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

[akis]

Yes the problem is present with running on batteries therefore the PSU/rectifier/transformer etc should be excluded.

[rf-loop]

Yes the problem is present with running on batteries therefore the PSU/rectifier/transformer etc should be excluded.

There are LOT of switch mode regulators inside oscilloscope.
With battery alone used for power there is only one what is not working.
There is two sources for raw 8.4V "main" power. Battery or external 110-240V.
After this there is lot of switch mode power supplies/regulators. Also in power supply module there need also do negative (-7.5V).

Power supply unit produce only raw 8.4V and -7.5V voltages. These are not (in mainstream) used directly but they are used for sub power supplies / regulators around of oscilloscope.

Some of these switch mode circuits produce now high amount of parasitic oscillations/ringing. (Why? Perhaps some components more deep parameters... all know that inductor is example 8.8uH but there is also many other things.. core material, internal parasitic capasitances...etc.. Also perhaps layout changes may have some role.

This whole case need more deep investigation.

Good thing is: These "noises" do NOT pollute signal pathways inside oscilloscope!

My recommendation for users and specailly now before this is solved:

take special care about probing!

We know now that clean signal example via well shielded and matched coaxial and there can not see  any sign of this problem.

And nearly as good situation is with oscilloscope probe IF you do not use probe GND "long wire" "inductor" for signal GND.  From oscilloscope GND arrive lot of "noise" signals ant they trawel via this wire and continue they "travel" to somewhere forward...   This GND wire is inductance for these high frequency components and it produce signal over this GND wire  etc bad efects.

You can reduce this problem if you need look low level signals.
If it is possible do NOT use probe GND wire at all.. take it away and do probing like this attached image, even if you handle low frequency signals. (becouse, part of this noise is high frequency and these need handle so that they do not pollute signal outside of oscilloscope using this GND wire reactance)

With slow horizontal speeds you see just spikes and random "chaos" but in real world they are high frequency "signals".   It can easy see using example other fast oscilloscope and spectrum analyzer. (also what you see depends samplerate becouse with low freq signals you use low horz speed and perhaps low samplearate what affect aliasing with these high freq components of signal (these noise components)

This image is good "thumb rule" how you need probe. And keep this GND connect as short as possible (around same as center pin..)

Image is borroved here

http://www.power-eetimes.com/en/evaluating-a-dc-to-dc-converter-in-the-lab.html?cmp_id=71&news_id=222904246

I have tested and example with 10mVpp 1kHz signal from enough clean signal generator and using Owon probe with 1x setting and scope 5mV/div give clean signal. With GND longwire... can see only  " total chaos".  (fatty trace where noise p-p is around same (10mVpp) as signal itself. )
And more fun. If probe 10x and signal 100mVpp  noise is nearly also 100mVpp.  Of course then what you see depends samplerate.

With short GND noise is gone.


This is NOT solution for this problem. But in many cases before real help, with this you can walk in many cases.

This just like first help before doctors come...

[GEuser]

 
After reading that i think the Doctor will be there quite soon imo ...

[rowbo]

Just purchased (Feb 2013) a OWON SDS7102 and I am also having the new noise issue on both channels.  As soon as I turned the unit on for the first time I noticed the background noise on the traces.
Its always present regardless of voltage setting, earthing , probes, no probes.  Amazing units have got out of factory with this issue.

Anyone have any further news on solution, .... it is still holiday in China, this may mean a big wait.
I guess I will need to contact the supplier about the issue as well.
 

[bill2010]

Just purchased (Feb 2013) a OWON SDS7102 and I am also having the new noise issue on both channels.  As soon as I turned the unit on for the first time I noticed the background noise on the traces.
Its always present regardless of voltage setting, earthing , probes, no probes.  Amazing units have got out of factory with this issue.

Anyone have any further news on solution, .... it is still holiday in China, this may mean a big wait.
I guess I will need to contact the supplier about the issue as well.
 

If you are seeing the noise even when using a fully shielded coaxial cable connection to the signal source then you are experiencing something different than what my unit is doing. Either way, agree that QC does not seem to be Owon's strong suit. Can you post any screen captures for reference?

[akis]

We can be more specific : 120mV pk-pk - how does it display ?

[rowbo]

I guess it is not a huge amount of noise, but it is there all the time.
Surely this is not normally for these scopes?

This image is a bit fuzzy, but all I have handy at present

[AndrejaKo]

The amount of noise seen on the bottom of the trace seems similar to what I get when I use the ground clip on the probe. It disappears with the springy type ground lead. Also my scope should be old enough not to be affected by current power supply issues.

