Year 2013 Oscilloscope Choices...

[hgg]

Bored@Worked I would have to say that in my case rf-loop was indeed unbiased, telling me all the
problems that the Siglent unit has.  He also sells Owon.   He also helped me with a lot of info while
me trying to buy the Siglent from a different seller in Germany!  So I think this is unfair.

eevblogfan where have you been !!! Now you are telling me...?   
I would definitely have to wait now!
Do you have a shipping date?

[hgg]

mojo-chan my only experience with Uni-T is the UT61E multimeter and I have to say that
the build quality is excellent, the accuracy is great especially if you consider the price range
and the probes are exceptional again for the price range.  Only negative is that they are a
bit stiff. They have an internal resistance of 0.06 Ohms.  I bought the Fluke TL175 probes and
they have a 0.05 Ohms...  I paid half the price of UT61E for the TL175 and I only got slightly
better probes.  Also Fluke does not mention anywhere the internal resistance of the TL175!
Not very professional I think.  If you want to measure low current its a very important factor.

Maybe other Uni-T products are indeed of bad quality.

[Hydrawerk]

Poor Uni-T has only one vote...  but maybe this is because nobody has any experience with it yet.

For some reason, UNI-T scopes are not much sold in English speaking countries. UNI-T doesn't update their English website. http://www.uni-trend.com.hk/
You must use their chinere website to get up-to-date information. What the hell? http://www.uni-trend.com.cn/cp.asp Even Owon has a better english website. 

[tinhead]

Watching two signals which have no relation is always difficult with an oscilloscope. Dual timebase is not triggering
on two different signals. That makes no sense because where is your reference point?

in case of these DSO the dual timebase is addition to ALT trigger, so you do have two referece points and
each of them can be set to own timebase/trigger. The word "dual timebase" can be easy misunderstood
by people knowing the "dual timebase" from analog scopes, which was complettly different thing.

More interessting is to know how far the dual timebase can be used, e.g. is 2ns/DIV for ch1 and 2ms/DIV for ch2
working smooth? Are there any relais clicks when set to such big difference? Is there visible difference (delay in chaniging between channels) when timebase for both channels is equal and when set to 2ns/DIV for ch1 and 2ms/DIV for ch2 ?

[tinhead]

The results are interesting indeed!

Rigol is clearly the winner but maybe this is because more people have worked with it.
It is the safest choice.

it is not the best choise, it is because Dave said "Rigol DS1000 is good", search the forum (or even other forums) and you
will see many such statements. What ppl didn't understood he didn't said "others are bad", because he haven't tested
others in this price class. Probably the worst i saw was as someone posted Dave's DS2xxx vs DS1xxx video comparision
and did use it as "proove" for DS1xxx quality ... where i'm asing myself where to f* the poster saw in that video something
good in DS1xxx? I saw only what BETTER and MUCH BETTER on DS2xxx and how the technology changed and the
bung for buck value - but not a single statement that DS1xxx is still usefull or BETTER than other chinese DSOs.

Anyway, i and others tested other chinese DSOs, where of course everybody have (regardless of the technical aspects)
some own preferences :

e.g. Dave likes knobs, so he have no problems to use single knob for two channels, where i don't like to spend time on
playing with knobs/buttions where it is not necessary - therefore separate knobs for me are the best.
marmad likes the thin Owon enclosure and don't like the Tektronix (and all the likes, e.g. Hantek/Tekway) where i
like it more when the DSO can't be moved by single button push; others like other things etc.

Before i decided to take Tekway DST1102B, i tested UNI-T (UT2102CE), Rigol (DS1102E) and Atten (ADS1102CE).
Where Rigol's bad point was primary display and the single knob crap, other failed with so many things that i even didn't
wrote it down. But that was YEARS ago, technology changed, so i would not say today "don't buy newer ATTEN/Siglent or
UNI-T because models i test are bad". Later i tested Tonghui (today old model already), Owon SDS and even
Rigol CA (due the price change, it wasn't in my price scope before) - not that i was about to buy another DSO, i was only
curious what other can/how they build. I did posted my findings in their own threads (if available, or simply somewhere),
some of my bad impressions has been even fixed by manufacturers (like UNI-T crap ADCs, UNI-T PCB<->PSU distance,
Owon crosstalk, Owon PCB layout and interferences from swithcing converter, tons of Hantek/Tekway things, etc.).
Others here (and other forums) did similar things (review/teardown/simple tests) as well, so lot of informations is available,
unfortunatelly not all current models has been tested so there are questions with no answer (like UNI-T 150k wfms/s ^^)

[EEVblog]

rf-loop thank you for the countless pages of unbiased information!

