New Hantek DSO2X1X models?

[dart1280]

There is no AWG calibration option on the  D10 I bought earlier this week, there is only a single calibration routine on my D10 (220210.00   3204) and that runs through all the voltage and timebase settings presumably setting them all to zero offsets - I didn't notice the AWG doing anything during the calibration.

[Algoma]

Hantek support likely has records of the original calbration values linked to your serial number that can be restored with a custom update package. At least for the units that are supposed to have them.

There has been some previous discussion in setting calibration values for the AWG. The DAC in this scope (DAC902E) is common to other Hantek scopes. Measureing the required Gain and Offset calibration values are likely a similar procedure.
https://www.ti.com/lit/ds/symlink/dac902.pdf

[JohnC]

Hantek support likely has records of the original calbration values linked to your serial number that can be restored with a custom update package. At least for the units that are supposed to have them.

There has been some previous discussion in setting calibration values for the AWG....

Thanks for the reply, Algoma. I'm not worried about AWG Calibration of a DSO2D1x, since those had "official support" of that function. If I buy a 2D1x I can just back up the Cache and EEPROM data to keep track of that.

I've read the earlier discussion about the calculation of AWG offsets, etc - however currently that seems too difficult. Since the AWG Calibration is stored in "firmware", I'm concerned whether the 2C1x models (with AWG hardware installed) will actually work OK after the hack conversion to 2D15 is performed.

It seems that a reasonable proportion of the C models have AWG hardware installed, and we know that the 100MHz models are identical to the 150MHz. So I'm wondering if Hantek is calibrating all the mainboards for AWG, and then they "de-spec" them by setting the HardwareID (model) after assembly of the C models.

Honestly I'm struggling to justify the extra AU$72 (about CA$70) for a pretty crude AWG in a genuine 2D15, so I'm really interested in hearing about how accurately the AWG works on a "converted" 2C10. 

[JohnC]

There is no AWG calibration option on the  D10 I bought earlier this week...

Yeah, my understanding is that there is no (practical) way of calibrating the AWG. I wish I could get a D10 like yours, however they won't be in stock locally until the week after the special "20% off" deal ends. So I have the choice of a 2C10 (and take a risk on AWG) or buy a 2D15 for AU$72 higher price. 

[vistorik]

Actually there is a practical way of calibrating the AWG.  Please read the attached file Calibration_upk.doc. The calculations are based on the ideas of michal_nt1.
The Calibration_upk.zip has the python's script that makes a dso3kb_calibration_restore.upk file.
To work on Windows you will need an embeddable package of Python 3. To work on Linux, you will need python3 installed and gnupg.

[JohnC]

Actually there is a practical way of calibrating the AWG.  Please read the attached file Calibration_upk.doc....

WOW, thanks very much for that information vistorik. I haven't seen that before, only the earlier post by  michal_nt1.

The Word Doc confirms my suspicion that the 2C1x don't contain the AWG Calibration values. The process using Python etc is somewhat complicated but reasonably straight forward I guess.

So, I suppose the question remains as to whether the Hantek Scope AWG is even worthwhile worrying about. I was looking for an upgrade to my old XR-2206 based Generators.

I'd appreciate comments about how it compares to cheap stand-alone units like FY-2300, JDS6600 or even the MHS-5400.

[DavidAlfa]

I think I can make a standalone UPK that does everything, without requiring a computer, python, etc.

Something like:
- Install "dso3kb_calibrate_gen.upk". The first pass it will reset calibration and reboot the DSO.

- Make the measurements and create "values.txt" in the usb drive, ex.:
     AY="632"
     BY="-618"

- Run "dso3kb_calibrate_gen.upk" again. It'll know it already resetted the calibration, running the second part this time,
  reading "values.txt" and calling a small program to compute the new values, generate "dds_calbration.dds" and update the eeprom.

[DavidAlfa]

Edit: Found the eeprom structure in the latest 4PDA post, stored at 0xE0 as plain text.

The C program seems to be almost finished.


This is my original calibration. Values are A=0.833157 B=13.485453.
For 1Vpp I get 976mV, no offset (or negligible):





First glance, running in the scope:

Code: [Select]

# cal
    Hantek DSO2x1x Wavegen calibration tool v0.1
    Usage:
    cal -r            - Reset calibration data
    cal -s <A> <B>    - Manually set calibration
    cal -c <Hi> <Lo>  - Generate new calibration
    cal -h | --help   - This help screen

    A reboot is always required to reload the new calibration data!