Can you switch to peak mode and zoom in on the noise there?

Also how clean is your environment with respect to electrical fields? Do any other scopes (if you have them) show any similar issues when near the SDS7102? Do you have any computers, telephones, CCFL sources or similar "dirty" devices nearby?

[rf-loop]

Here is some tests how probe GND affect - with scope what have this special "GND noise" issue

Later, after work is more ready I will add information also here. Now just lack of time.

[akis]

I will go and test now. But it seems your test frequency is 20KHz? How about looking at signals at 1KHz or so ? Surely at those frequencies the signal should be crystal clear. Also is your mode peak or sample?

[rf-loop]

I will go and test now. But it seems your test frequency is 20KHz? How about looking at signals at 1KHz or so ? Surely at those frequencies the signal should be crystal clear. Also is your mode peak or sample?

Noise is same what ever is test signal frequency. This oscilloscope GND noise is exactly same - even if test with DC. (Visibility is other case... it depends how you set horizontal speed and what is samplerate and what is sampling mode (with low samplerates peak mode give it)  Using 10-20us/div there can just see separate peaks (these most high level noise peaks). If go more deep it can easy find what these peaks include. They include high frequency oscillations even up to over 200MHz. (with low samplerates and normal acquire mode  these generate of course also aliazing (together how samples are displayed on the screen low resolution related to sampled points), so these displays may be very confusing)

20kHz was selected as compromize for better visibility to noise so that there also can see full cycle of test signal and so that I can easy adjust Owon and Tektronix TDS520 for same sampling rate.  (TDS520A is 500MSa/s  500MHz (repetitive) so becouse there is 500MHz analog front end in some tests there is selected 100MHz BW so that it is more compatible with Owon. If I use full BW Tek looks more noisy than Owon (also without any signal).
Owon analog BW is nearly 200MHz so it show littlebit more noise than Tek with 100MHz BW)  Here you find more tests also about Owon noise and other signal quality things.



About peak or normal. In these test images sampling "Normal". 

Peak do not make so much becouse in first (Scope GND noise test) part there is used maximum samplerate.

Peak mode give more if there is low samplearate. (becouse ADC work always with full speed and low samplerate means just only that more or less samples are dropped out. In peak mode it try find peak values also from these samples what are dropped out.
Example in one channel use ADC work always 1GSa/s  and if displayed samplerate is 100MSa/s it means that it takes every tenth sample and 9 samples just drop out. In peak mode it watch also these dropped out samples and "displayed" sample is (perhaps) highes in this 10 sample group or something like it. (of course it meand also that timing go wrong but this do not matter. (it can see if zoom in (stopped capture) and if know exactly signal and its peaks timing... )

(About this "noisy GND": work for analyzing oscilloscope internal circuits is under work. I have some "findings" but too early to talk more about these. Also priority go so that I will give all information to Owon first and later look what part of information is public  and what is not.)

Things measured without opening oscilloscope and these findings, they are of course public.  This GND noise spectrum is wide. From low freq up to around over 250MHz. This is why also this probe GND method make so big effect.

Oscilloscope have power supply. It is quite new version.  It also produce now more "noise" but this is only small part of this whole story. Noise produced PSU unit can also detect but this is not now this main problem.  I have  tested one  oscilloscope without its own PSU. I have used clean linear power supply for +8.4V and for -7.6V and run scope with it (Owon PSU removed) and result is nearly same.

[rowbo]

Added a couple of images showing the noises issue on scope purchased Feb 2013.  First is peak with probe at 1x and the second with probe at 10x

[akis]

At 2V/div it seems this "noise" is actually huge. Seems it is like a % of the signal, not a fixed amount. I am not sure anymore. Hoping there will be a simple software fix

[rf-loop]

At 2V/div it seems this "noise" is actually huge. Seems it is like a % of the signal, not a fixed amount. I am not sure anymore. Hoping there will be a simple software fix

This signal is probe comp signal. Is it clean signal for testing oscilloscope noise? No it is not. It is only ok for adjusting oscilloscope probe compensation and nothing else. And for this purpose it do not matter if there is some noise or not. It can be even lot of more noisy and still it can use for this kind of simple probe compensation adjustment. 

Also it is good to read all tests what have made for real noise tests and also it is good to understand digital oscilloscopes visible noise. These test results can read in this thread here and other threads here and also in other places. 
Inside oscilloscope in signal pathway there is not noise problem. If you put clean signal to oscilloscope input BNC this Owon can competite with well known good brand oscilloscopes. If some  people have not skills or knowledge and experience (or equipments to do tests)  - it is just other case. It need classroom and teacher.
It is also good  to read example Tektronix and Agilent materials about digital oscilloscopes noise.