Unbiased? He is a Siglent/Atten reseller, who in the past even ran a few sockpuppets here to promote his business. That and rather strange claims about Siglent quality.

Yes, rf-loop was posting under multiple accounts on this forum having conversations with himself about Siglent gear! 
That other account was banned.
The only reason rf-loop's main account wasn't banned is because he has and does provide good technical detail here.
Now that he's been "exposed" for those sock-puppet accounts, hopefully it won't happen again.

[EEVblog]

it is not the best choise, it is because Dave said "Rigol DS1000 is good", search the forum (or even other forums) and you
will see many such statements. What ppl didn't understood he didn't said "others are bad", because he haven't tested
others in this price class.

Correct.
And bare in mind that my Rigol review is now over 4 years old!
I cringe every time someone points to the review as if it's a current viewpoint.

e.g. Dave likes knobs, so he have no problems to use single knob for two channels, where i don't like to spend time on
playing with knobs/buttions where it is not necessary - therefore separate knobs for me are the best.

No, I greatly prefer separate knobs for each channel!

You have to remember that when I reviewed the Rigol it was by far the best value scope on the market, bar none, by a country mile. They had no competition.
Now there are at least 3-4 others scopes with better (on paper at least) specs for the same price.

[sorin]

...having conversations with himself about Siglent gear!

Are you serous?
I understand multiple accounts but "having conversations with himself", a normal person can't do that.
Maybe are 2-3 different persons from the same PC

[Wuerstchenhund]

I understand multiple accounts but "having conversations with himself", a normal person can't do that.

Sure, it's easy (and an often used tactic to manipulate discussions on forums).

[hgg]

Yes, rf-loop was posting under multiple accounts on this forum having conversations with himself about Siglent gear! 
That other account was banned.
The only reason rf-loop's main account wasn't banned is because he has and does provide good technical detail here.
Now that he's been "exposed" for those sock-puppet accounts, hopefully it won't happen again.

So its true.  Not very nice...  Maybe he regretted doing it.  We all make mistakes sometimes.
Anyway,

Dave please HEEELP!   
You have to do a review on the Siglent CML series!

People might save some money.  If it turns out good people will save some money by buying something
relatively cheap and fully functional.  It it turns out bad people will again save a lot more!
A win-win situation... except from you that will have to find one Siglent unit...   

[rf-loop]

Yes, rf-loop was posting under multiple accounts on this forum having conversations with himself about Siglent gear! 
That other account was banned.
The only reason rf-loop's main account wasn't banned is because he has and does provide good technical detail here.
Now that he's been "exposed" for those sock-puppet accounts, hopefully it won't happen again.

So its true.  Not very nice...  Maybe he regretted doing it.  We all make mistakes sometimes.
Anyway,

Dave please HEEELP!   
You have to do a review on the Siglent CML series!

People might save some money.  If it turns out good people will save some money by buying something
relatively cheap and fully functional.  It it turns out bad people will again save a lot more!
A win-win situation... except from you that will have to find one Siglent unit...   

The truth, and the truth which has been slightly modified.

Peoples can read truth about what I have talked using "aghp" account. (As long as they stay unmodified.)  [url]https://www.eevblog.com/forum/index.php?action=profile;u=10637;area=showposts;start=0/url]


Some place may also find what leads in history to this situation.
Two accounts without administrator accept was my mistake. Original purpose for this was much more simple than try make some kind of marketing work.  aghp is distributor.  One wrong word and some "xxxxx" company shoot me down. rf-loop (originally) can tell also things what distributor can not.  But perhaps this is too difficult to understand.

But if look content of aghp messages...   please do not read all, there is veeery dirty things.
Of course these messages are good reason to immediately shut down and ban or...
 

I have strong suspect it was not true reason for ban. But there was nice and easy reason to do it.
On this forum here is, and have been, many kind of peoples who have really been very difficult for other people but for those matters moderators tend to look the other way or sleep. Perhaps it is even entertainment for some individuals. It is easy to see adopted.
But I did a crime which apparently was extremely bad.

