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

[TomC]

I think, in these now showed test results there is not used any these modifications. If there is, then some capacitor is broken. With these modifications can not get these results what I see now there in signal pictures.

rf-loop,
These are test results with the new Owon PSU. Everything works OK as long as there is nothing connected to pin 5,  the bad duty cycle captures are with a probe or a 2" wire attached to pin 5 which acts as an antenna. Of course, I don't think it should be so sensitive, but it works OK as long as you don't attach anything extra to it. I suspect that the Owon engineer used resistors with too high a value for the voltage divider, probably to try to prevent power losses. Maybe lower value resistors with the same ratio would make this pin less sensitive.

[TomC]

Unfortunately, I have to stop my experiments for a while. I have some exams in the next two weeks for which I need to prepare and fixing the noise is currently draining too much of my resources.

Good luck with your exams, that comes first!
There'll be plenty of time later on to play!

[rf-loop]

I think, in these now showed test results there is not used any these modifications. If there is, then some capacitor is broken. With these modifications can not get these results what I see now there in signal pictures.

rf-loop,
These are test results with the new Owon PSU. Everything works OK as long as there is nothing connected to pin 5,  the bad duty cycle captures are with a probe or a 2" wire attached to pin 5 which acts as an antenna. Of course, I don't think it should be so sensitive, but it works OK as long as you don't attach anything extra to it. I suspect that the Owon engineer used resistors with too high a value for the voltage divider, probably to try to prevent power losses. Maybe lower value resistors with the same ratio would make this pin less sensitive.

earlier version (not know exactly what version this changed)
resistors was much lower value. Around 180 and 910ohm.

This output capacitor need also RF bypass C.

[TomC]

Today I started taking a look at the 8.4V converter. I found some surprises in the results and challenges in trying to scope them. The attachments are a set of signals I captured while scoping Q1's drain and D8's anode. Note that these captures are from the new Owon PSU that is currently installed in my scope.

The main challenge was dealing with the signal references, Q1 is in the primary side and is referenced to GND-A, D8 is in the secondary side and is referenced to GND-B. Note that connecting a ground lead to GND-A would cause a short circuit and as a minimum will blow the fuse, so that is out of the question. Connecting a ground lead to GND-B is not a good idea either, it probably wouldn't be catastrophic, but it will short out the CC current sense resistors (R14/R14A/R14B). So the only option is using GND-C which is also Chassis gnd. GND-C is separated from GND-B by no more than 125mV, so it's a fair reference for D8. However, it's a different story for Q1, because GND-A is nearly the same as GND-C half the time, but separated from it by 100s of volts the other half, it depends where on the mains AC cycle you are at. However, this knowledge is helpful, because you can focus on just the waveforms during the time that GND-A is nearly the same as GND-C and ignore the rest. Just use, for example, single trigger, until you capture something with the correct baseline.

Some of the captures were done while I powered the scope with my Variac. I wanted to see the effect of different mains voltages, and using the Variac I was able to vary the voltage between 100VAC and 140VAC. Unfortunately my Variac is almost as old as me and doesn't have a ground lug on it's power cord or on its output outlet. So I had to operate the scope without a ground and that changes the way GND-A is referenced. In this case it is floating in respect to GND-C and the signals referenced to it act as if the scope channel was AC coupled. So for those captures I manually set the signal's ground base line to the third graticule below the x axis. Of course in these cases the channel's ground marker should be ignored.

The surprises include the frequency, which varies depending on the mains voltage, and the converter's operation mode, which is CCM at low mains voltages, is in transition at around 120VAC, and is DCM at higher voltages.

#1 - Shows the effect of GND-A on Q1's signal. For this capture the scope was directly connected to the earthed mains (120VAC) and the persistence was set to 1s.

#2 & 3 - Shows that at 120VAC the operating mode is in transition, sometimes CCM, some times DCM. Also note the frequency, at 120VAC it's around 58KHz, even though the R7731A is programmed to operate at 65KHz.

#4 thru #9 - These were obtained using the Variac, and show the effect of the mains voltage on the operating mode. Also note the change in frequency. At 100VAC it's 65KHz and as the mains voltage increase the frequency decreases, the last image (140VAC) has a frequency of 56KHz.

[lemon]

...
The main challenge was dealing with the signal references, Q1 is in the primary side and is referenced to GND-A, D8 is in the secondary side and is referenced to GND-B. Note that connecting a ground lead to GND-A would cause a short circuit and as a minimum will blow the fuse, so that is out of the question. Connecting a ground lead to GND-B is not a good idea either, it probably wouldn't be catastrophic, but it will short out the CC current sense resistors (R14/R14A/R14B). So the only option is using GND-C which is also Chassis gnd. GND-C is separated from GND-B by no more than 125mV, so it's a fair reference for D8. However, it's a different story for Q1, because GND-A is nearly the same as GND-C half the time, but separated from it by 100s of volts the other half, it depends where on the mains AC cycle you are at. However, this knowledge is helpful, because you can focus on just the waveforms during the time that GND-A is nearly the same as GND-C and ignore the rest. Just use, for example, single trigger, until you capture something with the correct baseline.
...

