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

[TomC]

Hi Sarasir,

I took the time to read several articles on shielding theory, Faraday cages etc. I was expressly looking for something that would substantiate your theory that nothing, including conductive common mode currents can scape the shield. That only seems to be possible on a perfect shield (Faraday cage) with no openings which is hardly the case here. Everything I see reinforces the theory that shields are effective for radiated emissions, but can't stop the conductive components already present on the interface cables going in/out of the enclosure.

So I don't know what else to tell you!

Just a suggestion, can you try finding another source that substantiates your theory and post it! Of course I understand that this may not be possible if you think that this is an original discovery that no one ever thought of before.

As far as my theory is concerned, that's all it is, a theory, and believe me, it doesn't bother me at all if I'm wrong, that usually helps me retain the right answer/theory better!

And I don't doubt your technical ability and qualifications! I just can't come up with a different explanation for the experiment results, so at the risk of being embarrassingly wrong by persisting, I still have to say that based on this latest research my opinion still hasn't changed!

[rf-loop]

Hi TomC,

what you mean by external source? The time I made the pics no other noise was present with the Owon disconnected from mains (the terrible noise I had the other day never came back again after it disapeared the third day. Indeed must have been some HAM radio guy or whatever.).

All noise you see in the pics is generated inside the Owon, absolutely no doubt.

Theoretically the added 475 input capacitance of 20pF might reduce the displayed noise a little but I don't think that is significant. And the set up is the same with and without the oil can so why dealing with it?

The question why noise goes down when connecting the 475 to the setup was just btw. The same thing might happen as well if connecting a cubic of solid iron of 1m x 1m x 1m to the setup common ground. Must have something to do with the energy needed to drive a big mass - or antenna as well if you want - with HF.

Only the difference between the pics taken with the Owon in and out of the can counts in my opinion.

The displayed noise of the 475 shows just one snap shot of the noise due to it's more precise triggering whereas the Owon's display is rather unstable, has a tendency to trigger on different spikes each sweep.
But I haven't closely checked it's trigger abilities yet.

I do not know exactly how you have connected Owon+Tek parallel to signal generator  but I believe this connection add much more than just pure 20pF Tek input capacitance. So, this connection may reduce much more bandwidth than may expect if think only Tek input 20pF. You can try. Connect fast edge instead of this sinewave and yoy can see high reduction in risetime, and this Network may also peaking with several frequencies die to its possible impedance mismatching.

Just for clariy. If connect 50ohm cable to 1MOhm input and look 50ohm cable free end. There is this cable cpacitance and inductance "Network".

Also I note that you use 2mV/dif in Owon. There is 20MHz BW limit. This noise frequency components have lot of over 20MHz frequency components what are now highly attenuaated and peaks are much lower than they really are. 
If you want see around same with Tek oscilloscope and Owon ypu need do impedance matched real splitting Network from Signal gen output parallel to both scopes. Other ways can not see any real data.  But of course also this kind of simply test is useful.

Other thing is very big difference if we think how analog oscilloscope show noise what is "random" with scope eyes. This is real thing what have discussed and studied in many places. Owon itself can show  much more noise than this Tektronix. It can proof if you look some signal sou´rce what produce some known signal added with some random noise peaks.

Fact, I believe, is of course that this noise in this case is coming from Owon.

Btw, for trigger stability there is adjustment in Owon trigger settings.  (HF reject)




---------
About noise displaying with digital scope or analog scope.
(this is partially for these peoples who do not know and randomly read this topic and look these pictures where Tek analog scope do not show nearly any noise (or only small amount) and Owon screen is full of noise.



Try with 5ns width pulses  with 10ms period. What you can see on Tek275 display and what can see on the Owon display?
On the Tek display nearly nothing or nothing and Owon show  just clean full pulse without any problem.  Same for noise, depending noise characters and what signal is trigged.

[Sarasir]

Hi Sarasir,

I took the time to read several articles on shielding theory, Faraday cages etc. I was expressly looking for something that would substantiate your theory that nothing, including conductive common mode currents can scape the shield. That only seems to be possible on a perfect shield (Faraday cage) with no openings which is hardly the case here. Everything I see reinforces the theory that shields are effective for radiated emissions, but can't stop the conductive components already present on the interface cables going in/out of the enclosure.