Also I have made tests about noise, also used these Owon scopes what have this special "noisy GND" problem and still also these units give same kind of result as before (if keep care that noise do not externally pollute signal).

But now - in other hand -  this  IS real problem! And problem is just this "noisy GND" problem. It is real problem becouse it is difficult to avoid that this noise do not polute signal with normal probing methods. Even for DC there can not use normal probe GND. (becouse there is this RF noise walking  through GND)

Signal pathway inside oscilloscope is clean  -  this problem do not pollute signal pathway inside oscilloscope any markable amount.   Now with this problem this noise is added to signal externally.
I have demonstrated and proofed it many ways.
Also I have analyzed this noise and I can recognize different fractions of extraordinary abnormal noise and where from these come.

There IS problem what Owon need repair!

This problem is high EMI generated by oscilloscope and pollute oscilloscope GND with high amount of spread spectrum high frequency noise. And then externally this noise is "added"  to signal and there is many ways, example probe GND wire inductance is in markable role.

This problem have NOTHING to do with software inside oscilloscope. Nothing!  This is as pure HW problem as ever can be.

I have also showed (documentation include material what is not public at this time  -  it is not "open project") where in HW is main part of this problem.

I can repeat it here agen: Main part of this noise is generated in new version of adapter board inside oscilloscope. If you do not know what is adapter board.. it is board what connects together oscilloscope Main board, PSU and Display.  main problem is this new adapter board version. Other problem is new PSU but its role  is not so high as adapter board related to this noise what we have seen.

One is sure: this problem can not repair with software!

The problem can be corrected and treated only if its existence is known and admitted that there is a problem.

Whole case is now also under hard and busy work. Solution - what ever it is - is coming.

[akis]

One is sure: this problem can not repair with software!

The problem can be corrected and treated only if its existence is known and admitted that there is a problem.

Whole case is now also under hard and busy work. Solution - what ever it is - is coming.

If you say it is the new adapter board, can this board be replaced? Is it modular enough and easy to remove and replace or does it require a return to the factory? Also, can the adapter board be repaired somehow rather than replaced ? Is it a particular component on this board that needs replacing or is it a much bigger issue?

[rf-loop]

1.

If you say it is the new adapter board, can this board be replaced? Is it modular enough and easy to remove and replace or does it require a return to the factory?

2.
Also, can the adapter board be repaired somehow rather than replaced ?

3.
Is it a particular component on this board that needs replacing or is it a much bigger issue?

1.
I say: main reason is new adapter board version (add: there may be less variations between units).
 
Also main PSU is one source and it may have more variations between single units due to very strange parasitic oscillations in -7.6V supply (this works also with battery use).
 
Both can easy change. Scope open, old card or PSU out and new in. But of course it need some "skills" to do it. First need get solution and this is Owon who need find solution and desingn how they handle this case.

2.
It may be possible. If it need change some layout - no way to repair. Only change. If there need add some component it depends what and how. Most easy case is if there need only change some simple components.

3.
It depends what is real solution.

(also earlier version produce noise but now this noise is something totally diffent level.   I thing they have tried develop new adapter board version  better than old.  But something is now wrong... I do not like  public quessing without knowledge what there have happend... it may be even so that someone change some componet type after tests in mass production phase... who knows.)

[alexuresp]

http://youtu.be/6bKDudV2d0E
My video of GND noise.

[rf-loop]

Now some tests and later I will add more information here.

Now it is clear where from all  this extra noise come. (there is several sources)

[bill2010]

Now some tests and later I will add more information here.

Now it is clear where from all  this extra noise come. (there is several sources)

First, let me express my appreciation for all the time you have put into analyzing the problem. I know you sell these units and have a financial interest in their success, but in my mind you've gone above and beyond.  At this point it sounds like we need Owon to develop a solution and plan for fixing the units that have already been sold. I would greatly appreciate you keeping us updated on this process and advising us on how we should work with our suppliers to get our units fixed once Owon has provided an official response.

[Siri]

Greetings to all from Moscow. Take a beginner.
I apologize for my possibly incorrect English. I use a web translator.

