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

[lemon]

I quote some older post about how was the noise related with the battery option or not.
There was differente results between me and rf-loop.
Now after my mods there is no any significant difference between battery or ac plug.

Unfortunately, mine situation is different from what we are expected!
With well charged battery from the worst to best the sequence is: battery (?400mV), battery+ac main(?300mV), ac main (?150mV) - Cheked by Owon Probe after Self Cal procedure and Probe adjustment.

Yes, this is unexpected result.
I have seen before:
(a) battery+ac  most bad, 
(b) ac without installed battery (battery out) less than  a 
(c) with battery alone without ac connected,  less than  b

It need some more investigation. I have some small suspects but too early to say.

Thanks Flash2light for this sharp photo.

TomC, you have a lot of improvement.
This was you have done after the psu changed and adapter mode, 68mV peak to peak and now you are 36mV about:


But I have some of questions-observations
1) you have different time set, the first was 1ms and now is 500us, normally there is no significant difference but if there is a periodic noise, the time makes the difference.
2) The back cover was closed at the two measurement of noise or they was different? Perhaps, this changed the noise.
3) some times the residues decrease the electricity of components. Personally, I have see some difference to noise after a good bath and very well drying with isopropylic alchool.
4) what you meaning with "Base line noise", from what I see there is no any BW limited or something else.

But please to all member, let us make measurements with same conditions like a protocol.
I copy the directions of rf-loop for comparisons reasons:

Procedure for measurent of GND Noise
----------------------------------------------------
Make a Calibration procedure as manual refer (there is no nessesary if not the temperature variables ±5°C from the initial calibration)
Compensate the probe as manual described.
Change switch probe to 10x and set the same to the Probe Owon Menu (10X).
Connect together the probe tip and the probe aligator ground  to Probe Comp GND.
Fix the cable of probe like the attachment photo.
Turn CH2 off.
CH1 to DC, 50mV/Div.
Push Trigger 50%
Set the Acquire Length to 1M and the Acqu. Mode to Peak
Set the Horizontal speed to 100us
Set the measurement to Voltage peak to peak (Vp to Owon) and write the value.

[TomC]

But I have some of questions-observations
1) you have different time set, the first was 1ms and now is 500us, normally there is no significant difference but if there is a periodic noise, the time makes the difference.
2) The back cover was closed at the two measurement of noise or they was different? Perhaps, this changed the noise.
3) some times the residues decrease the electricity of components. Personally, I have see some difference to noise after a good bath and very well drying with isopropylic alchool.
4) what you meaning with "Base line noise", from what I see there is no any BW limited or something else.

But please to all member, let us make measurements with same conditions like a protocol.
I copy the directions of rf-loop for comparisons reasons:

Procedure for measurent of GND Noise
----------------------------------------------------
Make a Calibration procedure as manual refer (there is no nessesary if not the temperature variables ±5°C from the initial calibration)
Compensate the probe as manual described.
Change switch probe to 10x and set the same to the Probe Owon Menu (10X).
Connect together the probe tip and the probe aligator ground  to Probe Comp GND.
Fix the cable of probe like the attachment photo.
Turn CH2 off.
CH1 to DC, 50mV/Div.
Push Trigger 50%
Set the Acquire Length to 1M and the Acqu. Mode to Peak
Set the Horizontal speed to 100us
Set the measurement to Voltage peak to peak (Vp to Owon) and write the value.

The measurements were made with the scope completely back together (back cover closed etc.)

The baseline attachment is the noise when nothing is attached to the BNCs (no probes). Sometimes called baseline because it's supposedly what you start with, independent of ground noise for example.

The time base 500micro seconds or 1ms don't make much difference, but the trigger does. What I found is that with this low noise level, auto trigger with CH1 source is sporadic while it tries to adjust to noise spikes etc. You end up capturing something that is not the average. I suggest using AC line source for a more stable trigger.

Also, when you capture in peak detect mode, the capture length is always 10M. However if you do captures on other modes, you should set to 10M manually to make sure you capture all the noise. Try it and you'll see it makes a difference.

I prefer Average mode to get a more realistic picture of how the noise is going to affect your use of the scope when viewing real life waveforms. I use average 16 so it doesn't take too long to stabilize. Peak detect mode gives a snapshot of what's happening in the instant you do the capture, as far as the highest noise levels are, but these transients are not a good prediction of performance when you are using the scope normally. So I would like to see more people posting their average noise, if possible.

Got to go, it's lawn mowing day!

