OWON VDS1022I Quick Teardown (versus the Hantek 6022be)

[Gyro]

Ah ok, I read it as probes disconnected, probes shorted and GND - what I meant was probe tips Ch0 and Ch1 clipped together, ground clips disconnected.

The GND setting just does a flat line in s/w by the way.

EDIT: Screenshots for 5mV and 10mV added...

[camomille]

always screenshot X25
I do not see any problem !!

[camomille]

same conditions as you

[Gyro]

Very confused! 

I clearly need to investigate.

EDIT: I wonder if it's a H/W version thing. Or a fault developed on my and Nullarbor's scopes?


EDIT1:  What can I say guys, I seem to have got into the situation where ONLY Owon original V1.0.29 and V1.0.30 produce the correct scaling on Ch1 5mV. Neither camomille versions (original or portable), indman's composite 1.0.23 with V1.0.30 libraries, or even florentbr scale correctly for me. The behaviour is 100% consistent, switching between versions without disconnecting the scope or factory reset. Even going from Original V1.0.28 to V1.0.29 and back is consistent both ways. I am totally lost!

[indman]

Gyro, сan you show us the calibration window 5mV by pressing Ctrl+L on version 1.0.23 and version 1.0.30?

[Gyro]

@indman:

Thanks very much, you cracked it! No need for screenshots - in every non-working case, Ch1 5mV Coarse Gain is set to all zeros. In working versions, it is set to a sensible value - within a couple of digits of Ch2.     [EDIT: Changing the gain to a sensible value restores the sensible trace, confirmed]

I'm not sure how it has happened, or maybe worked its way across versions that I thought were ok. The V1.0.28 and V1.0.29 were fresh unpacks of the Owon Setup.exe's, but it is clearly the reason for what I've been seeing!

Many thanks - my brain was so fogged that I needed a quick sleep. I didn't even think of checking that the cal data was sensible.


P.S. I am wondering if the factory default cals maybe get stored in Flash in the ARM processor in each unit, that might possibly explain a 'cross contamination'. Maybe autoset has a fine tuning effect based on those settings, I don't know though.

[Gyro]

Follow-up:

I think we maybe don't fully understand how calibration settings work. Once I changed the Ch1 5mV settings and saved - right most button,  it is also works with Alain's and your versions, even after unplugging and changing USB ports! [EDIT: and previous Owon versions too]

Strangely the florentbr version still shown incorrect Ch1 5mV. 

A few tentative conclusions:

- The calibration data is stored in some non-volatile place. This could either be in Flash on the unit (which seems likely to me as Owon would want to do a basic cal on each unit during test), or some global (registry?) setting on the system. From the registry dump that you did yesterday, I don't see that this is possible.

- Assuming the above is correct, the florentbr version is not interrogating the unit to get factory setting and is maybe holding them somewhere else?

- Owon V1.0.29 - V1.0.30 are handling calibration data in a different way to previous versions ? ??

Still a little confused about the above thoughts.

[indman]

Strangely the florentbr version still shown incorrect Ch1 5mV. 
- Assuming the above is correct, the florentbr version is not interrogating the unit to get factory setting and is maybe holding them somewhere else?

Gyro, Is it possible that the explanations in the readme.md file from the florentbr can make a hint?
* moved calibration to user folder
* moved settings to user folder
* moved preferences to user folder
It would be very useful to understand correctly the purpose of the process of maintaining calibration settings.

[camomille]

@Gyro

If it helps you, I attach my original file Flashmemory.txt

https://ufile.io/2lajynvg

I think that if the flashmemory.txt file does not exist in the flash_txt folder, the VDS1022 must launch an "auto_self_cal" on startup.
or in Utility, you can launch a "Self Cal"

Alain

[Gyro]

@Gyro

If it helps you, I attach my original file Flashmemory.txt

https://ufile.io/2lajynvg

I think that if the flashmemory.txt file does not exist in the flash_txt folder, the VDS1022 must launch an "auto_self_cal" on startup.

Alain

Thanks Alain. I've downloaded the file, but suspect that it is related to your specific unit (hint, keep a backup of it). I was able to use the gain value I saw in V1.0.29 to set it for the other versions.

Strangely the florentbr version still shown incorrect Ch1 5mV. 
- Assuming the above is correct, the florentbr version is not interrogating the unit to get factory setting and is maybe holding them somewhere else?