    For the calibration to work, you must first reset the calibration, reboot,
    then set the WaveGen to 1KHz, 1Vpp, probe in 1x DC and measure:
    Hi:  Peak positive voltage in mv
    Lo:  Peak negative voltage in mv

    Then re-run the tool with the measures, ex. "cal -c 568 -630"


Resetting the calibration does A=1.000000 B=0.000000:

Code: [Select]

# cal -r
Reseting calibration data
Out:
        A:1.000000    0x3FF0000000000000
        B:0.000000    0x0000000000000000
EEPROM:
        1.0000000.000000
Success


Code: [Select]

# xxd /cache/dds_calbration.dat
00000000: 1000 1000 0000 0000 0000 0000 0000 f03f  ...............?
00000010: 0000 0000 0000 0000 0000 0000 0000 0000  ................
00000020: 0000 0000 0000 0000                      ........


Code: [Select]

# i2cdump -y 0 0x50
...
d0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff    ................
e0: 31 2e 30 30 30 30 30 30 30 2e 30 30 30 30 30 30    1.0000000.000000
f0: 10 10 18 05 15 04 22 ff ff ff ff ff ff ff ff ff    ??????".........

Then I rebooted and made the measurements. For 1KHz 1Vpp, I got 544mV and -608mV:





Ran the calibration:

Code: [Select]

# ./cal -c 544 -608
In:
        AY:544.000000  BY:-608.000000
Out:
        A:0.868056    0x3FEBC71C71C71C72
        B:7.372800    0x401D7DBF487FCB92
EEPROM:
        0.8680567.372800
Success


Code: [Select]

# xxd /cache/dds_calbration.dat
00000000: 1000 1000 0000 0000 721c c771 1cc7 eb3f  ........r..q...?
00000010: 0000 0000 0000 0000 0000 0000 0000 0000  ................
00000020: 92cb 7f48 bf7d 1d40                      ...H.}.@


Code: [Select]

# i2cdump -y 0 0x50
     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f    0123456789abcdef
...
d0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff    ................
e0: 30 2e 38 36 38 30 35 36 37 2e 33 37 32 38 30 30    0.8680567.372800
f0: 10 10 18 05 15 04 22 ff ff ff ff ff ff ff ff ff    ??????".........

The amplitude is better than original... but the offset isn't, 40mV.




Repeated with the probe in 1x mode, the lowest offset I got was 32mV. Not terrible, but original was better.
Perhabs the algorithm isn't completely correct?


Code and precompiled binary can be found in the src folder.
Depends on i2ctools, but will generate dds_calbration.dat without it, enough for testing.

Once I completely know this is working as expected, I'll make the automated UPK package

[vistorik]

The Word Doc confirms my suspicion that the 2C1x don't contain the AWG Calibration values.

This is not entirely so. At least the last four devices 2C10 reported by the participants in 4PDA forum (e.g. here: https://4pda.to/forum/index.php?showtopic=1018961&view=findpost&p=115355039, https://4pda.to/forum/index.php?showtopic=1018961&view=findpost&p=115643634) had all AWG hardware inside, there was a real calibration file, and the calibration coefficients were also on the PT24C02. But this of course does not guarantee that you will get the same.

[DavidAlfa]

The Word Doc confirms my suspicion that the 2C1x don't contain the AWG Calibration values.

This is not entirely so. At least the last four devices 2C10 reported by the participants in 4PDA forum (e.g. here: https://4pda.to/forum/index.php?showtopic=1018961&view=findpost&p=115355039, https://4pda.to/forum/index.php?showtopic=1018961&view=findpost&p=115643634) had all AWG hardware inside, there was a real calibration file, and the calibration coefficients were also on the PT24C02. But this of course does not guarantee that you will get the same.

The real question is: Did their devices really restore the AWG calibration from the eeprom after running platform-tools?
Currently the eeprom seems more like a scratchpad, some peple have their model and SN, others have some dates, others random stuff...


30mV error out of 1V is better than nothing, right?

Anyways, this doesn't seem to be the correct way to get F(x)=A(B+x)    Why B/10?

    A1 = Vmax - Vmin
    B1 = -Vmax - Vmin
    A = 1000.00 / A1
    B = B1 / A / 10.00

Anyways, new package, Generator calibration
Almost standalone, runs in two passes. It only requires a computer to write down the values to the usb drive.

Thanks 4PDA fellas!



Edit: The algorithm is definitely wrong.
After calibrating, try setting a 500mV signal. The offset is huge.