So I don't know what else to tell you!

Just a suggestion, can you try finding another source that substantiates your theory and post it! Of course I understand that this may not be possible if you think that this is an original discovery that no one ever thought of before.

As far as my theory is concerned, that's all it is, a theory, and believe me, it doesn't bother me at all if I'm wrong, that usually helps me retain the right answer/theory better!

And I don't doubt your technical ability and qualifications! I just can't come up with a different explanation for the experiment results, so at the risk of being embarrassingly wrong by persisting, I still have to say that based on this latest research my opinion still hasn't changed!

Hi TomC,

the shield is connected only at one point to the Owon and that is at the input BNC(s).
So all that is conducted by the shield is radiated emissions from the Owon (of course the shield will also sense all radiation from outside e.g. from mobile phone transmissions to TV broadcasts but those levels are far down compared to the noise we are dealing with - at least here where I'm working, and e.g. my Mobile is always off when I'm at home)

Why do you say hardly? The only open hole in the oil can shield is for the power cord and on this cord I have put 4 cores, with the cord wound around one time on two of them so to pass it through twice.

The other hole is closed by the BNC coupling that also connects the shield to the BNC connector.

Even though the shield is not perfect - just thin iron tin and not silver anodized - it brings down the noise to a level where it almost can not be seen anymore on the 475 whereas before the spikes where quite high.

And I never said that nothing can scape the shield! It's always a question of power what can go through what and what not.

And it's certainly not my invention. As I already mentioned inside of professional machines this principle is utilized time and again.

And I do not need any source to proof anything if I see that my theory works. Yes, sometimes it is good to read that you're on the right way or you might get some additional ideas.

Sorry, but I have to question the picture you have of my qualification because again you came up with this external noise crap implicating that I'm not able to verify whether the room here is clean or not. One time you may ask or say that but second time it's almost an insult for me.

And in general, of course it's not a problem at all what you beliefe or not, that's absolutely up to you and I'm not angry or whatever. It's just fun to fight for the truth and through our discussion my picture of the discussed mechanisms just clears up more and more.

[Sarasir]

I do not know exactly how you have connected Owon+Tek parallel to signal generator  but I believe this connection add much more than just pure 20pF Tek input capacitance. So, this connection may reduce much more bandwidth than may expect if think only Tek input 20pF. You can try. Connect fast edge instead of this sinewave and yoy can see high reduction in risetime, and this Network may also peaking with several frequencies die to its possible impedance mismatching.

Just for clariy. If connect 50ohm cable to 1MOhm input and look 50ohm cable free end. There is this cable cpacitance and inductance "Network".

Also I note that you use 2mV/dif in Owon. There is 20MHz BW limit. This noise frequency components have lot of over 20MHz frequency components what are now highly attenuaated and peaks are much lower than they really are. 
If you want see around same with Tek oscilloscope and Owon ypu need do impedance matched real splitting Network from Signal gen output parallel to both scopes. Other ways can not see any real data.  But of course also this kind of simply test is useful.

Other thing is very big difference if we think how analog oscilloscope show noise what is "random" with scope eyes. This is real thing what have discussed and studied in many places. Owon itself can show  much more noise than this Tektronix. It can proof if you look some signal sou´rce what produce some known signal added with some random noise peaks.

Fact, I believe, is of course that this noise in this case is coming from Owon.

Btw, for trigger stability there is adjustment in Owon trigger settings.  (HF reject)




---------
About noise displaying with digital scope or analog scope.
(this is partially for these peoples who do not know and randomly read this topic and look these pictures where Tek analog scope do not show nearly any noise (or only small amount) and Owon screen is full of noise.



Try with 5ns width pulses  with 10ms period. What you can see on Tek275 display and what can see on the Owon display?
On the Tek display nearly nothing or nothing and Owon show  just clean full pulse without any problem.  Same for noise, depending noise characters and what signal is trigged.

rf-loop I aggree with what you say. Btw the short BNC cables I used have exactly 53pF.

With my set up I do not want to make an exact measurement of noise levels and spectrum but just proove that it is perhaps the best way to get rid of the noise entirely. As I already said in an earlier post I'm not interested in qualifying something I do not want.

Once more, it's the amount of reduction that can be clearly seen in the pics that I consider a success. And as the connections are not altered between the Owon in and out of the oil can this reduction must be real.

