last +10 years doing the same, and i'm glad it's like that,
i don't want to pay development twice.
I don't want to be human! I want to see gamma rays, I want to hear X-rays, and I want to smell dark matter ...
I want to reach out with something other than these prehensile paws and feel the solar wind of a supernova flowing over me.
 

Offline madires

  • Super Contributor
  • ***
  • Posts: 7695
  • Country: de
  • A qualified hobbyist ;)
do everything with 1 chip.  7$ in silicon (we're always using largest die ) , 4$ in mask ( 4 million spread over 1 million chips now ) and 10$ devcost ( 10 million spread over 1 million ) and that chip costs you 21$ ..

you can't beat that price. The most profitable solution is to mass produce the high end chip and simply disable functions.  the cost of the silicon is nothing compared to the design and the startup cost !

that is the reality of semiconductor production.

I wouldn't expect anything else. If there would be different input circuits or whatever for the different models I'd accept also different prices. But if it's just a single byte in the firmware to determine the bandwidth while the hardware and the features are 100% the same for all models it's hard to explain the price difference. Why not just have the top model (2 and 4ch)? I know, you have to provide models in the right price range for the different target groups of buyers. In your example you estimated 5% buyers for the top model. If you do a mixed calculation over all the models and expected sales you'll end up with the price of the low-midrange model. If hardware and software are the same simply sell only the "top" model for the price of the low-midrange one. You will even sell more because the buyers see a fair pricing and top bandwidth becomes more affordable.
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 8515
  • Country: us
    • SiliconValleyGarage
uhhuh. but, so far we have only calculated the cost of the machine... if you sell your 1 million scopes you end up with 0$ at the end of the run... so where will you get the money for the next design ? ( which will be more expensive because you need to  develop new features , higher speed , ore expensive board and chip .. )
all your employees also need money ...

you need to get profit from each category. so 500.000 scopes at 100$ dollar profit means that for the 50.000 scope you need 1000$ profit.

thats how you balance it out. your low end scope is cheap and the high-end machines cost money. so each category attributes the same amount of profit.

and we haven't looked at the shareholders. the reason for existence of a company i to give shareholders a return on their investment... not to provide you with a scope at cost...
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
about the looks : the buttons on the rigol are too close together. the front panel is too 'busy'. i kinda like the spread-out look of the instek.

Gotta agree with Dave here - I don't know if you've used the Rigol, but the knobs and buttons don't feel too close together at all - they feel just right. Plus the horizontal scale knob is where it 'should be' - by the edge of the display.

And from what I can tell from the video, the UI is just much better overall on the Rigol - from the multifunction knob you can push any time to adjust the intensity  - to the use of two knobs (multifunction = fine & navigation = coarse) whenever you're adjusting a value with a large range. It might look more crowded, but from an interface standpoint it's much more clever.
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16272
  • Country: za
Good first version, at least the hardware is there, just needs a lot of UI changes and upgrades. Hopefully this gets back to the design team and they look into putting this into a firmware upgrade and a few production changes to improve it. At least the UI is field upgradable reasonably easily.

Looking at the plugins they look like standard PCI card slots ( good to 400MHz and already cheap and a well proven connector) and possibly they are using standard PC bus layout with only a few dozen pins allocated to do the additional functionality needed from the module. As the FG moved with the module I would think they used 16 differential pairs of lines as IO, and one pair for the FG output. the logic pods probably have a little intelligence in them as well, likely the comparators, trigger point voltage source and an ID chip for an internal I2C bus.

As to the differences I would say they make a single board for all four options, just not populating the input channels for the extras, but then just have the firmware with an option byte set to do the various options. Having a common input circuit makes manufacture easier, the difference in price between a 100 and a 400MHz opamp will not be much per unit when you buy in 100k lots, and buying only the one at high volume will probably work out cheaper than having the 2 stock items and then having to make 2 boards with different components and layout on each.
 

Offline tinhead

  • Super Contributor
  • ***
  • Posts: 1918
  • Country: 00
    • If you like my hacks, send me a donation
Plus the horizontal scale knob is where it 'should be' - by the edge of the display.

why??? The timebase section near trigger section makes more sense, far from display.
Near display other things need to be implemented (like intensity, multiknob, etc., but not horzontal scale).

This is for me so self-evident as the fact that Rigol DS2xxx is having crap buttons left and right side of the display,
and then two menu, and 4 menu top/buttom .. why? Is it really necessary to overload display near area with
useless things?
« Last Edit: May 26, 2013, 05:19:13 pm by tinhead »
I don't want to be human! I want to see gamma rays, I want to hear X-rays, and I want to smell dark matter ...
I want to reach out with something other than these prehensile paws and feel the solar wind of a supernova flowing over me.
 

Offline madires

  • Super Contributor
  • ***
  • Posts: 7695
  • Country: de
  • A qualified hobbyist ;)
uhhuh. but, so far we have only calculated the cost of the machine... if you sell your 1 million scopes you end up with 0$ at the end of the run... so where will you get the money for the next design ? ( which will be more expensive because you need to  develop new features , higher speed , ore expensive board and chip .. )
all your employees also need money ...

you need to get profit from each category. so 500.000 scopes at 100$ dollar profit means that for the 50.000 scope you need 1000$ profit.

thats how you balance it out. your low end scope is cheap and the high-end machines cost money. so each category attributes the same amount of profit.

and we haven't looked at the shareholders. the reason for existence of a company i to give shareholders a return on their investment... not to provide you with a scope at cost...