Original:
    A=0.833157
    B=13.485453
    Vgen=500mV, Vmax=252mV, Vmin=-244mV, Offset=8mV


Scope warmed for one hour, calibrated.
Generator calibration cleared, 1Vpp gives 568mV -616mV

Following andry2pda algorithm suggestions:

Code: [Select]

      A1 = AY - BY;
      A = 1000 / A1;
      B1 = -AY - BY;
      B = B1 / A / 10.0;
    A=0.844595
    B=5.6832003
    Vgen=500mV, Vmax=236mV, Vmin=-264mV, Offset=-28mV

Code: [Select]

      A1 = AY - BY;
      A = 1000 / A1;
      B1 = (-AY - BY)/2;
      B = B1 / A / 10.0;
    A=0.844595
    B=2.841600
    Vgen=500mV, Vmax=232mV, Vmin=-272mV, Offset=-40mV

[JohnC]

... Edit: The algorithm is definitely wrong.
After calibrating, try setting a 500mV signal. The offset is huge.

Thanks so much for the work on this, David 

Yes, I agree that the offset algorithm is definitely not correct, however the P-P calculation is perfect. Over the next few days I will check the original DOC that they issued, and see if I can work out where they have gone wrong.

Anyway, I bought a 2C10 which just arrived - haven't even plugged it in... but I teared it apart. 

It's motherboard rev DSO2D15_V1.7 2021.1.14 (like they all seem to have) and there is AWG HARDWARE inside. I will boot it up and check AWG Calibration values, etc. Perhaps if mine has a AWG Cal in Cache, we can "reverse engineer" the formula to calculate those values.

I'll be back soon with more info.

Cheers, John

[DavidAlfa]

Run Eeprom backup tool and Backup builder...
Send the files and I'll tell you if it contains calibration data 
(Or just run DSO2D15 converter after you made the backups, if it matches the adjusted output, it's fine then!)


@vistorik, I read you in 4PDA, I have account there, but they use russian captchas, so it's very hard for me to log in.
The result of the calibration can be seen in the file log.txt, in the usb drive. It's generated when running the calibration.
I don't really see why it would require generating the UPK package with the new calibration, when the tool does it in-place.
As it writes the values into the eeprom, you will never need to run the calibration again, the scope will read them whenever dds_calbration file is missing (After runnning platform-tools, etc).

I updated the calibration package, now instead 1Vpp, it can take any reference voltage.
Yep, adjusting the AWG using 600mVpp seem to do better:

[JohnC]

Run Eeprom backup tool and Backup builder...
Send the files and I'll tell you if it contains calibration data 
(Or just run DSO2D15 converter after you made the backups, if it matches the adjusted output, it's fine then!)
[SNIP]

Thanks, David 

I've reviewed the 4PDA thread and original DOC file, and I'm not convinced that the AWG offset calculation makes sense. I will comment further once I get my AWG running.

My 2C10 came with FW 3204 SW 220210

Is there still a need to run DSOFLASH backup? That's quite a complex (more risky?) process compared to your BackupBuilder UPK.

The bloody thing has a crack in the top plastic from transit damage. I am going to try to get a partial refund, rather than returning/exchanging it with the seller.

To allow an easy users to revert to defaults (eg: for warranty return), how difficult would it be for you to make 3 new UPKs:
dso3kb_2C10_conversion.upk
dso3kb_2C15_conversion.upk
dso3kb_2D10_conversion.upk

Thanks again, John

[vistorik]

To allow an easy users to revert to defaults (eg: for warranty return), how difficult would it be for you to make 3 new UPKs:
dso3kb_2C10_conversion.upk
dso3kb_2C15_conversion.upk
dso3kb_2D10_conversion.upk

Here are all these UPKs. But instead of using these rollback conversions, before changing the model number you can use the dso3kb_system.inf_backup.upk which creates a dso3kb_system.inf_restore_CNXXXXXXXXXXXXX.upk file, which restores the /cache/system.inf file.

[DavidAlfa]

No need of such, that's what Backup builder does, make your backup before doing anything, then whatever happens, restore it.

Keep it easy, I don't understand why people started flooding with lots of packages for every little thing, they're doing it worse.

[JohnC]

No need of such, that's what Backup builder does, make your backup before doing anything, then whatever happens, restore it.

Hi David, sorry to cause problems here!

One of the issues here is that there is little/no information inside the packages that shows what each UPK actually changes. For example, the readme for "dso3kb_2D15_conversion.upk" doesn't state what files get changed and how to revert back afterwards. So I was unsure that BackupBuilder could restore that factory Model Number data.

Anyway, I installed the 2C15 upgrade on my 2C10 and here is info from factory AWG calibration:

Process: Upgraded 2C10 to 2C15, then Restore Defaults, then Utility > Calibrate (the DSO). Probe at 1x.