[rf-loop]

I do not know exactly how you have connected Owon+Tek parallel to signal generator  but I believe this connection add much more than just pure 20pF Tek input capacitance. So, this connection may reduce much more bandwidth than may expect if think only Tek input 20pF. You can try. Connect fast edge instead of this sinewave and yoy can see high reduction in risetime, and this Network may also peaking with several frequencies die to its possible impedance mismatching.

Just for clariy. If connect 50ohm cable to 1MOhm input and look 50ohm cable free end. There is this cable cpacitance and inductance "Network".

Also I note that you use 2mV/dif in Owon. There is 20MHz BW limit. This noise frequency components have lot of over 20MHz frequency components what are now highly attenuaated and peaks are much lower than they really are. 
If you want see around same with Tek oscilloscope and Owon ypu need do impedance matched real splitting Network from Signal gen output parallel to both scopes. Other ways can not see any real data.  But of course also this kind of simply test is useful.

Other thing is very big difference if we think how analog oscilloscope show noise what is "random" with scope eyes. This is real thing what have discussed and studied in many places. Owon itself can show  much more noise than this Tektronix. It can proof if you look some signal sou´rce what produce some known signal added with some random noise peaks.

Fact, I believe, is of course that this noise in this case is coming from Owon.

Btw, for trigger stability there is adjustment in Owon trigger settings.  (HF reject)




---------
About noise displaying with digital scope or analog scope.
(this is partially for these peoples who do not know and randomly read this topic and look these pictures where Tek analog scope do not show nearly any noise (or only small amount) and Owon screen is full of noise.



Try with 5ns width pulses  with 10ms period. What you can see on Tek275 display and what can see on the Owon display?
On the Tek display nearly nothing or nothing and Owon show  just clean full pulse without any problem.  Same for noise, depending noise characters and what signal is trigged.

rf-loop I aggree with what you say. Btw the short BNC cables I used have exactly 53pF.

With my set up I do not want to make an exact measurement of noise levels and spectrum but just proove that it is perhaps the best way to get rid of the noise entirely. As I already said in an earlier post I'm not interested in qualifying something I do not want.

Once more, it's the amount of reduction that can be clearly seen in the pics that I consider a success. And as the connections are not altered between the Owon in and out of the oil can this reduction must be real.

Yes and my opinion was and is


1: Of course "oil can" reduct this noise, there is no any doubt.  Faraday box (well made) have worked in history and it works also today, as long as it is made ok.

Sidenote: Totally different case is example some "RF-shield" boxes installed example on the PCB what really are not faraday boxes due to fact they are in many cases more or less part of circuit (example multi point connection to GND where is RF currents and so, also every single point in GND area is different) And here comes also fact that Owon many SMPS circuits are designed without enough experience and knowledge about these things. This whole GND "Network" is just as "the magpie's nest".  If they design just new and good layout for PCB's they can reduct lot of noise but it need really know what is doing. Of course it also need that every single SMPS circuit out RF is blocked. But also, what they have forget is that SMPS circuit input side is just as important as output side. To common wide area GND from single SMPS circuit internal GND is accepted only and alone DC current flow. It can stay on the PCB and it do not transfer its its internal noise to traawel all around.  If designer do not know how to isolate these just By 2-5 components and right component selections and quite simple rules in PCB layout design it is better to pay final salary and kick out to play Mahjong on the street.




2: Previously and now:  " I believe, this noise in this case is coming from Owon. "

[Sarasir]

Yes, all correct!

This 'GND' network was also the reason to do the cage-of-Faraday-oil-can-test to see if there is a way to leave every thing just as it is (but they also must have achieved defenitely some improvements in their SMPS design as the new SDS9302 is notably less noisy) and just cage it so that nothing or very little of this noise is to be seen anymore from the outside world.

You know it as well when looking at the adapter PCB, all output currents of those little switchers  - or at least significant part of it - are forced to go through the Z-plane.... 

When I had my first SDS working in 'exploded view' with my self build extender and power GND connections done with wires directly to the main PCB there was much lesser noise on the Z-plane and almost no differences anymore (you also mentioned that in one of your past posts as well, that when moving around with the probe tip every two points (GND lead / probe tip) showed different noise level.)