TomC, you done very well documentation but I have confused a little.
Because I remember the blow of R7731 that I had before. I had the ground clip (multimeter) to Z-plate ... (230Vac).
From what I read the GND-A and GND-C is almost identical (they are connected to Z-plate together) but it is better to use only the GND-C or Z-plate.
Please, where was the probe ground clip at your measurement on Q1 and where was at the D8?

Note= the AOZ1094 has received and everything is OK.

[rf-loop]

About flyback freq. Did you note any Vdd variations related to AC powerline variations?
(R7731  freq  is some amount  sensitive for Vdd (and of course also temp) specially under 15V)
btw, what is this newest PSU version number?

[lemon]

...
btw, what is this newest PSU version number?

Don't laugh ( ), there isn't version number on it.
TomC here has upload some photos of upper and bottom of this psu.

[TomC]

TomC, you done very well documentation but I have confused a little.
Because I remember the blow of R7731 that I had before. I had the ground clip (multimeter) to Z-plate ... (230Vac).
From what I read the GND-A and GND-C is almost identical (they are connected to Z-plate together) but it is better to use only the GND-C or Z-plate.
Please, where was the probe ground clip at your measurement on Q1 and where was at the D8?

Note= the AOZ1094 has received and everything is OK.

Lemon, I'm really happy that your scope is fixed! Congratulations!
The ground leads for both probes were connected to GND-C (Z-plate) at all times. Please, don't connect the ground lead to GND-A, it's not connected to the Z-plate, and it will cause a short circuit similar to what you had before. See attachment, it shows what happens if you do this!

[TomC]

About flyback freq. Did you note any Vdd variations related to AC powerline variations?
(R7731  freq  is some amount  sensitive for Vdd (and of course also temp) specially under 15V)
btw, what is this newest PSU version number?

rf-loop,
I didn't check the IC's VDD during these tests, I will be doing that on the next set of tests. Theoretically, I don't expect significant variations, but I should know better than making theoretical predictions! Let's see if I get another surprise.

As Lemon said, Owon hadn't printed a version number on the new PSU board as of the date they shipped mine.

[TomC]

Today I only had time to measure the VDD pin of the R7731A with different mains voltages. The easiest way to do this is with a voltmeter, the ground probe on GND-A (Q1's heatsink), and the red probe on the cathode of D6 (which is connected to the VDD pin). Fortunately this time there were no surprises. The results were as follows:

100VAC mains --- 16.58VDC
140VAC mains --- 16.66VDC

I also wanted to view this voltage with the scope, however, this is not quite as easy. The problem once again is that if you connect the scope's ground lead to GND-A there is a short circuit. So again, I had to connect the ground leads to GND-C. I then connected CH1 to GND-A, and CH2 to the VDD pin. Then I used the math function CH2 - CH1 to view the voltage. The captures were done while I powered the scope with my Variac, and this time I improvised an earth ground connection so GND-A wouldn't be floating in respect to GND-C.

Note that when you scope anything on the primary side with the scope referenced to GND-C, the signals you see are riding on a 50/60Hz half wave rectified AC signal. This is normal, because GND-A is connected to GND-C via the rectifier diode D4. So that's the reason the attachments look the way they do. Nevertheless, the difference between the two signals is the voltage in question.

To keep the two signals within the confines of the screen I have then overlapping each other in some captures. In those cases it seems that only channel 2 is in the picture, however, CH1 is just hidden underneath. If either one of the signal is partially off screen the measurements are not accurate. This is illustrated on the last attachment.

#1 - Shows the VDD voltage with 100VAC mains, in this capture the signals are not overlapping and since they are referenced to the same vertical division you can visually see and measure the voltage difference between them.

#2 - This is the same as #1 but the signals are overlapping.

#3 - Shows the VDD voltage with 140VAC mains, in this capture the signals are overlapping and since they are not referenced to the same vertical division you can't visually see and measure the voltage difference between them.

#4 - This is the same as #3 but the signals are not overlapping. It shows the problem with accuracy when signals are too large for the screen.

Note that the cursors used to get the average math trace voltage use CH1's V/Div scale. You can't select math as the source for the cursors, one more thing I wish Owon would fix. As a result, since CH1 is set to 20V/Div, you have to divide by 10 to match the Math 2V/Div setting. This yields a voltage reading of 16.4V.

[rf-loop]

Without differential probe these measurenments are difficult, and also controller chip internal smoke leaks easy out and without internal smoke it can not work.
Later I will check also some things after I upgrade some scopes with new PSU and Adapter boards and I hope I have then time for do some mesurements.