Edit: The capture length is not 10M, however, if you forget to set the length to 10M, the noise reading will be almost if not equally as accurate as if you had. The reason is that on peak-detect mode the scope first selects and keeps the lowest and highest samples, then throws away other excess samples so that the image fits in the memory length that has been selected.

[rf-loop]

Also, when you capture in peak detect mode, the capture length is always 10M.

Really?
Afaik, not.


Peak mode do not have so much advantage if scope use one channel 1GSa/s or two channel 500MSa/s.

As we know if one channel in use ADC run allways  1x 1GSa/s mode and if 2 channel in use 2x500MSa/s mode. This is independent of time base setting of course. If example selected one channel and 1k memory and 500us/div there read 100kSa/s. But system run 1GSa/s.  But it collect to 1k sampling buffer only every 10000th of sample and do not care about these other samples at all. Simplified: In peak mode it is not blind when it drops out these 9999 samples and keep one.. it find highest value and lowest value there.. (becouse still 100kSa/s there is now sequentially ...highest, lowest, highest... after every 10us.)

I do not know how it exactly then build displayed image etc details so there may be some tiny differencies if use 1GSa normal mode or slower but peak mode. In practice not much.

Set 1k memory, 500us/div, peak and display mode dots.
capture just something noisy.
Stop scope. Zoom out ti 1ms/div you see there is 10div
Zoom in to 10us/div. Count sample points... there is 1000

Same but acquire mode normal.
Count sample points, there is 1000.



(nosignal5mVpeak1k.bin.txt) remove  .txt
There is 1000 sample points.

[lemon]

...

The baseline attachment is the noise when nothing is attached to the BNCs (no probes). Sometimes called baseline because it's supposedly what you start with, independent of ground noise for example.

...

Got to go, it's lawn mowing day!

Thanks for explaination TomC, but seems like crazy to me a 10mV base line, normaly should to was to 1mV.
Look, at the last capture that rf-loop upload at the 5mV. The base line noise is 1mV (full BW like yours).

[TomC]

Also, when you capture in peak detect mode, the capture length is always 10M.

Really?
Afaik, not.


Peak mode do not have so much advantage if scope use one channel 1GSa/s or two channel 500MSa/s.

As we know if one channel in use ADC run allways  1x 1GSa/s mode and if 2 channel in use 2x500MSa/s mode. This is independent of time base setting of course. If example selected one channel and 1k memory and 500us/div there read 100kSa/s. But system run 1GSa/s.  But it collect to 1k sampling buffer only every 10000th of sample and do not care about these other samples at all. Simplified: In peak mode it is not blind when it drops out these 9999 samples and keep one.. it find highest value and lowest value there.. (becouse still 100kSa/s there is now sequentially ...highest, lowest, highest... after every 10us.)

I do not know how it exactly then build displayed image etc details so there may be some tiny differencies if use 1GSa normal mode or slower but peak mode. In practice not much.

Set 1k memory, 500us/div, peak and display mode dots.
capture just something noisy.
Stop scope. Zoom out ti 1ms/div you see there is 10div
Zoom in to 10us/div. Count sample points... there is 1000

Same but acquire mode normal.
Count sample points, there is 1000.



(nosignal5mVpeak1k.bin.txt) remove  .txt
There is 1000 sample points.

You are right rf-loop. What I meant to say is that on peak detect mode you capture just the highs and lows of the signal and ignore everything else, so the noise reading is pretty much the same even if you forget to set the length to 10M. I was in a hurry and tried to shorten what I was writing and it came out misleading. Thanks for pointing it out

[TomC]

...

The baseline attachment is the noise when nothing is attached to the BNCs (no probes). Sometimes called baseline because it's supposedly what you start with, independent of ground noise for example.

...

Got to go, it's lawn mowing day!

Thanks for explaination TomC, but seems like crazy to me a 10mV base line, normaly should to was to 1mV.
Look, at the last capture that rf-loop upload at the 5mV. The base line noise is 1mV (full BW like yours).

Lemon, it is 1mV, but the probe setting is set to x10, I didn't bother to change it, so the scope multiplies by 10 whatever it thinks is coming from the BNC.

I got done mowing about an hour ago and had to take a rest

[TomC]

Here is a little more information on peak detect mode. May help clarify the last few posts on that subject.
The entire article in PDF is also attached.

[TomC]

After a lot of tinkering, prodding Owon to come up with solutions, and some major modifications, the GND noise level of several of our scopes is down to under 50mVpp. It also seems that members that bought their scopes recently are getting decent GND noise levels right out of the box.

For me, the GND noise level was never a deal breaker, but rather like a nagging little gnat that you can't swat no matter how hard you try. It can't hurt you, but you'll go to considerable lengths to get rid of it.