Yesterday I cut a window into my oil can for to see the reduction through the shielding directly on Owons screen and not have to use the Tek 475 anymore to see that.

And the pulse bursts also present in Owons noise 'melange' where back - I think you know what I mean.

So now I know that the display is not only a near field radiator as TomC calls it but also contributes significantly to the noise that's forced out of the BNCs on the other end.

Checked right away what's there of todays technology and found transparant and conducting film from various manufacterors that is also offered for EMI shielding, especially for displays of course (but also used to make solar cells).

Last to say is that I started to cover the window again but with a wire grid but could not finish it yesterday. So today will see further....

[Sarasir]

Hi TomC + rf-loop,

I'm preparing the pics of the oil can test part 2 (without the 475) but not through yet - perhaps later...

In between found somthing that might be interesting to read:

http://www.lowemfoffice.com/low_emf_computer_display.htm

[Sarasir]

Hi TomC and rf-loop,

so here is my first set of pics of the new Owon (Adapter PCB v3.3) without the oil can.
Opposite to my first impression there is still a lot of noise, but it's different from what I remember from my first Owon. The constant noise is much less now but there is noise 'flying by' that I haven't seen before and I think it is generated by the display and/or it's drivers. They changed the display power design slightly. Now there are no wire leads for the backlight anymore. It is supplied now through the ribbon as well. And they omitted one ground post of the adapter PCB close to the ribbon connector. I haven't checked further yet for other changes. The main PCB is the same version but with a National Semiconductor ADC (ADC 08D1500) which is probably the same as the TI one.

It took several hundreds of shots to catch all kinds of noise present bcause some of it is 'flying by' just occasionally - the bursts and the cloud like one. The spikes are more frequent.

I did shots in 2mV and 5mV sensitivity each with the SG505 output on and off.

Next I will post pics taken with the same settings but with the Owon inside the oil can in which I cut a window covered then with a wire grid.

[Sarasir]

...and here the pics of the test with the oil can shield:

Btw, I did all pics at 2mV and 5mV because rf-loop said that the BW limit at 2mV causes lesser noise to be seen. There is a difference as the pics show but far not as much as one would expect. That is I think because of the gain/bandwidth product of the gain stage(s) in the preamp.

The last two pics show the difference with the cores on the power cord removed. That gives an idea of the quality of the line filter in the Owon SMPS....

[Sarasir]

Yeah guys (TomC, hope some time you will say something again but anyways, I will post my results for everybody who's interested in.),

...and here are two pics I did after replacing the original input shield of my second SDS 9302 with the one I have made for my first SDS 9302. The Test set up is as in my post above using the oil can shield (Was very hard work removing the input shield from the old main PCB without killing it entirely).

What you see here is the only noise left. There are no bursts and no cloud like noise clusters anymore, so I post only this two pics with the SG505 output off and on and SDS 9302's input at 2mV.

Thanks God the noise that has disapeared now and of which I thought that it might come from the display is very likely generated by the digital circuitry of the main PCB -  most of it I think from the ADC - and so can be and obviously is shielded off pretty well now.

(For to see my design of a new input shield check my post on 167:
https://www.eevblog.com/forum/testgear/review-of-owon-sds7102/2490/)

Next I will remove all power supply return current from the Z-plane. That means to isolate the main SMPS and the adapter PCB groundwise (the mounting posts!) and then connect them by strong wires to each other and further to the main PCB. Also the upper two posts connecting the main PCB to the Z-plane will be replaced by plasic types. The Z-plane then will become part of an internal cage of Faraday and will be connected to the whole circuit only through the two posts close to the input BNCs.

Will post the results then.

Btw, got a new main PCB for my first SDS9302 from Owon through my dealer for 440 Euros, also equipped with the NS/TI ADC. Perhaps Owon now preferes the ADC08D1500 and stopped using the MXT2815 because of performance/reliability reasons. I do not think that NS/TI's price is lower then the Chinese made one, but maybe its also because of availability reasons.

[TomC]

Hi Sarasir,
The latest noise results look great!

Sorry if you felt ignored but I didn't have much to add to the last two sets of pics, just more noise on one set and less noise on the other. I'm glad you were able to reduce it with better input stage shielding.

It seems you have a clear picture of what you want to do next, so I'll wait for the next installment of results.

