REVIEW - Owon SDS7102 - A look at the SDS series from Owon

[patb]

Big thanks for the review part 1.

UPDATE:
I've just read your last post and the firmware of this DSO doesn't look good at all. I only hope they still consider it as work-in-progress, and not the final version. I am also a little bit worried about hi-speed signals on Owon, but i'm not expert here. Usability was probably on the bottom of the list when they designed this product. What a shame. Btw, have you contacted Owon yet regarding all the issues?

[tinhead]

No Equivalent Time Sampling available.  That kind of sucks - I just assumed all modern DSOs had this - it's very handy for fast repetitive signals in digital circuits.

right, but this makes only sense when the input stage/bw filter is really allowing to have a benefit. For example a cheap
60MHz Tonghui DSO (with -3db on 75MHz) when sampling with 400MSs can show perfectly 300MHz signals in equivalent
mode (of course with reduced amplitude) where on 200MHz Hantek sampling with 1GSs it does not matter what you enable - the waveform looks always (almost) the same - even if i remove bw filter (or set it to 900MHz) equ. sampling is not really
working as expected. So well, is it importnat? Depends, i can work with avg sampling too.

No way to turn off sin(x) - as mentioned above.

i personaly never used linear interpolation on the TDS754D, so i'm not missing it on the HanTekway.
But for sure, additional feature is always welcome.

When using the frequency counter (on the trigger channel), the count becomes more and more incorrect the further the trigger level is moved below the zero line of the waveform.

the freq. counter should in principle work up to plus/minus Vrms trigger level for sinus wave (at least ±354mV for 1Vpp) ,
if not then it's not good enough.

[alm]

I have tons of digital circuitry with which I could generate, at the very least, fast, repetitive clock signals.

Try to find a signal with a very fast rise / fall time (eg. 1ns), but preferably a fairly low repetition rate (eg. 1MHz). This gives you a very broad range of (odd) harmonics, and is a good measure for frequency response. Since you don't have a faster scope, you can't verify the actual shape, but you can compare the two scopes.

Although the 100MHz .pngs from the Owon don't look too bad, what is odd is that I have the display set to dots, not vectors, and the sine wave coming from the Hantek is exactly 2 points per period; i.e. a triangle wave.  Why is the Owon showing a sine?  Is it because I can't turn off sin(x) on the Owon scope?  The 2 .pngs after the 100MHz screen caps show the saved 100MHz waveform brought into the Owon PC software - two times saved, one after the other - with different sample points on the waveform.  Again, I'm not sure if this is due to the instability of the DDS3X25 at this high frequency, but it doesn't look good to me.

Seems like the signal is indeed under sampled. Is it sampling at 500MS/s? Does turning off one channel help? Note that a signal near its rated bandwidth is expected to look like a sine. The rated bandwidth is the -3dB point, and it will roll off with about -6dB/octave (sometimes even faster). That means that the fundamental will be about 30% down, and the harmonics even more. Attenuating the harmonics makes any signal look sinusoidal.

No Equivalent Time Sampling available.  That kind of sucks - I just assumed all modern DSOs had this - it's very handy for fast repetitive signals in digital circuits.

Apart from clock, few digital signals are repetitive unless you can change the program or can trigger on bit patterns. ETS is often poorly implemented in cheap scopes IMO. To make use of the extra samples, the trigger stability needs to be very good. In a modern scope, real-time sampling rate is usually good enough for the analog bandwidth. Not much point in sampling a 100MHz sine at 10GS/s.

[marmad]

right, but this makes only sense when the input stage/bw filter is really allowing to have a benefit. For example a cheap 60MHz Tonghui DSO (with -3db on 75MHz) when sampling with 400MSs can show perfectly 300MHz signals in equivalent mode (of course with reduced amplitude) where on 200MHz Hantek sampling with 1GSs it does not matter what you enable - the waveform looks always (almost) the same - even if i remove bw filter (or set it to 900MHz) equ. sampling is not really working as expected. So well, is it importnat? Depends, i can work with avg sampling too

Apart from clock, few digital signals are repetitive unless you can change the program or can trigger on bit patterns. ETS is often poorly implemented in cheap scopes IMO. To make use of the extra samples, the trigger stability needs to be very good. In a modern scope, real-time sampling rate is usually good enough for the analog bandwidth. Not much point in sampling a 100MHz sine at 10GS/s.

Well, working with µCs, it's easy to change firmware to generate repetitive signals for debugging - but you both make good points since I didn't realize ETS was poorly implemented on some (many?) of these cheaper scopes.  I tested it on the Rigol with somewhat higher frequency signals and it seemed to work fairly well.

the freq. counter should in principle work up to plus/minus Vrms trigger level for sinus wave (at least ±354mV for 1Vpp) ,
if not then it's not good enough.