I do not defend the act, however, in my view, but the truth is the best.

If these words are now enough reasons for ban rf-loop then it is. Then I know more about this...

[hgg]

Ok,
I think I know the perfect way to make Dave review the Siglent CML oscilloscope.

All we have to do is throw a broken one in the Dumpster !!!     

[rf-loop]

Watching two signals which have no relation is always difficult with an oscilloscope. Dual timebase is not triggering
on two different signals. That makes no sense because where is your reference point?

in case of these DSO the dual timebase is addition to ALT trigger, so you do have two referece points and each of them can be set to own timebase/trigger. The word "dual timebase" can be easy misunderstood  by people knowing the "dual timebase" from analog scopes, which was complettly different thing.

More interessting is to know how far the dual timebase can be used, e.g. is 2ns/DIV for ch1 and 2ms/DIV for ch2 working smooth? Are there any relais clicks when set to such big difference? Is there visible difference (delay in chaniging between channels) when timebase for both channels is equal and when set to 2ns/DIV for ch1 and 2ms/DIV for ch2 ?

Here in these images what you ask, and litlebit more.
Relay clicks. No. In this mode both channels its own 500MSa/s ADC group in all time/div settings.

Between 2.5ns -100ns/div all is ok.
If other channel is in this speed area and other channel is 250ns/div or more slow, it produce systematic 2ns shift in other channel. I have not (of course) tested all possible settings combinations but it seems just systematic. Somehow it loose one sample period (2ns). This is bug.
(if need, it can correct using skew adjustment but then if forget it on it may lead some time to error. Of course if  in ALT mode one channel is example 10ms/div then other channel update rate is also slow. It is ALTernative mode.

Bot channels trigger can select free separately, both channel time/div can select free, both channel can selected for math (but there is only one Math process so ther can not be dual math process)

There all images are from ALT mode!
Most of images signal is equal to both channel. (One signal splitted to both channel and made so that there is minimal time difference(<<50ps)
In last picture (bottom picture in "8_SDS1102CML_ALT_........"
There CH1 sinewave is fully unsyncronized with CH2 signal. They come from totally separate sources, and source equipments not  connected to workshop frequency reference distribution.

[marmad]

Bot channels trigger can select free separately, both channel time/div can select free, both channel can selected for math (but there is only one Math process so ther can not be dual math process)

Nice; it's a pity it's not a standard feature on DSOs. So they're using individual, non-interleaved ADCs on each channel?

[rf-loop]

Bot channels trigger can select free separately, both channel time/div can select free, both channel can selected for math (but there is only one Math process so ther can not be dual math process)

Nice; it's a pity it's not a standard feature on DSOs. So they're using individual, non-interleaved ADCs on each channel?

There is 10 ADC clocked with 100MHz. In dual channel mode there is 5ADC + 5ADC and both groups of 5ADC are internally interleaved (5 + 5 phase shifted clocks). In one channel mode with high speeds these are combined to one 10ADC group and interleaved. (10 phase shifted clocks)
It is this "classic" consept as example in Rigol DS1000E (10x ADC) and as is in Hantek 5000B (there is 8xADC). These 10ADC is 5 pcs double ADC IC.  I do not know sure but looks like same ADC what Rigol and Hantek also use. 
If they use same ADC family what is used in Owon SDS and Rigol DS2000 result can be better (and littlebit more expensive solution)

[tinhead]

Here in these images what you ask, and litlebit more.
Relay clicks. No. In this mode both channels its own 500MSa/s ADC group in all time/div settings.

...

Bot channels trigger can select free separately, both channel time/div can select free

that looks indeed good, thanks for pictures/clarification.

[Hydrawerk]

There is 10 ADC clocked with 100MHz. In dual channel mode there is 5ADC + 5ADC and both groups of 5ADC are internally interleaved (5 + 5 phase shifted clocks). In one channel mode with high speeds these are combined to one 10ADC group and interleaved. (10 phase shifted clocks)

Not sure if it's much important, but for some reason I prefer scopes with real one chip ADC 1GS/s per channel.

[nctnico]

I agree. At high sample rates the jitter must be closely controlled so supress noise. With 10 interleaved ADCs and shifted clocks it will be very hard to keep the jitter below a few picoseconds.