[rf-loop]

Hi sarasir.

Nice result after your extremely hard work.

But I will (agen) nopte that you use 2mV/div ehere analog front end is 20MHz limited. Lot of SMPS circuits noise is around 50MHz to 200MHz. For this kind of noise 2mV setting risetime is not enough for show full peak levels) Only where is full BW is front end final stage and ADC input of course.

Perhaps Owon now preferes the ADC08D1500 and stopped using the MXT2815 because of performance/reliability reasons. I do not think that NS/TI's price is lower then the Chinese made one, but maybe its also because of availability reasons.

What is probkem with RuiFeng reliability? Perfomance is open question.. there NS/TI win more or less.
Chines (not specified) Company may destroy what ever IC reliability just simply By using wrong assembly methods and wrong PCB topology design (and wrong circuit design - some times they forget  to read (and underrstand) data sheets and addendum informations) One example is how they forget data sheets recommendations is how they arrange GND's. (example separate AGND and DGND) Some may think that GND is same after 2mm trace. Yes it is for DC and only for DC.


I do not know MXT2815 situation today but some time ago Owon  stop using some RuiFeng ADC versions in some models and surprise it happend after Rigol start massive use of RuiFeng ADC's.  This may explain some things. Who knows...

You have made interenting experimental work.

[Sarasir]

Hi TomC and rf-loop,

Im very glad that there's no bad feelings left. Thank's a lot for understanding. Just want to add that when struggling for solutions and truth mostly there is also friction, inside of oneself and also in between partners and collegues. I consider this even as a natural and also healthy part of the whole process. Bringing in the whole of you including all your knowledge, doubts and emotions is in my view best prerequisite to achieve the max...

And I agree what rf-loop says but also already mentioned that the difference in between 2mV/5mV test results is not as big as one would think (partly gain/bandwith product I guess, but didn't give it hard thinking yet).  As you can see above (oil can test 1+2) all types of noise was present in both 2 and 5mV, with some differencies of course but just little . Also what to consider is that BW Limit @ 20MHz does not mean that the frequency response stops there immediately.

And further thinking about brings up the question at what point is certain kind of the noise - now shielded off by the input shield - introduced into the chain. Before the BF filter or after?

...and to be honest, I was just to lazy - and also very hungry - yesterday to take the SDS out of the oil can again, switch to 5mV and put it back in what also includes everytime removing the cores, the BNC coupling, and the lid and putting all back on... and still wanted to post the result right away.

[Sarasir]

Hi TomC and rf-loop,

as I disasembled my first test approach which was to externalize all switchers (now it looks quite rediculous to me that I thought it might really work) few days ago to get my first SDS 9202 together and working again I started thinking about how to denoise this switchers the best way. You know my first try was adding lots of tantalums and coils what actually spoiled the power up and down rules and also created overshoots and ringing which Im pretty sure now killed the ADC. I read the data sheet from NS/TI and found another restriction (in addition to the over/undervoltage restriction) which is that the supply voltage of the circuit feeding the input must go down first and then ADC's supply. Doing it vice versa also could kill the beast.

And by reading papers about denoising and about ESR of different types of capacitors, I had to realize that tantalums - even chip types - are absolutely out today in this aspect (how easy was it in the old days when just a few tantalum beads did the job....)

I feel ashamed because of my ignorance, and lazyness by using this old components but this little ceramic chips are always a challenge to me to solder especially with my old eyes.
Only on the backlight switcher I had already used a chain of ceramic C chips because there is this restriction of 10uF max at the output.

So, anyways for three days now I've been sorting out my stock of ceramic SMDs - thanks god I have this little Wavetek LRC meter which is a very handy thing - but still it's something else especially if you had put mountains of all kinds of PCBs on the oven last year and heated them up and just drew off all SMDs - boxes full of SMDs I have, I tell you but, thanks God have done some preliminary sorting already in the past.

And immediately I did some preliminary testing:

My simple idea is why not replacing a 10uF ceramic chip - as it is used by Owon as a standart on most switchers in paralell with a smaller 1uF chip and another even smaller one - e.g. with 10 x 1uF chips to push down the ESR? And, as I expected of course it works pretty well so far.