There is coming some new, and I think, Owon have also learned something.
--------
 

[lemon]

Today I only had time to measure the VDD pin of the R7731A with different mains voltages. The easiest way to do this is with a voltmeter, the ground probe on GND-A (Q1's heatsink), and the red probe on the cathode of D6 (which is connected to the VDD pin). Fortunately this time there were no surprises. The results were as follows:

100VAC mains --- 16.58VDC
140VAC mains --- 16.66VDC
...

Also, I have done a measurement to R7731N (socket dip-8) on PCB-T115-J Rev6 with 230Vac mains (50Hz) and the Vdd was 18.09Vdc after a half an hour of operation. Starts cold with 18.5 almost and slowly stabilized to 18.09Vdc

[TomC]

Also, I have done a measurement to R7731N (socket dip-8) on PCB-T115-J Rev6 with 230Vac mains (50Hz) and the Vdd was 18.09Vdc after a half an hour of operation. Starts cold with 18.5 almost and slowly stabilized to 18.09Vdc

That's interesting, I wonder if the difference in voltage from the R7731A is due to the higher mains voltage or something else, like for example, a slightly different flyback transformer.

Lemon, if you got some time, could you capture waveforms of Q1 & D8 like I did in my previous post. I would like to see what they look like at 230VAC. I don't have a suitable step up transformer to do this myself.

[lemon]

TomC, with my pleasure but confirmed please the connections to my attachment photo.

[TomC]

TomC, with my pleasure but confirmed please the connections to my attachment photo.

Lemon, that's the correct connections to measure the VDD of the R7731N IC with the scope, you can post that too for a more complete set of measurements. However, the capture I had in mind was for Q1 and D8 like in this previous post:

https://www.eevblog.com/forum/testgear/review-of-owon-sds7102/msg282577/#msg282577

I was just mowing and I'm away from the lab for a little while (I'm using the laptop from the garage and I'm not allowed in the house until I clean up). As soon as I come in I'll give you more detailed instructions on how to connect the scope.

[lemon]

OK, TomC take your time that you are needing and I am here to measure whatever you want.

I prepare a work about AZ1094 circuit...it has a lot of interesting!

[lemon]

Well, as I have told before, I was thinking to search about AOZ1094AI circuits. All we know that this dc-dc converter takes the 8.4V from PSU board and convert to 5.6V output.

Before, I analyze the circuit and see how this various components effect the output noise, I've done some measurements to have them as reference. All these measurements is from modificate version 3.0 as described here before..

# Sample ref = is the gnd noise with Sample Acq. Method. Range 40-52mV
# Peak ref = is the gnd noise with Peak Acq. Method. Range 40-52mV
# Average ref = is the gnd noise with Average (16) Acq. Method. Range 30-34mV, the noise is smaller (<28mV) if time base set to 500usec vs 100usec.
# Noise 5.6 ref = is the noise at output of 5.6V (it has measured with short gnd clip at the second inductor position).
# L3_ref = is the wave captured on L3 inductor (on air)
# D127_ref = is the wave captured on D127 diode (on air)
# AOZ1094_ref = is the wave captured on IC 1094 (on air)

(it continues...)

[TomC]

OK, TomC take your time that you are needing and I am here to measure whatever you want.

I prepare a work about AZ1094 circuit...it has a lot of interesting!

That'll be interesting!

Your PSU board is different from mine, so I attached images of where I think it'll be easiest to hook up the probes. CH1 should be on D8's anode, and CH2 on Q1's drain. Both ground leads to the Z-plate. Make sure the probes are set to X10 because the peak voltage at Q1's drain may be more than 400V with 230VAC mains. While the scope is running you'll see an image like #1, jumping up and down. When the baseline of the CH2 trace is level with the ground marker it is as closely referenced to the Z-plate ground as you can get. To get a capture like #2 I used single trigger. Sometimes the baseline will be at the correct place, so when this happens I capture the image.

[lemon]

I returned back all the relate circuit about AOZ1094 to factory setup and take an output noise measurement on 2nd inductor position (right photo mark 3).
For the measurement, I used a short gnd clip.
Attention the initial version 3.0 there isn't 10uH inductor at this position, this inductor was adding by Owon (with the hand) after.



# Output Noise 5.6, with factory setup, no 2nd inductor = Ohh, what is difference and what peaks are they? Sure the designer was drunk! From 36.8mV jumped to 93.6mV!
The second wave (yellow) is the gnd-noise, both of tip and gnd grip on Z-plate.

(it continues...)

[lemon]

Before I continue, must analyze a little the topology of circuits and marks the differences with adapter version 3.2

I repeat the above compare photo btw the two adapter versions for easy. Left is version 3.2, right the version 3.0



As we can see, at the version 3.2 has corrected the badly routing and cutting of ground planes that previous version was made (point 1).
Removed the smd capacitor C36 (point 2), thus adding a second inductor 5.6uH with some smd decoupling capacitors (point 3).