Now that the GND noise level of my scope is in the mid thirties of mVpp, I wanted to see how this improvement would potentially affect the way I use the scope. One of the main reasons for wanting a low GND noise level is to make it possible to view and trigger on low level signals while still enjoying the convenience of using the long ground lead.

The attached captures show the lowest level signals that allow my scope to properly trigger and consistently display the correct trigger frequency. It is possible to obtain a stable trigger at lower amplitudes, but the trigger frequency displayed is not correct. The threshold where this transition occurs in my scope is around 36-38mVpp.

I obtained images with both the long ground lead and the ground clip. The signal generator was set to the lowest level that allowed the correct trigger frequency to be displayed while using the long ground lead. Then the ground clip was used to capture the same signal without altering the signal generator setting. At low frequencies the two images are almost identical. At higher frequencies the effect of the long ground lead inductance starts to become apparent. At 30MHz the amplitude of the two images is several millivolts apart. The higher amplitude obtained with the long ground lead. Perhaps is not such a good idea to use the long ground lead much beyond this frequency if accurate readings are needed.

[lemon]

TomC had you measure before changes what was the threshold for right triggering?

I think that these measures (noise + triggering) there is no way to obtain with the old psu board?

I hadn't test before to my scope the level for right triggering but with very low signals there is no way to examined because noise.
There is some documentation here (it demands login to see the captures) about before and after mods. With noisy gnd and long ground clip 50mVpp square signals there is no way to examined, after mods 1MHz/2mVrms sinus signal is almost perfect with a long gnd clip. But there is no reference about triggering.

It is interesting if other members have some tests before and after and specialy triggering operation.

[AndrejaKo]

Hi there guys!

I decided to do some experimenting with ground noise problems, since it seems that I have a particularly "interesting" unit. I have a "thick" part of noise which is around 300 mV peak to peak and some huge spikes which are between 2 V and 3 V (even worse with 150 MHz probes I'm normally using). Their period is around 80 ms.  I have an older through-hole version of the power supply marked PCB-T115-J rev4 and 3.0 adapter. From time to time, there will be a huge single peak of 5 V to 6 V as well. One interesting thing is that this PSU is using the "Z-plate" as heatsink for D8.

To get to the source of the noise I connected one probe to the ground pin on the probe compensation connector and moved the other probe around the power supply. From what I can see, most of the peaks seem to be synchronized to the waveform on the coil marked L2.  Unfortunately, my power supply does not have coil L3 (which from what I can see if used as a filter) and instead seems to be filtered only with an electrolytic capacitor. I placed two 100 nF 1206 SMD capacitors and two 47 nF 0805 SMD capacitors over the electrolytic. In addition to that, near the power connector there were unpopulated places for ceramic through-hole capacitor and for one electrolytic capacitor on the -7.6 V rail. I placed one 470 µF capacitor and one 100 nF disk ceramic capacitor at the power connector. In addition to that, I placed same set of 4 SMD capacitors in parallel to the electrolytic.

Next, I added two 100 nF 1206 SMD capacitors in parallel with all electrolytic capacitors on the secondary side of the power supply. I also replaced the 1N4007 diodes mentioned few pages back with UF4007, but in my case that made no measurable improvement. My guess is that the reason for that is the large amount of noise that is cancelling all the (minor) improvements I made.

As you can probably guess, I'm poking in the dark here trying to come up with something that would help. My next plan is to get some 1 µF 1206 and 0805 SMD capacitors and place them in parallel with the ones I already have on the -7.6 V rail. 

First attached image is how everything looked like with only ferrite on the power cable going from the PSU to the adapter board.  Next two images are made after adding the capacitors on the -7.6 V rail and show relationship between the noise peaks and the waveform on L2.  Final image is what I get when I use same settings as on the first image. Unfortunately, there seems to be no improvement (in fact I could say it's a bit worse now).

So does anyone have any smart ideas what to try now? I'm starting to run out of ideas and am considering undoing the changes I made since I don't actually have any problems running the scope for my needs. It's just that knowing that the ground noise is there is bugging me.

[lemon]

AndrejaKo you have the most  psu board that I have seen!

Can you upload the two faces photos of this board? One for components and one for soldering routing.

I have a  through-hole version also but it doen't use the Z-plate as heatsink for D8.

[rf-loop]

I have a  through-hole version also but it doen't use the Z-plate as heatsink for D8.

It was in very first versions  and if there is also this totally different  front end electromechanical  construction then scope  is perhaps this very extremely early version.

[mswhin63]

I have seen that noise while waving USB cables and HUB close to the scope.

[Carrington]

I have an older through-hole version of the power supply marked PCB-T115-J rev4 and 3.0 adapter.

You can make a picture and upload it to the forum.
Thanks. 

If the adapter is similar to the following link. Could you try to make the mod. indicated?
https://www.eevblog.com/forum/testgear/review-of-owon-sds7102/msg265881/#msg265881

[Carrington]

It was in very first versions  and if there is also this totally different  front end electromechanical  construction then scope  is perhaps this very extremely early version.

Sure sure, that explains everything...

AndrejaKo: Why you bought one of the first? Very bad, very bad...
I still wonder why I buy one ~7 months ago. 

[AndrejaKo]

Well the scope's serial number is SDS71021211XXX (came with 2.6.2 firmware if I remember correctly) and the main board, which has been replaced, is SDS71021209XXX. I got it in June 2012. It seemed like a good idea at the moment.

Anyway, I can't post pictures right now, since I don't have a digital camera at the moment and I assure you that you do not want to see photographs from my smartphone. I'm working on obtaining a camera, so hopefully by the end of the day, I'll be able to post pictures.

Also I must say that my computer is relatively close to the scope, so I'll take one more batch of screenshots in an empty room, just to be 100% sure that the noise isn't external.

[lemon]

11th week of 2012 is somewhere at the end of April.
If I remeber well, before this period the Owon had done some major changes to the hw. No you haven't the early version of this scope but you have a bbbbaaadddllllyyyy psu board!

If you can't upload photos, you can find some helping for TomC's psu diagram circuit (it is based on an newer version with smd IC) and if you wish I can to  upload some photos from mine version. I don't know perhaps to help they to some mods.

But it is very strange to me to have such a lot of noise with ferrite to carbon cable between psu and adaptor boards. If you used a large ferrite and you have a such noise perhaps this psu board is for trash! There is no way to pass the EMI Standards!

[TomC]

Hi there guys!

I decided to do some experimenting with ground noise problems, since it seems that I have a particularly "interesting" unit. I have a "thick" part of noise which is around 300 mV peak to peak and some huge spikes which are between 2 V and 3 V (even worse with 150 MHz probes I'm normally using). Their period is around 80 ms.  I have an older through-hole version of the power supply marked PCB-T115-J rev4 and 3.0 adapter. From time to time, there will be a huge single peak of 5 V to 6 V as well. One interesting thing is that this PSU is using the "Z-plate" as heatsink for D8.

To get to the source of the noise I connected one probe to the ground pin on the probe compensation connector and moved the other probe around the power supply. From what I can see, most of the peaks seem to be synchronized to the waveform on the coil marked L2.  Unfortunately, my power supply does not have coil L3 (which from what I can see if used as a filter) and instead seems to be filtered only with an electrolytic capacitor. I placed two 100 nF 1206 SMD capacitors and two 47 nF 0805 SMD capacitors over the electrolytic. In addition to that, near the power connector there were unpopulated places for ceramic through-hole capacitor and for one electrolytic capacitor on the -7.6 V rail. I placed one 470 µF capacitor and one 100 nF disk ceramic capacitor at the power connector. In addition to that, I placed same set of 4 SMD capacitors in parallel to the electrolytic.

Next, I added two 100 nF 1206 SMD capacitors in parallel with all electrolytic capacitors on the secondary side of the power supply. I also replaced the 1N4007 diodes mentioned few pages back with UF4007, but in my case that made no measurable improvement. My guess is that the reason for that is the large amount of noise that is cancelling all the (minor) improvements I made.

As you can probably guess, I'm poking in the dark here trying to come up with something that would help. My next plan is to get some 1 µF 1206 and 0805 SMD capacitors and place them in parallel with the ones I already have on the -7.6 V rail. 

First attached image is how everything looked like with only ferrite on the power cable going from the PSU to the adapter board.  Next two images are made after adding the capacitors on the -7.6 V rail and show relationship between the noise peaks and the waveform on L2.  Final image is what I get when I use same settings as on the first image. Unfortunately, there seems to be no improvement (in fact I could say it's a bit worse now).

So does anyone have any smart ideas what to try now? I'm starting to run out of ideas and am considering undoing the changes I made since I don't actually have any problems running the scope for my needs. It's just that knowing that the ground noise is there is bugging me.

AndrejaKo, Since you have all that noise around L2, I would check C9, C10 and C11 out of the circuit and replace them if necessary. The amount of noise you have seems more like a malfunction than ground noise. Just a thought!

[Carrington]

AndrejaKo, Since you have all that noise around L2, I would check C9 and C11 out of the circuit and replace them if necessary. The amount of noise you have seems more like a malfunction than ground noise. Just a thought!

The truth is that yes, it seems something like that.
Maybe rf-loop has seen something like this before.

[TomC]

TomC had you measure before changes what was the threshold for right triggering?

I think that these measures (noise + triggering) there is no way to obtain with the old psu board?

I hadn't test before to my scope the level for right triggering but with very low signals there is no way to examined because noise.
There is some documentation here (it demand login to see the captures) about before and after mods. With noisy gnd and long ground clip 50mVpp square signals there is no way to examined, after mods 1MHz/2mVrms sinus signal is almost perfect with a long gnd clip. But there is no reference about triggering.

It is interesting if other members have some tests before and after and specialy triggering operation.

I have checked this before the mods, unfortunately I didn't document it, wish I had! Relying strictly on memory, I don't think I had any luck with amplitudes below 75-100mVpp. I also remember that the waveforms with the long ground lead were distorted compared to the ground clip, specially at the lower frequencies where the density of the noise spikes is higher.

I'm also a member of ForumUp, I don't think I posted anything there yet, but I'm pretty sure I've seen some of your posts, and of course rf-loop's (aghp in that forum).

I would also like to see other members post their results for similar tests. After all, I think this is more telling of the scope's performance than just the GND noise test.

[TomC]

AndrejaKo, Since you have all that noise around L2, I would check C9, C10 and C11 out of the circuit and replace them if necessary. The amount of noise you have seems more like a malfunction than ground noise. Just a thought!

I would also check, R8 and C7, the snubber for D4.

By the way, the components I mentioned in both posts are referenced to the schematic in the Service Manual, which is a thru hole version similar to what you have. I attached a copy in case you don't have it.

[lemon]

AndrejaKo, don't forget to have in memo the following procedure.

Procedure for measurement of GND Noise
----------------------------------------------------
Make a Calibration procedure as manual refer (there is no nessesary if not the temperature variables ±5°C from the initial calibration)
Compensate the probe as manual described.
Change switch probe to 10x and set the same to the Probe Owon Menu (10X).
Connect together the probe tip and the probe aligator ground  to Probe Comp GND.
Fix the cable of probe like the attachment photo.
Turn CH2 off.
CH1 to DC, 50mV/Div.
Push Trigger 50%
Set the Acquire Length to 1M and the Acqu. Mode to Peak
Set the Horizontal speed to 100us
Set the measurement to Voltage peak to peak (Vp to Owon) and write the value.

[AndrejaKo]

First, sorry for the confusion. My readings were made using battery only! Next, my L2 is L3 on this new schematic and is same as L2 on TomC's reverse-engineered schematic! It's the coil used by the MC34063A to generate negative voltage. Compared to the new schematic, some components are missing on my board. First, there was only one green 470  µF 16 V electrolytic capacitor right next to the IC. There are (originally unpopulated) foorprints for another electrolytic and ceramic capacitors next to the connector.

Anyway, I found components that to me look like the snubber for the diode (but I didn't use that, since power is coming from the battery, right?). They are marked C4 and R11. C4 shows 2.287 nF, which is pretty close to expected value, from what I can see. R11 gives 21.7 ohm reading on my multimeter, so that too looks OK. The soldering of R11 was bad in my opinion, so resoldered it.

My problem with the capacitors is that I don' have an ESR tester at the moment and my multimeters can't measure large electrolytic capacitors. I'm making a crappy one from schematics of an old Poptronix kit, so maybe that will help and maybe not.

As for the ferrite, I'll post a picture of it when I get a camera. It's about the size posted maybe 20 pages back.

I'm off to get a camera now, so hopefully my next post will have photos of the PSU itself. Thanks for all the ideas!

[lemon]

@ AndrejaKo = I'll upload mine psu board PCB-T115-J rev6, probably will be help you a more becouse they differs only by revision.
From the forum of Electronics and Owon SmartDS oscilloscopes have the information that now all the new SDS have a probe gnd short as accessoires.
I have made mine by diy method...
 

[lemon]

@ AndrejaKo = about ESR of electrolytic capacitors on psu board.

First, I have the same concern about what values have they. If you will see at my older post I changed them with Nippon NCC KY Series (same uF/V) with a low ESR.
When I take off chinesse capacitors and measure ESR them, they had a very low ESR!
But the new ones had better behaviour at filtering to high peaks of noise. See my post 1302 at page 87 how they filtering all the peaks of the noise!
But, I believe you can achieved the same and better result with a better decoupling is some areas of circuit, there is no need to change the electrolytic capacitors.