I used the Tek SG503 (up to 260MHz constant amplitude sine wave generator) and alternately the Tek PG506 (50 Mhz suare wave generator) and of course my 475A 250Mhz analog scope.
Used a little piece of tin sheet and BNC cable + 50 Ohm ternination + BNC connector soldered to the sheet on one side and another BNC connector and probe adapter and 10 x probe on the other side to the scope. In between I soldered different chips to the groung plane (tin sheet) and a wire across connecting all together.

The task now is to find out the best chip combination inside of the limit that is given by the original capacitance amount with the maximum attenuation of the HF.

And my question to you is, do you have any experience made in this already?

Remarkably is that I haven't seen this, as it looks to me, simple trick in any paper I have browsed through yet dealing with denoising. Of course Owons approach is to be found - e.g. three different sized chips in paralell - but not more. Maybe it's because of economical reasons but if you look at the low price of this capacitor chips in mass production I can not really see this. 

[TomC]

Hi Sarasir,

Personally I didn't use this approach on my SDS7102, however, many of our members used smaller capacitors to more successfully shunt the higher frequencies. I don't remember seeing more than 3 capacitors in parallel and don't know if using more may significantly improve the situation, maybe your experiment will answer this. At the time, I think the main thrust was to identify places on the circuit where these capacitors may be of benefit as well as existing locations on the board where they could be installed without or with the least amount of modifications to the existing circuit board traces.

[Sarasir]

Hi TomC,

with the SG 503 generator at max output level and frequency (3.5V P-P / 270 MHz displayed by the 475A at the 50 Ohm termination and no connection to the capacitors) and the 475A with 10x probe and at max sensitivity (5mV with 10x probe = 50mV):

Testing the Owon standart 10uF + 1uF + 100nF  (decending chip size 3.2mm/2mm/1.5mm) will give a residual level of 150mV.

Using 10 x 1uF of 2mm size instead the level goes down to 70 mV.

So it is not so much compared to the efford and I think that it is some kind of a zero sequence (Math) which means that it will go towards zero more and more by adding more and more capacitors but never will reach zero actually.

But by adding just a little ferrite bead (two holes) in front and another 100nF/1.5mm to the Owon standart the level went down from 150mv to 15mV.

The level using 10 x 1uF and the ferrite bead went down to 10mV.

Adding another 1uF to the Owon standart + 100nf + bead the level went down to < 10mv, so that's it in a way - so far!.

The guys where right (what you said above)! Just wonder why Owon does not use ferrite beads.

Added later:
It just came to my mind that I tested without DC running through the bead, so with up to 1A it will certainly look different!


The thing with the mechanical arrangement is very critical in deed. The connection must be absolutely straight forward what means the connecting wire must go precisely from one C to the next C and no short cuts at all allowed just all arranged in a straight line. Connecting the out wire just little - e.g. 3mm - before the end of the chain will raise the level right away remarkably - almost double or so.

[TomC]

Hi Sarasir,

Interesting experiment!

As I see it, the interaction of the mechanical arrangement is consistent with the theory behind the use of smaller capacitors for high frequency filtering since longer leads increase the inductance which is precisely what smaller capacitors reduce.

I also think the ferrite bead will still be effective with a DC bias, specially since the two hole configuration you are using seems to be common mode which should result in a null field for the DC bias current with little chance of saturation.

[lemon]

Hi to all.
I read the thread occasionally and I haven't understand which is the area of the scope at the Sarasir mod. Is there any photo?

By the way, Owon upload a new fw with 1st August of date for SDS71021312xxx, SDS71021319xxx, SDS71021327xxx, SDS71021334xxx, SDS71021344xxx, SDS71021352xxx
and
SDS71021407xxx, SDS71021414xxx, SDS71021418xxx series.

Same changing log with previous update (May), but for some reason split these series at the firmware.

[TomC]

Hi lemon,

Nice to hear from you again!

I'll try to summarize what Sarasir has been doing until he has a chance to answer your post himself:

In general there has been a number of experiments trying to come up with new ways of eliminating the ground noise on his SDS9302s. The early experiments involved removing the DC-DC switchers from the Owon boards, adding filtering & shielding, and reconnecting them via long cables. This approach has now been abandoned because of complications caused by the long cables.

During this time he also modified the shield for the scope's input section

https://www.eevblog.com/forum/testgear/review-of-owon-sds7102/2490/)

These mods successfully improved the noise situation and were later added to his second SDS9302.