I disassembled and deeply researched my SDS7102(V) (s/n 1234xxxx). Power supply 8.4V and-7.6V not the cause of this noise. The problem of "GND-noise" comes from the DC/DC down-converter with an output voltage of +5.5V. Its located in the center of the L-shaped adapter board and operating frequency of about 480 kHz conversion. It has a wide range of harmonics that spread on the ground surface of the device. In my version of the adapter board this chip marked Z1094AI.
Even when on battery power, away from potential sources of electromagnetic interference - I live in a country house and conducted special experiments with the release of the house - the problem remains. If you join the probe to the outer part (GND) of the inputs, it is seen that "needle" noise - RF pollutions start to become numb from the ground to the input of the oscilloscope. Well enough to connect a short piece of wire to the probe (or even just put the factory with a hook at the tip of the probe) - capacitive displacement currents arise from the ground, the ground surface radiated interference caught probe. The same thing happens when testing external circuits with weak signals - the circuit under test begins to work as an antenna.

The best solution I see the replacement step-down converter +8.4/+5.5V to conventional linear regulator. Unfortunately, it will be at the expense of total energy consumption, but will not this annoying problem in all a very good oscilloscope.

[rf-loop]

You have littlebit older SDS.
Can you tell what version of PSU and Adabter board is inside your oscilloscope.

Newest version I know and have inspected are
PSU: T115-N Rev 3.  (partially SMD components but still single layer board)
(also +8.4V and specially -7.6V SMPS radiate high amount of EMI)

Adapter board: SDS7102_CON_V3.2   (2012.07.24)

On adapter board can you look if buck's flywheel schottky diodes are B140.

(note for possible 5.5V mod : in batery use this "8.4V" drops very very down!)

[akis]

I do not get it, why would you use switched mode supplies and converters inside an instrument that should be free of all possible sources of contamination and noise ? Why not use linear supplies with precise windings on the transformers so as to generate the least possible heat ? How hard can it be?

[Siri]

Hello rf-loop,

Power board is the same: PCB-T115-N Rev 3

Adapter board: SDS7102_CON_V3.0 (2011.11.14)


I conducted an audit of the scheme for the noise.
Very small portion of the RF noise produces an inverting converter circuit -7.6V, performed on the chip MC34063. Printed circuit topology of this chip on the SDS has the design flaw: ground printed track of M34063 scheme is relatively narrow (high inductance), a C-shaped.




In this case, the distal end of the C-track for some reason connected to the metal case power switch. As a result of the end of the track is "hot" for the RF-harmonic converter and switch plate - a good antenna for emission of pollution.



 When touched the metal of power switch (GND on the designers' plans) insulated screwdriver when removing the back cover, the level of radiated EMI increases substantially. When you touch the ground with a isolated screwdriver is very close - on the output  power connector of PSU or on Z-plate, this increase in noise is not happening.

At first I thought it was only at this point the problem and did some upgrades. Was cut off from the end of the ground printed track from the switch pins.



The main problem GND-noise persists. After that I added additional MLCC ceramic capacitors (low ESR at RF) on the input and output voltage line MC34063, and also made a wide screen-shield/ground bus on place MC34063 (screen-shield is not in the photo below).



Emission spectrum of a local 34063 changed, it disappeared some high-frequency components, but the main problem as it turned out not that.

Further research led to the bug DC/DC step-down converter +8.4 / +5.5 V. Long story as it was found. That step-down +5.5 V makes a noise and makes this slyly, "needles" switching pulse (differential current) running out on the ground of adapter board. Perhaps this can be reduced by replacing the electrolytic capacitors in the circuit at LOW ESR (chineses save at this point), maybe it will improve another topology .. But I fundamentally output is seen only in the replacement Z1094 linear voltage regulator 5V.

By the way, a little "noise" and following the +5.5 V step-down converter U21 (AST4060A) with an output voltage +3.3 V, but his contribution to the "GND-noise" is not so big.

Depending GND-noise on the battery voltage (6.6 ... 8.4V), I had not noticed. Output voltage of the converter +5,5V does not change and noise equally.

[bill2010]

I do not get it, why would you use switched mode supplies and converters inside an instrument that should be free of all possible sources of contamination and noise ? Why not use linear supplies with precise windings on the transformers so as to generate the least possible heat ? How hard can it be?

Two words, size and cost.  A linear supply with the same current rating would be heavier and cost more to produce. Efficiency is probably a distant third since you need more voltage margin on a linear to achieve the same level of regulation.

[rf-loop]

Depending GND-noise on the battery voltage (6.6 ... 8.4V), I had not noticed. Output voltage of the converter +5,5V does not change and noise equally.

Nice work Siri!

I did not mean noise as I told that notice about battery. I mean if think linear regulator for 5.5V
Battery terminals shut off treshold is as low as 5.2V. (it may littlebit vary between single batteries (this control is inside battery) When it is just below 6V  oscilloscope still works normally and it still do not crash even if battery terminal drops to 5.2V.

Your adapter board is different.