[Wuerstchenhund]

Not sure if it's much important, but for some reason I prefer scopes with real one chip ADC 1GS/s per channel.

The thing is that 1GHz ADCs are still more expensive than ten 100MHz ADCs, and therefore won't be used in low cost DSOs.

At the end of the day, you get what you pay for.

[hgg]

Is there a way to actually measure this jitter on each oscilloscope?

[tinhead]

you don't need to measure jitter, the whole discussion about jitter is simply bullshit.

Why? well, let explain based on Rigol DS1000E.
The clock source (for FPGA) is based on a crystal and two gates (oscilator and driver), this circuit give us max. 2ps jitter. So far everything ok, 2ps aperture jitter and 8bit (oh well, AD9288 give us best case 7.4bit) is good for 250MHz signal. But then we have FPGA, using DCM (or PLL in other similar DSOs), which produce extra 30-50ps jitter. This is already only good enough for 7.4bit by 40MHz signal (ups, so far about 8bit 100MHz DSO). My numbers are based on AN501A and other app notes. However they (app note autors) didn't take into count the fact that Rigol (and all the other chinese DSOs using similar circuit) not using dedicated DCM/PLL clock out pins, but regular i/o pins. In principle we have now 7.4bit by 20MHz. But that would be in perfect world, all chinse manufacturers have to save money, so you will not find enough blocking caps in all low cost chinese DSOs. With combination of this together with cheap PSU you can assume the ENOB is reduced once again, i assume we somewhere at 6bit now. What next, yeah, the crap unshielded PSU in few cm distance from ADCs. Is it all? Shit no, Rigol, and the other manufacturer as well, using overlocked DCs. It is hard to say how bad is it, i did masured only influence on Hantek DSOs (as they clocking faster) and yes, that was not good at all as well.  What else, hmm, ahh, right. Due the FPGA/regular i/o pins situation the skew between clocks is more or less "randomized", hehe, so all manufacturer working with "cross zero" calibration to align data. However, this method works afterall not that great, when sampling with wrong skew the resulting data is dominated by interleave distortion (which you can see on all these DSOs, especially single shot). There are of course other aspects not making these DSOs better, i've made Hantek schematics, "A Helene" made Rigol DS1000E schematic - check both, check pictures of Rigol/Atten/UNI-T/Hantek/Siglent/etc. PCB and you will find that they very similar. Very similar in case of "price driven design" as well, look for me as someone (probably one of the EEs) said "who really care, we reduced already so many things that the whole DSO didn't make sense anymore, so let's cut few other things as well" ^^
I don't remember what DSOs was this, but i saw clearly unpopulaed clock chip and ADCs clocked with FPGA, like if someone decided "later" to produce crap instead of proper designed thing.

Of course, with some software tricks, the "resultion" can be increased a bit, all these manufacturers working with similar tricks.
So the question about jitter is pointless, this are low budget DSOs, simply take one from your list and you will be happy as much as with others.

Btw, these DSOs are better in the case of interleave distortion:
(only when two channels enabled, single channel still interleaved - but clocked by proper low jitter source)
 Rigol DS1xxxCA series
 Owon SDS series
 Siglent CFL series
 Hameg HMO series
 Rigol DS2xxx series
 Instek GDS2000A series
 Agilent DSOX2000 series

Btw, you can easily buy 20k USD DSO and still see lot of interleave distortion, this shows how complex is to achieve good results.

[tinhead]

and here some other interessting informations

[hgg]

Very nice info!

So, jitter is dependent on frequency. The higher the frequency the more the jitter. 

Lets say that we have a known low jitter signal for a specific frequency.  If we feed that to the
oscilloscope under test, use the lowest time base and volts/div and switch on persistence,
wouldn't we be able to measure phase and amplitude jitter with some certainty?

Because jitter is depended on thermal noise, is it possible to minimize it by using for example
Peltier cooling on the ADCs?   Just a thought.   Water cooled oscilloscopes... 

[nctnico]

No you got that wrong. The amount of jitter that can be tolerated depends on the frequency and the number of bits. Imagine you are sampling a sine wave in the zero crossing (where the sine wave's slew rate is at its highest). Jitter causes the sampling point to move a bit so it is not exactly in the zero crossing. If there is a lot of jitter the sampling point will move so far away that the wrong value is sampled which causes a faulty reading.

[hgg]

So, how can we measure the jitter of our oscilloscope?