The next experiment involved the use of a Faraday cage implemented with an oil can to see its effect on the ground noise. Later a window to view the scope's screen was added to the Faraday cage. Both experiments showed noise reduction, I'm not sure if a permanent mod is planned based on these experiments.

The latest experiments involve testing different combinations of small capacitors and ferrite beads for their ability to filter the output of a high frequency signal generator. The intent is to use the best performing combinations as filters for the DC-DC switchers.

[lemon]

Hi lemon,

Nice to hear from you again!

I'll try to summarize what Sarasir has been doing until he has a chance to answer your post himself:

Thanks TomC, it is very kind from you. I appreciate it a much.

In general there has been a number of experiments trying to come up with new ways of eliminating the ground noise on his SDS9302s. The early experiments involved removing the DC-DC switchers from the Owon boards, adding filtering & shielding, and reconnecting them via long cables. This approach has now been abandoned because of complications caused by the long cables.

Something like that doing a Russian member (BAHHAOO or like this) before (do you remember, he has very well improvement). Very effective approach but difficult cause the demand of redesign all the power boards.

During this time he also modified the shield for the scope's input section
https://www.eevblog.com/forum/testgear/review-of-owon-sds7102/2490/)
These mods successfully improved the noise situation and were later added to his second SDS9302.

I saw it and likes me...

The next experiment involved the use of a Faraday cage implemented with an oil can to see its effect on the ground noise. Later a window to view the scope's screen was added to the Faraday cage. Both experiments showed noise reduction, I'm not sure if a permanent mod is planned based on these experiments.

The Faraday cage was for PSU Board, something like I was make before or the Faraday cage was for the probes?

The latest experiments involve testing different combinations of small capacitors and ferrite beads for their ability to filter the output of a high frequency signal generator. The intent is to use the best performing combinations as filters for the DC-DC switchers.

I supposed that this mod is for adapter board, isn't it? I read the last two pages but I can't find for what dc-dc switchers, he was refereed?

[Sarasir]

Hi lemmon,

thanks a lot TomC for summarizing my attemts - great!

another reason for to abandon the removal of all switchers from the scope, cleaning their DC outputs and connecting them back to the scope using long cables was that this experiment showed that there is also a lot of noise generated by the digital circiutry on the Main PCB itself.

And by using long cables with additional filtering the power up and down sequencies where disturbed along with ringing caused by e.g. the ADC switching from low power mode to normal operation (haven't mentioned that yet actually I think). That load change caused spikes on the DC power lines that together with a wrong DC voltages power on/off timing killed the ADC on my first SDS 9302.

Next idea was the Oil Can test as a consequence of pictures and inside views gained through profound and extended discussions with TomC and also rf-loop who threw in his knowledge time and again:

The Faraday cage experiment was to put the whole SDS into an old oil can and connecting the can only to the input BNCs so to catch up all emissions radiated by the SDS and feed them back to the input ground side - in a way nulling them out - thus eliminating their impact on anything that is connected to the input.

That worked to a certain extend quite well but also showed that there must be another leak through which noise is coupled to the input. And that was the poor input shielding.

(I already noted that before and had redesigned it concurrently with the switcher removal - thanks God - but it was on the broken PCB so had to work it over to the new one!!! - read below!)

And, to continue, it's not the original shield actually but it is the way it is connected to the main PCB. By soldering it to the PCB on only three or four points as is was on my 1st and 2nd SDS (upper shield I think three and lower shield four soldering points, on 1st SDS just three points on both I think!) the shields could not work properly for RF only for LF it was perfect.
(Remember just a few millimeters make a big difference in the e.g. 100 to 300 MHz range - not to talk about the area above!).

Everybody who wants to fix this might just solder the existing shields all around seemlessly  to the PCB (watch the white areas where the PCB tracks pass under the shield and do not attemt to solder there!)  and - also very importand - add some circular spring - look at my photos - to connect the BNC connectors directly to the shield - not as originally only to the PCB (that I think is also very important for a maximum input shielding improvment)

The negative aspect of this is that whenever some repair is needed inside the shields access is quite difficult. I would mean to remove the shields and whoever has already done somethig like that knows what I'm talking about. Only this is why I redesigned the shields, build a frame and screwed the original shields to it on top - just to sustain accsess to the inside.