Gyro, Is it possible that the explanations in the readme.md file from the florentbr can make a hint?
* moved calibration to user folder
* moved settings to user folder
* moved preferences to user folder
It would be very useful to understand correctly the purpose of the process of maintaining calibration settings.

Interesting, yes there do seem to be multiple files involved in the calibration data. As I mentioned back in Reply #468, I have previously seen some new and unexpected zero offsets that I hadn't seen before. Looking at file dates, it was apparent that both  '\flash_text\flashmemory.txt'  and '\configuration\com.owon.uppersoft.dso\pref.properties' appeared to be involved, I needed to copy both of those from my previous version to get the offsets looking normal again (they could of course also be trimmed in the cal menu).

From Alain's translation of the calibration screen, it is clear that there are buttons to save to flashmemory.txt only, or write to the device (scope) and Factory settings. What doesn't seem clear is if/when the flashmemory.txt and device data get exchanged or overwritten.

Without fully understanding this I suspect that there may be a danger of accidentally 'infecting' each other with our own scope calibrations! 


P.S. Sorry to you both for your time taken in searching for a non-existent bug!

[camomille]

in Utility, you can launch a "Self Cal"

[Gyro]

Yes, I know (there is the centre 'Calibration' button for it in the calibration screen too, that does the same thing). From my experience with the zero offsets that I mentioned earlier though, it doesn't seem to do a full accurate cal - I think it is based on the saved factory parameters (I tried both self-cal and factory reset). The zero line result isn't optimal afterwards anyway.

It's not a big issue to calibrate it anyway (Indman did it) - it's only an 8 bit scope (like most) so it's never going to have DMM type accuracy. Even on the highest range, it's only 5V/div (X1) so no high voltages are needed. I'm still a little confused about the difference between 'Zero Amplitude' and 'Zero Compensation' they seem to have the same effect but one is maybe ADC zero and the other input opamps offset for best linearity. I need to have a play.

[camomille]

To help a little, I tried to follow the actions of "Gain", "Amplitude" and "Compensation" but without success.
What I learned is that "zeroamplitude" in the program changes its name and is called "Step" (as in flashmemory.txt.
If it can be useful

[Gyro]

Thanks Alain, yes "step" sounds useful. It could be that Zero Amplitude, in conjunction with Coarse Gain can be used to optimize the ADC scaling (eg. as close to 1 bit per mV as possible) and the Zero Compensation is then used to bring the line to zero.

We'll learn all there is to know one day! 

[camomille]

By analyzing the diagrams of the circuits (analog side),
 there are few possibilities of programming of these circuits,
I see only the 2 reference voltages of the AD9288 and the gain and the offset on the 2 AD603.
I think it corresponds to our 3 programmable voltages.

[Gyro]

Good first interpretation.

Looking at my unit, the AD9288 RREFinA and REFinB are tied to REFout (pins 5,7,6), that makes sense as it's the most accurate reference source available.

With those two references fixed, I think that all three programmable values relate to the two AD603s. Checking around the AD603s, Pin 2 (GNEG - negative voltage increases gain) on both channels are linked together and decoupled to ground - these must be connected to a bias network somewhere (I didn't fully trace it out) and so I don't think they are involved in calibrations (there is no common cal setting for both channels).

Pin 1 (GPOS - positive voltage increases gain) of each AD603 is connect to an RC network, which in turn is fed by one half of an SGM8292 dual high voltage rail to rail opamp, on my unit, on the back of the board, near the ARM CPU. These form gain networks which are then driven by PB0.5 and PB0.7 port pins on the CPU. I Suspect that these are controlling the AD603 gain pin levels using PWM (the alternate functions listed on PB0.5 and .7 list ADC but not DAC).

I think this brings us back to the Firmware controlling gain and, zero and offset via those two pins, doing the appropriate compensations in code. Coarse Gain is fairly self-evident. I suspect that Zero Amplitude and Zero Compensation work together, to give both zero, and equal measurements / deflection for equal positive and negative voltages.

That's my best guess anyway.

(My earlier thoughts about Zero Compensation compensating opamp offsets is probably wrong - they would only be relevant at low signal levels (Zero Compensation exists on all ranges) and would be automatically included anyway in the the Zero adjustments provided).

Other interpretations very welcome!


P.S. I noticed in the AD9288 datasheet that a 10 bit pin compatible upgrade is available. That could be fun if we had full control of the Firmware and FPGA code. Alas...