Strange, I haven't been able to replicate that freq.counter bug again - it definitely was happening when I was measuring 2 channels - I played with it for awhile wondering what the hell it was doing... but the last few times I've tried to repeat it - I can't.  So it's certainly happening with some combination of settings - but not others.

Here are a few other .pngs - following a suggestion from tinhead (@tinhead - I could only check noise level at 5mv/div - at 2mv/div setting 20MHz filter is on automatically), plus some reading on interleave distortion. I think the file names are self-explanatory.  Note that the single shot 0.1MHz, 1MHz, 10MHz, and 50MHz sine waves used by the FFTs are generated by the Hantek with 2000, 200, 20, and 4 points respectively.

BTW, from all I can tell, the Owon does sample at it's full speed (1GS/s single - 500MS/s double) when using the whole 10Mpts.

[tinhead]

Here are a few other .pngs - following a suggestion from tinhead (@tinhead - I could only check noise level at 5mv/div - at 2mv/div setting 20MHz filter is on automatically)

yeah "it's not a bug its'a feature", you have probably to disable it manually again.
I did patched away that's "feature" on HanTekway, if i wish to enable something then only if i wish.
Why some developers can't understand that something like autofilter need to be implemented
as user selectable option, that's 10 lines in code.

BTW, from all I can tell, the Owon does sample at it's full speed (1GS/s single - 500MS/s double) when using the whole 10Mpts.

so you did single shot, zoomed out and the visible record length is 10ms when in single chan mode
and 2 x 20ms when both chans enabled ? If so, then that's an overkill for many other DSOs.

EDIT: i forgot to mention something, Owon can do an CSV export, look on the header of that file (or count of data lines),
even if we think it is sampling with 1GSs and does have 10Mpoints it can be still 800MSs and 8Mpoints - both are resulting
in exact the same record length.

[saturation]

FYI marmad on these issues:

I do get a reproducible 75 MHz out of the 3x25 with the factory software viewed on the Rigol.  Its fairly similar to what you've posted.  Here from an old post on the 3x25 thread:



Curious on the Owon sine at 100 MHz.  On the Rigol, now using mechatrommer's software, the reduced data points does show in the interpolation as a distorted sine, here's mechas screen grabs:

I could not replicate the 75MHz result I had with the Rigol on the Owon.  I'm not sure if this is due to the fact that 75MHz from the Hantek is not particularly stable - and I managed to get lucky with the Rigol - or, more likely, problems with the Owon.

 Why is the Owon showing a sine?  Is it because I can't turn off sin(x) on the Owon scope?  The 2 .pngs after the 100MHz screen caps show the saved 100MHz waveform brought into the Owon PC software -

No Equivalent Time Sampling available.  ...

[Mechatrommer]

I do get a reproducible 75 MHz out of the 3x25 with the factory software viewed on the Rigol.

how do you produce that clean 75MHz signal out of 3x25? what factory software?

[marmad]

so you did single shot, zoomed out and the visible record length is 10ms when in single chan mode and 2 x 20ms when both chans enabled ?

@tinhead - Yes, that's exactly what I did.  But let me tell you, the 10Mpts of memory, just like the gorgeous display, is another example of nice hardware - which is not taken full advantage of by the software.  When you are 'zoomed in' on a large record (small s/div), the encoder knob still moves you horizontally at the same speed through the record as always - it takes forever to move around it - and it's easy as pie to get lost.  Why not, for example, if the scope is stopped and you're just browsing the data, have the encoder movement different and faster for zooming and jumping through the long record - as well as change the display feedback to better locate your position? Also, the Owon PC software is crap, so trying to look at these long records there is tedious - but that I could rewrite, so it's not such a problem.

If so, then that's an overkill for many other DSOs.

Sure, for most things, but if you need to log data over time, it can be a godsend.

Curious on the Owon sine at 100 MHz.  On the Rigol, now using mechatrommer's software, the reduced data points does show in the interpolation as a distorted sine

@saturation - yes, the 3X25 only uses 2 points to represent a 100MHz sine wave so it's actually more like a perfect triangle wave. But since I can't switch off sin(x) on the Owon, there's no way for me to see the actual data points it is sampling (it always shows the interpolation) unless I save the waveform and take it somewhere else to look at it.  I wish I had a high quality function generator to really test this scope with.

I tell you, this has to be the most frustrating thing I've purchased lately - leaving aside the unknown electronics for the moment - except for the stiff membrane switches and the slightly cheap-feeling smaller encoder knobs -  the design, ergonomics, and hardware (at least on the outside) are attractive and very appealing.  Who owns a cheap Chinese DSO but doesn't want a thinner desktop model, or an 8" 800x600 screen, or an optional 4 hour lithium battery, or a whisper quiet fan, or a smarter trigger pass output, or VGA output (it's great to see the waveforms really huge from a distance away when probing), or 10Mpts of memory?  You can't use this Owon and not feel that this is the direction many of the Chinese DSO makers will be moving in the future.  But the firmware is discouraging and there is so much untapped potential. 