And the filtering test with beads added is for all switchers. Mainly on the adapter PCB but also on PSU PCB and perhaps also the 1.9V switcher on the main PCB (needs checking if really necessary for not to kill another ADC).

All very complex because of it's caotic nature and it takes a lot of thinking to gain a clear view of all interactions and to find the right keys....

[Sarasir]

... and after thinking a while about all and reading again I want to add, yes of course I will try to come up with an over all modification to free the SDSs from noise as much as possible with reasonable efford.

There is one thing left and the oil can test result has lead me to that and that is the role the z-plane acts in this drama.

Thinking after the oil can/Faraday cage test results where present about how to realize an internal cage I focused on the z-plane as it is the biggest metal inside and therefore certainly must be somehow included in that or actually should even be the base of an internal Faraday cage.

And suddenly I became aware of what role this z-plane has been given by this Owon engineers in their naive thinking:

They use it to transport the return currents of all switchers. Wow!!!! I thought, that might be the Gordian Knot we have looked at all the time but could not cut it because we didn't recognize it.

(I' talking only about the time since I've joined here - posts before I read a lot but still not all yet)

In other words, we where speaking of radiation, emission, antennas, noise and spikes but we never focused on what the biggest part of this noise radiating antenna actually is.

It is the z-plane, and it is because the Owon engineers have forced it into that role by making it conducting/transporting all that strong, noisy and chaotic return currents.

The thing MUST radiate like hell - no other choice.

So, the next and hopfully final test is to free this z-plane from all return currents and by this making it a true shield that is only connected to the most sensitive point that we want to be clean, the input BNCs. And I'm looking forward to the result with great tension, because I have a certain feeling that this might be the hidden key we where searching for such a long time....

[lemon]

Sarasir, thanks for the clarification and explaination for what you have done until now!

Now, I have clearly all of them in my mind...

The Z-plate and how this related with the noise, had a lot of conversation before (older messages).
rf-loop many times has referred about this.

I don't think so that the Owon uses the z-plate for better returned currents of switchers.
I think that in their mind, was the better shielding.

Your thought to remove all the returned currents from z-plane, will be the better solution for the "noise-free" of this scope.
I had made a lot of modification on adapter board (like rf-loop did) and some more to the backlight circuit.
When I attached the new psu & adapter board that Owon warranty cover to me, I have stopped any modification to this board.
With the new noise level I am been satisfied, but I can to experiment with the olders boards...

[Sarasir]

Hi lemon,

The Z-plate and how this related with the noise, had a lot of conversation before (older messages).
rf-loop many times has referred about this.

I don't think so that the Owon uses the z-plate for better returned currents of switchers.
I think that in their mind, was the better shielding.

yes, I know, rf-loop said as well that there is differential noise everywhere on the z-plane.

But nobody so far has seen it as a result of return current RF voltage drops yet (to my knowledge), rather as coupled  through the electrical and magnetical fields to the z-plane from adapter PCB. That in my opinion is a very important difference.

If the return currents go through the Z-plane there will be RF voltage drops due to inductivities of the z-plane resulting in different noise potentials all around the z-plane especially in the area around the big aperture where main PCB and adapter PCB are connected. That is almost like if you want to use it as antenna of a transmitter. I can not understand how they could think of the z-plane as a shield when the same time it functions almost as a dipole antenna or - I'm not too familiar with antenna technology - or as the feeding element of a multi-element-antenna.

Replacing z-plane as a conductor by wires should reduce radiation tremendously as the area that is radiating is reduced significantly. And if that is not enough these wires could also be shielded.

In former discussions with TomC you can see how we arrived at the picture, seeing the whole problem as a RF-frequencies-radiation-balancing-thing - just if you didn't get that yet entirely.

I'm working on my first SDS to get it running again - got a new main PCB - and will do the modification then on it. Will take a while...

[OldTimer]

Hi!

I am considering purchasing this scope and have read all of the available manuals etc. but I had a couple of questions that were not answered by reading the manual.

1) Does this scope have a waveform averaging function?  If so, what is the range of samples that it can take?
2) For the FFT function, what is the range of points that can be selected?
3) Can this scope perform an FFT on a portion of the displayed waveform?

Thanks for all of your help!

OldTimer