[camomille]

I will analyze your deduction, in this area you are much faster than me, I started to make the schema on paper around the AD603.
I made a discovery, the circuit "8061" (AD8061 / 8063) can isolate each channel by pin 5.
FPGA is just a logical matrix, but to decode it is another story.

[camomille]

For me, it is the outputs of the 2 DAC (IDAC0 / IDAC1) that go to SGM8272, respectively PB0.10 pin 47 and PB0.11 pin 48 of the 64 Pin TQFP SIM3U1xx .

[Gyro]

  Doh, yes, of course, you're right - I was looking at the 80QFN datasheet footprint, I was too busy balancing the board vertically while probing both sides!  Yes, IDAC0/1 it is then. It makes a lot more more sense than PWM.

I have to take the screws out again later to check the AD8061s. According to my old photo they are indeed marked 8061 and according to the datasheet, only the 8063 has the disable pin.

I am still wondering the about the 'mechanics' of the 'Self Cal' operation. Whether it is an offset cancellation (I think what Tek call 'signal path compensation') or whether it is gain too. I suspect that some of the secrets are hidden under the shield.

[camomille]

But you too are right because there is a second SGM8272 underneath and he is attacked by logical outputs.
I just changed the image above.

it is good 8061 but surely a manufacturer with the box of 8063

[Gyro]

Ah, good - it's always nice to be a little bit right.

The AD8061s are definitely 8061 on my board revision - SOT23-5 pin package, not SOT23-6 (8063), so lacking enable.

I'd never paid any attention to the SGM8272 on the top side, from its position I think I assumed that it was PSU related. Following the outputs leads me to the two LMH6551 differential output opamps which are the stages between the AD603s and the AD9288 ADC. On my board they go to a large tacked-on Tantalum decoupling capacitor before going to the bias network on the LMH6551 Pin 1 (input -). This looks to be in non-inverting configuration, so it must be an offset adjustment for the LMH6551. [EDIT: Vcm (common mode input) of the LMH6551s are driven from the ADC Vref Out, as Fig. 33 in the datasheet (Driving an ADC)]

I suppose this makes sense because by the time the signal has reached the LMH6551 it has already been scaled to a fixed range by the AD603, input attenuator relays etc. so will work as a range independent offset adjustment going into the ADC (but still adjustable on a range by range basis).

Ok, so now we have the Coarse gain adjustment (PWM?) driven into the AD603s from PB0.0 / PB0.2 via the bottom SGM and offset adjustment at the LMH6551s driven from PB0.10 / PB0.11 (10 bit IDAC?) via the top SGM. It's making sense.

EDIT: The picture of my board revision (from Reply #12), showing the tacked-on Tants...



P.S. I suspect that your board doesn't have the U29 footprint - an interesting evolutionary relic!

[Gyro]

One more thought... The s/w will normally expect the ADC exactly half scale code to be Analogue zero. I wonder if a change in what it expects here could be regarded as a 'Zero Amplitude' adjustment?

EDIT: No, too many digits. I was forgetting it's only 8 bits.

[camomille]

Just a small precision before attacking the day, the DACs are 10 bits current output (0.5, 1, 2 ma) .
The circuits marked "YJAA" are SGM2019A.

[Gyro]

Yes, lots of nice local regulation on the analogue supplies.

I've just been doing some quick measurements and found that both Zero Amplitude and Zero Compensation affect the IDAC outputs to the LMH6551s. I was hoping that it would be just one or the other.

As expected, the Coarse Gain adjustment changes the GPOS gain adjustment pins of the AD603s.


P.S. I still wonder how it performs the Self-Cal. Watching the screen, it seems to adjust  the zero positions of both traces at the centre of the screen, and then do the same again at some positive displacement on the screen. It could be that somewhere (under the screening can?) it switches in a positive reference, tapped of an analogue supply so that it can determine what offsets there are in the system.

Alternatively, it might be offsetting by changing the Zero Amplitude setting and then optimising the Zero Offset setting at two positions. I'm not sure why it would do this (apart from maybe optimising ADC linearity?). With a 10 bit IDAC it should be possible to set the zero offset of an 8 bit ADC quite closely.

I can't see any way of Self-Cal setting the Coarse Gain. It would either need a separate reference or feed back of the ADC ref. I think this is a manual cal only.

[camomille]

You have a separate reference in SIM3U156 :

PB0.8 pin 49 VREFGROUND
PB0.9 pin 48 VREF