What's wrong with simply adding up all costs, future R&D, profits and so on and dividing that by 1M scopes (considering 2 and 4ch versions of course)? The method you described creates artificially low price scopes and high price scopes which are the exact same product but one byte in the firmware. The high price scopes subsidize the low price ones. In average the company will make the same profit by numbers. But if the company is able to sell the scope for less than the competitors (with the artificially pricing) while making the desired profit it will even sell more scopes because of that. The scope will be more expensive than the low price version of the competitor, but it offers 300MHz vs. 70MHz at the price of the 150MHz, maybe it's just 200-300 bucks more than the 70MHz. That wont work for the $250-350 market but for the market above $1000.

Let's assume for example:

Vendor #1
300MHz 2ch $1400

Vendor #2
70MHz 2ch $1000
100MHz 2ch $1200
200MHz 2ch $1700
300MHz 2ch $2300

Both vendors got nearly the same models with the same features and done a proper calculation. Vendor #2 offers bandwidth just by software option (all models same hardware and features). Which scope would you buy if you're looking for a 2ch with a bandwidth of >= 100MHz (assuming the scope offers all the features you need)?


 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
why??? The timebase section near trigger section makes more sense, far from display.

More sense? One could mount a logical UI argument for positioning it anywhere - but it comes down to what you prefer. I prefer it near the display - and so does Dave, it seems. But you don't.
 
Quote
Near display other things need to be implemented (like intensity, multiknob, etc., but not horzontal scale).

Yes - those things ARE implemented near the display on the Rigol - as is the horizontal scale - which makes sense to me.

Quote
This is for me so self-evident as the fact that Rigol DS2xxx is having crap buttons left and right side of the display, and then two menu, and 4 menu top/buttom .. why? Is it really necessary to overload display near area with useless things?

There is nothing useless near the display on the Rigol - everything has a function - and in any case, none of the buttons are 'in the way' of the display. To me, the interface is fluid and intuitive. But really, I learned a long time ago that trying to argue about interface preferences is a waste of breath - either's Dave's complaints rang true for you - or they didn't - but there is no correct answer.
« Last Edit: May 26, 2013, 10:32:41 pm by marmad »
 

Offline tinhead

  • Super Contributor
  • ***
  • Posts: 1918
  • Country: 00
    • If you like my hacks, send me a donation
... but there is no correct answer.

i worked in last 15yrs (mostly) only with TEK and Yokogawa scopes (ok, in last 3 yrs also ~2.5 with Tekway), this is why i can easy recognize/identify
things what looks&feel TEK (Yokogawa is here anyway special). Dave did it with Agilent gears, so he can identify himself better
with their UI/look&feel.

So the correct answer is - use what you like to use (but be open for new things as well)
« Last Edit: May 26, 2013, 10:47:01 pm by tinhead »
I don't want to be human! I want to see gamma rays, I want to hear X-rays, and I want to smell dark matter ...
I want to reach out with something other than these prehensile paws and feel the solar wind of a supernova flowing over me.
 

Offline Hydrawerk

  • Super Contributor
  • ***
  • Posts: 2599
  • Country: 00
Well, if you get used to Tek and Yokogawa, then you find Rigol completely different. But I like the dedicated measurement menu on left side of display...
Amazing machines. https://www.youtube.com/user/denha (It is not me...)
 

Offline nitro2k01

  • Frequent Contributor
  • **
  • Posts: 843
  • Country: 00
I believe the problem at 52 minutes into the video is an aliasing problem. The sample rate (the rate of samples actually used for the display) and frequency of the signal happen to line up in such a way that you get a line like that. Annoying.
Whoa! How the hell did Dave know that Bob is my uncle? Amazing!
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
So the correct answer is - use what you like to use (but be open for new things as well)

BTW, I hope you noticed in my original comment about the placement of the horizontal scale knob I wrote the phrase 'should be' in quotes - to indicate that I was joking - and I knew it was just personal preference. ;)
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
@Dave - Perhaps some different combination of the persistence time and intensity level gives a better result for the waveform grading?

Yeah, perhaps. This was a direct side by side comparison though, and in this particular case it's came out 2nd best.
It may indeed work better under other conditions.
 

Offline IanB

  • Super Contributor
  • ***
  • Posts: 11790
  • Country: us
Not sure I will watch to the end of the video (1 h 12 m long? Are you kidding me?), but at the beginning I do not find anything bad about the control layout. Everything is nicely organized and separated. There is no obvious case of multi-function knobs, which is a good thing for usability.

(There is a maximum watchable length for videos of 20-25 min. Seriously. Please get that machete and edit out the filler. Slash and burn baby, slash and burn!)
 

Offline marmad

  • Super Contributor
  • ***
  • Posts: 2979
  • Country: aq
    • DaysAlive
There is a maximum watchable length for videos of 20-25 min. Seriously.

Says who?

Quote
Please get that machete and edit out the filler. Slash and burn baby, slash and burn!

Just skip the parts you don't want to watch. I enjoyed watching the entire thing - it made my train trip to Paris more enjoyable. In fact, there are so many textbook examples of bad choices in the firmware (the method of setting the frequency of the FG was a classic) that I'm thinking of using parts of it in the next class that I talk about interface design in.
« Last Edit: May 27, 2013, 01:32:25 am by marmad »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf