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

[TomC]

Hi Sarasir,
Very interesting changes. Thanks for the pictures!

On the SDS7102 the noise is much more bothersome when using the X10 probe setting with the long ground lead at the 50mV/div setting. Those are the test conditions for the <30mVpp noise figure I quoted before. At 20mV/div, which is the lowest possible setting in this case, the noise is less because the bandwidth is limited to 20MHz. Also, if the X1 probe setting is used the noise is much less because the probe then limits the bandwidth to about 6MHz.

I noticed from your pictures that your unit uses the previous version of the PSU, T115-N Rev3. This is the PSU that was also originally installed in my unit. While I was trying to modify it I made a complete schematic of this particular version, I attached a copy in case this may be useful to you. My final modifications consisted of adding a belly strap to TR1 and replacing L2 with a hand wound toroid. I'm still using this PSU in my unit, the noise level with these mods and a ferrite on the ribbon cable are as good or better than what I get with the new style PSU.

I also attached images of the new style PSU. On the printed circuit side you can see a number of SMD capacitors that were added by Owon to reduce noise. They also isolated the Power Switch's casing from GND-C to prevent it from radiating noise. In addition, they changed the connection of D8's heat sink just as you did on your unit. So it is no longer connected to the anodes of D8 and instead is connected to GND-C. Other than these changes, the schematic I made is the same for this newer PSU version.

Again, I would like to see more of what you are doing, and let me know if there is anything I can do to help!

[Sarasir]

Hi TomC,

thanks God finaly sombody to talk to who knows what I'm speaking about.
And thanks for the Scheme/pics. I will check them later 'cause just got up. Wow!
Of course I will post everything I did with all notes/schemes/photos. Just all scribbled on loose paper and needs to be checked again before posting. And these was the first pics I made. You know when I'm working I do not spent so much time on documentation but I know I schould - but everything's here...

Until later - now I'm on skedule....

[Hydrawerk]

For such an expensive 1500 Dollars scope I would expect a better internal construction. 
http://www.testequipmentdepot.com/owon/sds-series/sds9302.htm

[Sarasir]

yes of course I see what you mean Hydrawerk. It's quite some money and as I already said I couldn't believe it.  And if you look at the necessary mods that I believe will do, it's just cents: Some tin boxes, inductors with tantalums or equivalent and an extra little PCB for all the switchers and in turn the interface PCB then can be made smaller. On the Chinese market almost nothing I guess. And therefore they are ruining their name and reputation. And everything else is nicely done. The guys of OWON should study their own culture 'cause there is a chinese saying: Missing it by a little can lead you very far astray.

But you know, I'm learning new things and there is also a lot of fun and if we made it we feel like kings. What else can you ask for in life? Money? And I tell you I made quite some money in my life with electronics through what I've learned from all my crazy projects, really.... no school makes you learn better and you will never forget it anymore.

TomC,
have everything together but will post it tomorrow. I'm tired now and it's late.
For now I will attach two pdfs that helped me a lot to develope understanding of what goes on in these OWONS. Didn't  torture myself with the formulas just the text works out everything clearly...

[Hydrawerk]

This SDS 9302 is for some specific needs when you need 300MHz bandwidth and nothing else, but for a hobbyist 100MHz might be enough.

[TomC]

Hi Sarasir,
Those are two very interesting papers!

I also have an extensive collection of PDFs that deal with SMPS design, noise reduction, etc. So I will add these new papers to my collection, thank you!

There is also a paper on the theory of operation of the T115-N Rev 3 PSU that I started writing before trying to modify my PSU for lower noise, see attachment. I didn't finish it so it is not complete but it may be helpful if you haven't figured out yet what some of the components are supposed to do. Note that the information in red were notes to myself for things to check and try out, later I found out that most of this didn't require any action, but I never updated the document. Also I never updated the information on the snubbers, although later on I found out that Owon used R2CD snubbers that rely on the operation of slow recovery diodes like the 1N4007. Anyway, hope you can get something out of it.

By the way, I agree with you that trying to improve these scopes is a great learning experience. In my case, it made me investigate deeply into the operation of SMPS, which I only understood superficially before. I also think that a 300MHz scope, even just for hobby use, can provide you with useful details that otherwise would be invisible. And that alone justifies the cost to me if you can afford it. I would have gotten a 9302 also, unfortunately, justifying the cost to my wife wasn't quite that easy.

[rman_nl]

As Mark Model video  shows (23:30-24:10)we can see the output (using the vga ) of scope on a desktop screen in a Window (or Ubuntu?) environment. What hardware and/or software do I need?
I'm using now Ubuntu 14.04/Windows 7 with a hdmi cable connection to my 24 inch screen.

 

[rf-loop]

As Mark Model video  shows (23:30-24:10)we can see the output (using the vga ) of scope on a desktop screen in a Window (or Ubuntu?) environment. What hardware and/or software do I need?
I'm using now Ubuntu 14.04/Windows 7 with a hdmi cable connection to my 24 inch screen.

 

There is direct VGA output on the SDS7102V.  (native resolution 800x600 and pixel to pixel realtime copy from Owon own TFT display))
Direct VGA cable to monitor, (or VGA to what ever adapter) thats it and also it is only way to get realtime parallel display.

[Sarasir]

Hi TomC,

I just read through your Theory Of Operation paper and schematic for the OWON SMPS and was quite impressed. Increased my understanding of whats going on inside these things significantly.

And, by the way, found the data sheet for the MOSFET and will attach it.

First about the residual noise and your noise figures.

In general I have to say that when testing/calibrating scopes you never use probes. The scope must perform the same whatever you connect to the input. So, stating any results by feeding the scope via a probe puts two things in one pot and nobody can tell which component is responsiuble for what and how much.
The only step in a calibration procedure of a scope where you need something like a probe is the input capacitane adjustment and this - at least in Tek labs - is done with a so called NORMALIZER a simulated 'perfect' probe with a known and calibrated capacitance matching the specified input capacitanc of the scope. And you adjust the input capacitance of the scope then for a flat square wave like you would do later with a probe that in turn will have to be adjusted to match the scope's input capacitance.


As a reply to you noise figures I will describe the noise problem of the OWON now as I experienced it when testing the machine and try to interpret it a bit of course according to my current knowledge (I might be wrong because of having overseen something - but don't think so):

Test setup example: Connect the output of whatever device to be maintenanced or tested to CH I input and a probe to CH II input.
Then shorten the tip of the probe with the groundstrap of the probe. How much noise do you see on screen provided the shorted tip end of the probe is as far away from the scope as possible?
Answer: nothing or almost nothing (no noise). Then take the ground strap off from the tip and touch the scopes input connector which is supposed to be solid ground. And what do you see on screen? Answer: Lots of spikes!!!
And now start tipping around inside of the device to be maintenanced/tested.
What will you see?
I can not tell at what level or what phase or spectrum but it will be noise everywhere except if you are lucky and find a point of Zero for all frequencies (not very likely).
I also depends on whether and how the device is grounded by itself, through the mains cord or whatever.

Conclusion is, that the OWON pollutes every conducting thing you connect to it's ground as good as it can. Using long wires and in addition grounding the thing to your central heating radiator the OWON has a very hard job. Is the thing a floating battery operated device and the connection to the OWON very short and done with HF wire it will pollute the thing entirely but you will not be able the see it on screen because the tip of your probe might be in perfect phase with the thing connected to OWON's ground. But after connecting something else with big mass or even grounded to the thing your measurements will show different again.

Its not the hot side only of the input that is pulluted! It's the COLD SIDE AS WELL.  And that means actually you can not predict the displayed noise at all.

So I have to say that any noise figure you give can only be true for the individual setup you had.


To the attempts of OWON reducing the noise by circuit mods I can only say that the result can be only marginal. They should know better. Of course as you pointed out in your paper 'slowing down' the slopes would help but  there are hard limits as you wrote (Power consumption / component sizes).

How much reduction did you get with the belly strap and hand wound toroid? and what material did you use, chopper foil?

And another thing I would like to ask you: please can you check what is written on the backlight switcher controller and the screen supply switcher controller. I can not fine the sheets in the net and the screen supply is a bit complex as it prvides the power up sequence for the various display voltages. Found the spec sheet of the display with power up sequence detailed but no app circuit given. I will retrace it but you know missing a little and killing the diplay wouldn't be so great. Datasheets are always a good backup also if some resistor breaks when unsoldering or jumps away.

And my postings - all a bit rough and preliminary. The final design of the smal converters I will do after I have everything together.

1. 3.3V buck converter: In the box is what was on the I/F PCB just added Inductors + tantalums as filter

2. 5.6V buck converter: removed all components and replaced it with a linear 7805 + diode for 0.6V increase. maybe will keep it but will decrease battery time should I ever need it. To be decided later.

3. 1.9V Buck converter: to be found on the main PCB. removed all parts using hot air but replaced the controller then with the one that did the 5.6V before. The reason was that the original was supplied by the +5,6V converter and by taking it off the main PCB could feed it directly with the 8.4V also taking away current from the 5.6 now linear regulator.
And for the former 1.9V controller 8.4V would be above or just  max specs. As the former 5.6v controller has an enable input I could use this to keep up with the correct power up sequence and also power on/off dependencies. (1.9V only present if 5.6V present) It works fine.

So that's it for now. have removed the backlight switcher but not boxed yet. And will continue working on the display switcher.

Next photos will be the input shielding I did and minor mechanical mods for to make a bit more space between button PCB and main PCB (.6mm) but have to dissasemble the thing again before.

Keeping up the good sprit,
Sarasir

[Sarasir]

....and here the MOSFET data sheet (Change the extension to rar and unpack it)....

[Carrington]

@ Sarasir:

Below the shielding metal sheets of the input stage there is a 5V LDO, connected to ~5.54V (i.e Vdo <0.54). And if Owon did not put any filters (F.Bead) in the input of the analog power for the DAC, less to the input of the analog power for the input stage. This LDO is able to reject only some frequencies <1MHz. Similar for the 7905.

Good luck!

[Carrington]

2. 5.6V buck converter: removed all components and replaced it with a linear 7805 + diode for 0.6V increase. maybe will keep it but will decrease battery time should I ever need it. To be decided later.

Humm... Many amps for an 7805, at least 2 Amp.

3. 1.9V Buck converter: to be found on the main PCB. removed all parts using hot air but replaced the controller then with the one that did the 5.6V before. The reason was that the original was supplied by the +5,6V converter and by taking it off the main PCB could feed it directly with the 8.4V also taking away current from the 5.6 now linear regulator.
And for the former 1.9V controller 8.4V would be above or just  max specs. As the former 5.6v controller has an enable input I could use this to keep up with the correct power up sequence and also power on/off dependencies. (1.9V only present if 5.6V present) It works fine.

Edit:
Ok, now I understand, you got rid of a few amp.

[Sarasir]

Hi Carrington,
thanks a lot! I recognized this 5V regulator when redesigning the input shielding but did not think about it any further as I thought its linear and therefore should give no no problem.


TomC,

here I have attached the remainding pics.

[Carrington]

@ Sarasir:
Could you replace all this with batteries or a linear PSU? Just to see the noise floor.




I suppose that you are using ferrites on all the cables in and out the metal caps.
All these cables can be a big problem.

[Sarasir]

first time when I used the 7805 was before I had removed the +1.9 switcher from the main PCB that hangs on the + 5.6V and I tell you it worked without the 7805 getting real hot (I placed it on the main chassis where normally the battery compartment is with a mica and screwed a flat spring to one of the threads to hold it down.

After I had removed the +1.9 switcher and connected it directly to the 8.4 volts the +5.6 line drew about 1Amp and the 7805C - version can do I think 1.5A. Before it was - not sure now, was some time ago - less than 1.5Amps. so maybe they have changed certain ICs to versions with lower consumption in my maschine.

But, as I wrote thats only preliminary. If finally I should use a linear Reg at all I will of course use a LowDrop one with higher amp rating.

[Sarasir]

...of course you could, but the efford...

No, my concept to get a definite result is to take all switchers out of the machine and put them in closed tin boxes with enough filtering so that they are just clean. Connect them with long cables to the scope and use enough capacitors on the scope side to eliminate the wires influences (HF wise).

Then to ckeck if there is any noise from the digital circuitry (core ++) going into the analog side, and if then it's getting very comlicated as one has to thinker around on the main PCB digitizer circuit and control and therefor have to retrace the circuit first as a start for to know whats going on at all.

I already recognized that they provided no split ground for digital and analog and I think you can guess what that means....

[Carrington]

LDO: https://www.eevblog.com/forum/testgear/review-of-owon-sds7102/msg312500/#msg312500

And don't use the internal LCD, is another source of noise, including the connector that joins main the boar with  the DC/DC board.
Use only VGA output.

[Carrington]

I already recognized that they provided no split ground for digital and analog and I think you can guess what that means....

This, is not always necessary/mandatory. A "no split ground" may be even better.

Trust me, this thing is a nightmare.

[Sarasir]

nothing against what you say - it depends on the way the ground distribution is done.

And of course I noticed already the noise radiated from the display but there is no way to say anything defenitely about that as long as these many connecting lines between Main PCB and display all running along on the I/F PCB in close vacinity under 4 of these little switchers. This I/F PCB acts as a perfect coupling device to make sure that all spikes are properly distributed over the entire machine.
Another thing that must have been designed/decided by a dev engineer who rather should be kept away in an assylum for mental cases.

You see the display itself works at a relatively low voltage level signalwise, haven't checked that yet - perhaps around 3V whereas the switchers work from 8.4V and some amps are flowing.

So what would you guess makes more noise, a bunch of low current / low voltage signals or high current square waves at up to 8.4V.

The other thing is that the ground side of the switchers on the I/F PCB are all connected with several screws the the chassis and, very important BEFORE THEY ARE PROPERLY SHIELDED AND FILTERED. If you want to do that right simply remove shield and filter them first and then you connect them to that point in the whole circuitry where most of the amps are needed and nowhere else. This way you create a predictable environment. And the signal display lines running along the I/F PCB after all switchers have been removed can be even wrapped around with some thin metal foil should they also radiate significant noise. And should the radiation going out from the front of the dislay be too much for you you get yourelf one of this conducting but transparent microfabrics that Tek offered as an option for all of its scopes for environments that are highly sensitive to noise radiation. You know, a normal CRT radiates as well out of the screen a little...

I think we can handle all of this. And after we have designed a small pcb for all the little switcher beasts and put them inside of a small tin box that will be screwed through isolated posts above the I/F PCB with no noise going outside of this box and the outputs connected with if you want shielded cables to where the voltages should really go I'm sure we have that thing cured.

And, I tell you if this spreads around the price for this machines will fall and whoever feels like may offer a kit with all inside what we did (will not be me for sure will be off to new shores then) to convert this bastard to a decent device.

Just search ebay for this owons as I did recently. Countless offers came up. Piles of this machines must be out waiting for uninformed buyers. Do you know what I means....?

[TomC]

Hi Sarasir,

I'm amazed at the number of mods you have made, have you devised a way of quantifying the improvement after each mod? Personally, I'm always curious about the effort/improvement ratio. Not that it matters much if the goal is near perfection, and it seems to me that you are well on your way toward achieving that!

As far as the noise tests, I'll try to explain what we have been doing so far and why. It all starts with the symptoms of what we've been calling GND noise in the SDS7102s. On scopes that are not otherwise defective, it manifests itself when the probe, set to X10, is used in combination with the 6" ground lead, or strap as you call it. On these scope the base line noise is negligible, about 1mV, and signals, including low level ones, are virtually noise free when the spring ground clip is used instead of the 6" ground lead. However, when the 6" ground lead is used there is HF noise with spikes riding on the signal, this noise persists even if both the probe tip and the 6" ground lead are connected to the scope's ground lug. The amount of noise varied quite a bit from unit to unit, in my case, it was in the range of 100-150mVpp, with the highest spectrum amplitude at around 100MHz.

Similar symptoms have been well documented when signals on a DUT are polluted with common mode noise. Scope manufacturers warn in their literature about this effect and often provide probe accessories to allow shorter ground connections, for example, the spring ground clip that comes with the Owon probes. Common mode noise is normally invisible to the scope, which is designed to view differential mode signals. However, because the 6" ground lead causes an impedance imbalance due to its inductance, the common mode noise is converted to differential mode noise, which is then naturally displayed by the scope.

The disturbing fact in this case, is that the common mode noise originates within the Owon, and even when viewing signals from a clean DUT, the user is deprived from the convenience of using the 6" ground lead. Member rf-loop, which is also an Owon dealer, was the first one to identify the SMPS and peripheral DC/DC converters as the culprits. To help members determine if their scopes had this problem and to what degree, a methodology accessible to most users was devised to check/test the ground noise. This basically involved using the scope probe set to X10, and connecting both the tip and the 6" ground lead to the scope's ground lug. The scope was then set to 50mV/div, 100us/div, maximum sampling rate, and several other details, before taking a reading. See the following post for all the details:

https://www.eevblog.com/forum/testgear/review-of-owon-sds7102/msg255469/#msg255469

Several approaches toward minimizing this problem emerged from a number of our members with varying degrees of success. My approach was to try and mitigate the noise at the source by altering the circuit so it wouldn't cause so much noise. For my purpose, I was satisfied with the results. But many other members obtained similar results using other approaches, and Owon obtained similar results with the new PSU and a mod to the adapter board. None of these provided a perfect solution, but the noise was reduced to the point that it was possible to view low level signals while using the 6" ground lead.

I attached a few PDFs that were particularly helpful to me. If you are interested, next I can post images of some of the test results, like before and after noise levels.

[Sarasir]

Hi Tomc,

I' sorry but I think that you want to qualify a 'deplorable state of affairs' (got it from the Dictionary) and what the heck should that help as we already know what's it's cause.

It's like counting the red spots after somebody got the measels instead of giving the right medicine.

You qualify and count red spots if you want to optimize a thing not when you have a basic, well known and documented problem.

At 100MHz a short piece of wire acts already like a coil. So what does it help using probes to qualify the prob here. It might help to find out that there is a prob but thats it.

What you explaind is absolutely right - common mode / differential mode - but is again just a refined description of the problem.

You know a machine radiating HF is like a transmitter. you have antennas, traps, ample return pathes, stray capacitances/inductivities and whatever. It will act differently if you move it from a desk made out of dry plastic to a desk made out of wood that stood before outside in the rain.

Radiation return path takes place through the mains cord but also through the electrical/magnetical fields. All of this has absolutely chaotic nature here!

If you do not want a machine to act as a transmitter kill or shield the source of the signal that makes this machine act like a transmitter not qualify how good or bad it acts as a transmitter.

Concentrate on what has to be done to make the thing work fine.

[Sarasir]

Hi TomC,

to reply to you first question:

I already wrote that in one of my first posts I think, after each step I made the noise got down. I did not take any readings because as long as there is reduction after every step made I know I'm on the right way, and the noise was so high and unpredictable that, as I already pointed out I didn't feel like wasting my time with qualifying something I do not want (I love to qualify good behaviour though).

But to make myself clear for not to be missunderstood, of course you need to qualify problems if you want to claim refunding or compensation for damage, malfunction or bad workmenship towards the seller/manufacturer (I'm not so good in leagal english but think you know what I mean).

One more thought came to my mind and that is that I have the feeling that you wanted to find THE ULTIMATE SHORTCUT through witch simply by replacing ar adding a few cheap components the problem will be solved.

And I tell you and I'm pretty sure about that, it doesn't exist.... here we have to go the hard way!

Please forgive me...

P.S.: But you already said that you think I'm on the right way...

[Sarasir]

....and please can everybody check the I/F or Adapter PCB for the marking on the controllers (ICs) of the backlight and display supply switchers. I would need the datasheets and already searched the net but there are many smaller companies who do chips like that. So far I have not found any who matches somehow partly the code/markings on mine and I hope that  OWONS out of other production time have either chips with different markings or equivalent chips from other companies or non coded markings or whatever might help to identify them.

The backlight controller is quite simple but has an control input that is connected to the main PCB. I guess it is for the intensity control. Without that it could be replaced easily. Or one would have to check the format of the control signal and then look for a replacement.

And the display circuit is the most complex circuit of these switchers as I already mentioned.

The location of these ICs is upper right where the two wires (black and red) to the display are connected and bottom right where the ribbon to the display is connected.

The marks on the chips in my machine are:

upper one: OBT   24J     PDVH
lower one: 1780     TB51

So please guys collect information if you can spent some time supporting my approach.

[BBAAHHOO]

Isolation of power sources have already done ... was the noise from the LCD screen ...

https://www.eevblog.com/forum/testgear/review-of-owon-sds7102/1455/

[BBAAHHOO]

https://www.eevblog.com/forum/testgear/review-of-owon-sds7102/1500/