I have to decide tomorrow (or Friday morning at the latest) if I will keep the Owon or return it and go for the Hantek.  It would be kind of a shame to relinquish the flash new hardware, but I can see by all I've read that the Hantek firmware is better (and the electronics are proven) - and even though I would end up with a DSO without as many initial bells and whistles, I'd have one which is less frustrating and probably more useful in the long term.

[saturation]

Sorry mecha, its not without postprocessing, its using the original Hantek software with averaging.  Here's are the full settings:

Analog Ch  State   Scale    Position   Coupling  BW Limit  Invert
CH1        On     50.0mV/   16.0mV     DC        Off       Off

Analog Ch  Impedance   Probe
CH1        1M Ohm      1X

Time    Time Ref    Main Scale    Delay
Main    Center      5.000ns/      0.000000s

Trigger  Source      Slope    Mode      Coupling     Level    Holdoff
Edge     CH1         Rising   Auto      DC           -16.0mV  500ns

Acquisition    Sampling    Averages    Memory Depth    Sample Rate
Average        Realtime    256         Normal          500.0MSa   

Math   Source  Window    Display Screen    Scale
FFT    CH1     Rectangle      Full        10.0mVrms/


The last two images are your software with averaging, and without averaging but at 100 MHz.

I do get a reproducible 75 MHz out of the 3x25 with the factory software viewed on the Rigol.

how do you produce that clean 75MHz signal out of 3x25? what factory software?

[bertchai]

A stupid question to ask.... 
So everyone like Hantek more?  Any comments on Atten like ADS1102CML?

I am looking to buy my first 100MHz DSO.....

[marmad]

I changed my opinion about the measurement system on the Owon (Note to myself: RTFM); mistake due to a stupid visual assumption on my part - and have softened somewhat to the menu system.  I've edited the original post, back one page in the thread.

Here are some new .pngs of 1MHz, 50MHz, 80MHz, and 200MHz LVCMOS square waves, made single channel, full sampling rate (except the 1MHz = 500MSa/s) and two FFTs.

Note: Check out the measurements for the 200MHz signal. It lists '?' for frequency in the screen capture (although the counter lists the frequency correctly) - strangely, on the actual screen while running, it lists '200MHz'.  Also, the rise time and fall time are listed around ~1.5ns.

Again, after the various hacks around here, one has to wonder how much is actually different between this scope and the SDS8102; the next one in the line.

[marmad]

Usability was probably on the bottom of the list when they designed this product. What a shame. Btw, have you contacted Owon yet regarding all the issues?

Well, I've softened my view a bit about the firmware (see previous post) - it still has problems for sure - and some features are missing I wouldn't mind having - but I wouldn't go so far as to say usability was at the bottom of their list.  Many of the nice hardware features (such as VGA out) are great for usability.  Anyway, I'll try to cover all these things in part 2 of my review of the device.

Btw, have you contacted Owon yet regarding all the issues?

No reason to at this point - I can still return the scope for a refund if I like.  If I decide to keep it (and I'm still testing and making up my mind), I'll send them my suggestions and any bugs I find - but it appears (from what I can discover online) that Owon doesn't offer firmware upgrades directly to users - they instead follow the policy of dealer upgrades (ship or carry-in plus service charge) - so I doubt they'll make an effort to change 'usability' aspects of the firmware.  If I keep the scope, perhaps I'll try to change the firmware myself.

[Bren]

Well, after reading some of your (marmad) posts, I think I've decided against the SDS7102. The Hantek/Tekway 100MHz looks like the way to go, you get a scope with much more functionality and with some time and effort you can double the BW.


Love the SDS 10Mpts, 8" TFT screen, Trigger Out and VGA Out, 100MHz, 500MS/s (dual channel), Ultra thin design.
But with all these seemingly great specs, the typically "selectable" options like sin(x)/x, and 20Mhz BW Limit, being mandatory?

I like to tell people what I want on my pizza. Not the other way around.
Sad that the SDS looks more like a work in progress than a finished product to me, and it may all be because of firmware.


On that note,
Anyone know where to get a Hantek DSO5102BM or Tekway DST1102B cheap in Canada?

[marmad]

But with all these seemingly great specs, the typically "selectable" options like sin(x)/x, and 20Mhz BW Limit, being mandatory?

Two things:  linear interpolation is something you rarely use - it certainly can be handy for checking sampling around the bandwidth region, but not something you use often in practice.  Also, as tinhead mentioned, the Hantek/Tekway doesn't have it either.

The 20Mhz BW limit only came on automatically at the lowest horizontal setting: 2ns.  And, strangely enough, I haven't been able to repeat that happening as well.  So some combination of settings I had (and I had both channels on at the time) triggered it - but it's not doing it anymore.

[marmad]

Added speed calculations of USB to PC transfer one page back at original post.