AWG: Sine 1kHz @ 600mV
100mV/Div gives AY 296, BY -296   = 592mV (offset zero)
  50mV/Div gives AY 294, BY -296   = 590mV (real offset -1.0mV)
  20mV/Div gives AY 289.6, BY -296 = 585.6V (real offset -3.2mV)
  10mV/Div gives AY 283.2, BY -290.8 = 574mV (real offset -3.8mV)

AWG: Sine 1kHz @ 1000mV
200mV/Div gives AY 504, BY -496   = 1000mV (real offset +4.0mV)
100mV/Div gives AY 496, BY -496   = 992mV (zero offset)
  50mV/Div gives AY 488, BY -494   = 982mV (real offset -3mV)
  20mV/Div gives AY 480.8, BY -492.8 = 973.6V (real offset -6.0mV)
  10mV/Div gives AY 472.8, BY -483.6 = 956.4mV (real offset -5.4mV)

Note 1. At low mV/div I measured AY and BY separately after scrolling V Position. In each case I cursored to the outside surface of the noise envelope.

Note 2. My "Real Offset" is the offset from Zero Volts ie: (-AY - BY)/2

WHAT CAN WE LEARN from my test?

1. The factory offset calibration must be for 100mV/DIV

2. When display mV/Div is reduced, the Measured Amplitude reduces, and offset shifts to negative. 

3. The DSO measurements cannot be trusted, so it's pointless to use it to calibrate it's own AWG 

Cheers, John

[DavidAlfa]

One of the issues here is that there is little/no information inside the packages that shows what each UPK actually changes.

So apparently this is not enough...

Before doing anything, save your important data!
- Make a flash backup (Check DSO Flash Tool below)
- Run Backup Builder (In Mods section)
- Run Eeprom backup (In Mods section)

Do I also have to write a complete Wiki page about every single low-level detail? Read the Hacking thread.

Most people here is lazy as f***, not even reading the FAQ most of the time, what the heck, I already did too much

[aztecboy]

David has documented a wealth of info in the hacking thread, it really is worth reading in it's entirety.

[DavidAlfa]

I've moved the FAQ here (Page 8 ), there was no sense on having it separately, getting lost, and being spammed anyways  .

[arturmariojr]

Well,

Last weekend I placed some few more capacitors (.47uF tantalun) to filter output from switching power suppluy (SPS) of this scope, trying to get less noise measurements. There was no improvement at all.

So, I'll replace every SPS on mother board and display backs for linear ones, using still the main SPS and, for the last moving, replacing it too.

At each replacement, I'll try some configurations to see if the terrible waveforms I'm getting with my scope, with an anoying disturbing signal, get better and clean shapes.

Any advise?

[pcprogrammer]

You can't replace some of the switchers because they generate voltages higher then what the main supply gives.

This it the case for U37 on the user interface board, which is for the LCD bias voltages. Creates both positive and negative voltages.

Most likely also the case for U34 which generates the voltage for the back light. This one uses a PWM signal to control the brightness.

See the thread on the schematics. https://www.eevblog.com/forum/testgear/hantek-dso2xxx-schematics/msg4250503/#msg4250503 The ones for the user interface board are done. Working on the main board now, but have not looked at the power supply yet.

[arturmariojr]

Thanks for replying!
I haven't seen that thread yet!

Excellent! Someone is trying to reverse schematics, as Hantek seems have forgotten     their scope. I coudn't see the voltages there.
(Hey, the "someone" working on schematics is you! Congratulations and many thanks!!!  )

As you said, I'll try first the simple SPW (switched power suppply) first: 1,1V and 1,9V . I thought to connect the wire from linnear PS directly to probe points on pcb and remove toroids and SIC ICs from board.

If these changes do not solve the problematic noise, I'll try another trafo with needed voltage for display LCD and brightness. Let's see if it will be possible, otherwise, it is very possible sell my scope as parts unit, as it is unusable and untrustable at all  .

Best wishes with your hard work!
Artur

[pcprogrammer]

I coudn't see the voltages there.

I did not do the calculations or live measurements on it yet.

It is quite a bit of work to go through all the components and connections with a microscope and a multi-meter, but I'm getting there eventually.

Success with your noise canceling quest

[DavidAlfa]

The scope has some noise, but on the average cheap spec, completely usable, or did you expect 16-bit 10Gsps for that price?
It's just bad for very little signals, ex. 50mV and below.
People tried feeding linear power supplies without any noticeable difference.
The PCB isn't great either, most signals are on the top layer instead being "sandwiched" between gnd planes, thus generating and being exposed to interferences.
If you're having huge noise spikes, well, that's explained in the first lines of the FAQ, try fw3203 if you have 3200/3202/3204.

[arturmariojr]

My advice: stay far away from these trash. If you need anything more than a toy with a big screen, don't spend neither 1usd on them.