Siglent SDG1032X mini teardown (hidden self adjust)

[gianluigi]

https://www.eevblog.com/forum/testgear/siglent-sdg1000x-waveform-generators/msg1057306/#msg1057306
If you want in... you already in... ok, with telnet then, I can only replace this with known password.

Which suggests this 30 MHz model might be able to be improved to 60 MHz. 

Who's game to try ?

I'm interested too!

[Johnny B Good]

Hi,

it looks like there is a possibility to solder in a TCXO like on the SDG1025.
This would give the possibility to fine trim the frequency accuracy.

Did someone already try this?

Chris

 I've done this type of upgrade... but not with any Siglent gear. I've upgraded the shitty 50ppm 50MHz smd xo chip used on the main board of an FY6600-60M, first to a 50MHz TCXO board and then to a CQE 10MHz OCXO (as can be found in Symmetricon kit) to drive a 3N502 clock multiplier chip to generate the required 50MHz clock for the FPGA to produce a jitterless clock reference very unlike the one produced by that similarly shite smd xo chip in the SDG1032X (I note the SDG1025 uses a 25MHz XO), that I just took delivery of yesterday afternoon.

 I've attached a short movie which demonstrates this issue quite nicely.  It starts off by showing the SDS2504X Plus being triggered from ch1 (fed from said FY6600-60M's 2nd channel which had to be tuned to 9.999986500MHz as shown in the attached jpg image - the FY6600 is locked to my 10MHz GPSDO reference) with ch2 showing the not very square wave reference output from the SDG1032X - substituting this for ch1 set to 10MHz only changes the wave shape to a sine but not the jittery behaviour which remains the same, making a complete mockery of its claims for low jitter performance 

 Just over halfway through this 20 second's worth of movie, I change the trigger source to ch3 which is showing the RFS, syntonised to the GPSDO courtesy of ch1 of the FY6600 which is outputting a nice stable Sinc pulse at 10MHz (which is a damn sight better than the SDG1032X can do at 5MHz with this flavour of pulse - its shape "squirms and wriggles" quite horribly (the FY6600 shows just a tiny hint of this "squirm" at 10MHz)).

 Now, I recall seeing this jittery behaviour very early on in my ownership of the FY6600 when trying to match frequencies against a bunch of DIP14 XOs, quite possibly before I removed the lump of excrement that Feeltech laughingly refer to as a 50MHz 50ppm smd XO chip and replaced it with an off-board 50MHz TCXO reference. It was rather like chasing Will 'o' the Whisp to get the FY6600 to match the drifty DIP XOs. ISTR that this game became a bit easier after the TCXO upgrade which leaves me guessing that all of these shite 50ppm smd XO chips have this nasty behaviour and what I'm seeing in my nice shiny brand new SDG1032X is simply the consequence of myopic bean counteritus on Siglent's part. It's either that or else my example has a serious fault in its 10MHz reference oscillator department.

 I've just tested a 5MHz and a 16MHz DIP14 XO against the FY6600 (which btw, can be tuned in 1uHz increments over its full frequency range which in spite of Siglent's uHz increments claim, there seems no way to achieve at a frequency setting of 10MHz). Both DIP XOs merely showed the expected thermal drift with absolutely no sign of the jitter being so clearly demonstrated by the SDG1032X' reference output or a 10MHz sine wave from either channel output. Connecting an external 10MHz reference to the SGD1032X immediately eliminates all such jitter from its output.

 Now, I don't expect miracles of stability when running from an internal 10MHz 50ppm smd XO but this weird jitter behaviour is most definitely not acceptable in a signal generator that claims to have only 300ps rms of jitter (not the pk to pk of 300ps of jitter I've observed in the FY6600-60M's sine wave and Sinc pulse outputs, aka ~100ps rms).

 As you can imagine, I'm now rather keen to know whether this jittery behaviour is normal for the SDG1000X series or whether I've just suffered the misfortune of landing up with a "Lemon". IOW, would anyone be willing to test their SDG1000X against a stable frequency reference to see how prevalent, if at all, this jittery 10MHz internal reference issue is?

[EDIT 2020-10-24]

 After some discussion in another topic thread, I eventually concluded that the issue was simply down to a fault in or around the XO chip and returned it for a full refund and purchased from elsewhere, an identical looking SDG2042X to replace it.

 However, that doesn't mean I can retire my much abused (modified) FY6600-60M any time soon - it still has so much to offer (like a frequency setting interface that's a positive joy to use compared to the user hostile interface that all of Siglent's AWG models are currently cursed with).

John

[Johnny B Good]

https://www.eevblog.com/forum/testgear/siglent-sdg1000x-waveform-generators/msg1057306/#msg1057306
If you want in... you already in... ok, with telnet then, I can only replace this with known password.

Which suggests this 30 MHz model might be able to be improved to 60 MHz. 

Who's game to try ?

I'm interested too!

 For those who haven't already discovered it, a less risky hacking method to upgrade to the 1062X was revealed back in January here:-

https://www.eevblog.com/forum/testgear/siglent-sdg1000x-waveform-generators/msg2876058/#msg2876058

Basically you use a dummy firmware update file to gain root access without modifying the existing firmware which is safer than actually downgrading to an earlier firmware version with known root password. You can then follow the instructions to backup and then edit the config files during that session. When done, you simply shut down, pull the pen drive out and reboot into the existing and unaltered firmware version as per any normal reboot. Checking the system info will show the model number changed to SDG1062X and you can dial in 60MHz for sine and square waves.

 Important, follow the thread to get the whole picture and take note of rf-loop's advice to e0ne199. When I was following this breadcrumb trail to upgrade one of these 1032s should I ever finally choose one over the 2042, I gathered quite a few files and studied them until I was reasonably confident I'd be able to follow the correct mix of instructions and not end up in a dead end from which I wouldn't be able to back out of (backing up the key original config files being my insurance against this outcome).

 The need to actually run through this process has gone away but looking at what files I had gathered, I get the distinct feeling I would have to re-study them all over again just to regain the understanding I believe I had attained just a couple of months back.

 Using the dummy firmware update file and keeping backup copies of any files you are going to change should remove the risk of turning it into a 'brick'.

HTH and HAND

John

[velozipapa]

could you distribute this again, i cannot download it anymore

[Johnny B Good]

could you distribute this again, i cannot download it anymore

 The attached zip seems to be what you need but as I said, events made such a mod with the SDG1000 model redundant so I never got to try it out. The instructions are presented in two formats, a text file and a pdf file. The pdf is the easier to follow set of instructions ( they're both exactly the same instructions).

 I can't guarantee that this recipe is the best and safest one to follow but it does seem to avoid the risks associated with using an earlier firmware update version which had a known login user name and password so you might want to seek further advice from others better versed in the subject of hacking Siglent kit before testing the waters with both feet at once.