The worst all of this is the badly ground plane routing. For this reason, the member Siri had suggested the return of diode D127. The member Carrington had suggested the joined btw the two parts of ground plane. Both of them have the same effort to output noise. Both of them connect the two ground parts to one.

Following photo shows how this joined effects the output noise to 5.6V path. Very well, 41.6mV from 93.6mV only with better ground plane, closely to 36.8mV ref. measurement.
The second wave (yellow) is the gnd-noise, both of tip and gnd grip on Z-plate.

(it continues...)

[TomC]

Following photo shows how this joined effects the output noise to 5.6V path. Very well, 41.6mV from 93.6mV only with better ground plane, closely to 36.8mV ref. measurement.
The second wave (yellow) is the gnd-noise, both of tip and gnd grip on Z-plate.

(it continues...)

Very interesting! 

I'm wondering why you are using 20mV/Div (restricts bandwidth to 20MHz), are the signals too small if you use 50mV/Div?

[TomC]

Today I wanted to measure the ripple and peak voltage at the output of the unregulated supply. This is best viewed with a scope. Once again, as in post #1817,  the scope's ground lead can't be connected to GND-A. So the ground leads are connected to GND-C, CH1 is connected to GND-A, and CH2 is connected to the cathode of D1. Again the math function CH2 - CH1 is used to view the signal. The captures were done while I powered the scope with my Variac, again using an improvised earth ground connection so GND-A wouldn't be floating in respect to GND-C.

#1 Shows the peak voltage & ripple with 100VAC mains.

#2 Shows the peak voltage & ripple with 125VAC mains.

#3 Shows the peak voltage & ripple with 140VAC mains.

[lemon]

OK, TomC take your time that you are needing and I am here to measure whatever you want.

I prepare a work about AZ1094 circuit...it has a lot of interesting!

That'll be interesting!

Your PSU board is different from mine, so I attached images of where I think it'll be easiest to hook up the probes. CH1 should be on D8's anode, and CH2 on Q1's drain. Both ground leads to the Z-plate. Make sure the probes are set to X10 because the peak voltage at Q1's drain may be more than 400V with 230VAC mains. While the scope is running you'll see an image like #1, jumping up and down. When the baseline of the CH2 trace is level with the ground marker it is as closely referenced to the Z-plate ground as you can get. To get a capture like #2 I used single trigger. Sometimes the baseline will be at the correct place, so when this happens I capture the image.

TomC, thanks for helping...

Here it is what you ask.

[lemon]

Following photo shows how this joined effects the output noise to 5.6V path. Very well, 41.6mV from 93.6mV only with better ground plane, closely to 36.8mV ref. measurement.
The second wave (yellow) is the gnd-noise, both of tip and gnd grip on Z-plate.

(it continues...)

Very interesting! 

I'm wondering why you are using 20mV/Div (restricts bandwidth to 20MHz), are the signals too small if you use 50mV/Div?

Your thought is right, this is my  . Unfortunately I haven't the reference noise with full BW but I'll see after...
Again some captures with full BW.
There is D217 180° turn at the two captures, at the first there is no any 2nd inductor, at the second there is a 10uH inductor like Owon does by hand.
The result at the second captures is worst and I don't know who from Owon thought that solution!

[lemon]

I think is time to analyze the circuit according to datasheet and evaluation schematic board.

I have enclosed two attachment schematic, the typical and the evaluation.

The Vin (pin1) of AOZ1094 demands some decoupling capacitors. The typical has 22uF electr. caps, the evaluation has two 22uF electr. caps parallel. Owon has a 100uF electr. with two parallel smd caps (C33, C95) with unkown values. I think there is a need to measure the noise and ripple at that position (at the end of C33 caps).

The pin6 is enable pin, Owon just connect with Vin, typical and evaluate schematic set a 1nF decoupling capacitor. I believe that this must adding.

The pin5 is external loop compensation pin. Here the datasheets seems to be identical with the choice of Owon. The evaluate schematic set the position C7 open for tuning (?).

The pin2 & 3 are two grounds, digital and analogue respectively.

The pin4 is used to determine the output voltage via a resistor divider between the output
and GND FB pin. Here the choice of Owon is good with one resistor more adding at the output (R21). The evaluate schematic set  the position C8 open for tuning.

The pin7 & 8 are joined together. From this point Owon puts the D127 to gnd and through the L3 inductor with 3 caps parallel, C36/5uF, C45/12uF and C44/470uF.
This way differs some from evaluate schematic. At the evaluate schematic there is RC parallel to D127 and at the output there are two electr. caps of 22uF. I think that needs 1uF smd parallel to diode. The output needs more investigation with two inductors or pi filters.