The Siglent SDG2042X Thread

[tautech]

Well, my original supplier (Elektor) simply isn't answering my request, so I contacted a Dutch Siglent dealer, and they are now trying to get a quote for a new CPU board from Siglent.  But they're working on it. 

If needed you can contact Siglent in Hamburg directly, explain Elektor has not been timely in offering support and/or point them to EEVblog (thread or post) to seek assistance.
As I've mentioned member Lightages had little trouble sourcing a replacement PCB for his SDG1k series AWG.
http://www.siglenteu.com/lxwms.aspx?id=1110

But apparently, a request like mine happens very rarely, and so getting Siglent to fish a replacement board out of their logistics chain and put a price on it is going to take some time.

The only thing I can do now is be patient.

[Timpert]

Just today I got the message that Siglent will repair the generator for me, I have to send it to them. The price is very reasonable, and I want a working generator, so I'll do it. Kudos go to eleshop.nl for picking up the after sales service!

[McBryce]

Hi all,
     finally got some time to properly try out most of the features of my new SDG2042X last night and came across something strange:

If I choose a square wave at 25MHz I get a 25MHz wave with a rise time of 8.2ns (pretty much what the datasheet implies - 9ns). However, if I now press the sweep button and leave the parameters at "sweep start" 25MHz and "sweep end" at 25MHz, the unit is now outputing a much nicer 25MHz square wave with a 3.0ns rise time??

Can anyone else confirm this on their unit?

McBryce.

[alank2]

I can confirm this - and I have pointed it out to Siglent.

[McBryce]

So it's a bug that improves performance?? So why can't it just do square waves with a rise time of 3ns all of the time?

McBryce.

[alank2]

So it's a bug that improves performance?? So why can't it just do square waves with a rise time of 3ns all of the time?
McBryce.

Apparently - I assume there is an issue where the IC used to generate the waveform simply performs a little better when in "sweep" mode.

[PartialDischarge]

So it's a bug that improves performance?? So why can't it just do square waves with a rise time of 3ns all of the time?
McBryce.

Apparently - I assume there is an issue where the IC used to generate the waveform simply performs a little better when in "sweep" mode.

Not sure about that. The spec of 8.4ns applies for Arb mode. In DDS mode I can get rise times of 4.5ns, maybe the sweep mode uses this mode...

[McBryce]

Exactly. So why aren't they using the DDS mode to create the square wave signals all of the time?

McBryce.

[tautech]

Exactly. So why aren't they using the DDS mode to create the square wave signals all of the time?

McBryce.

IMO the answer is signal fidelity.
We all know the faster you drive signals the harder it is to maintain good fidelity.
If it was not a SG it would not matter too much about signal shape.

Re-read the website claims and tell me I'm wrong.
http://www.siglentamerica.com/pdxx.aspx?id=1364&T=2&tid=16

[jjoonathan]

Here is the superposition of 20,21,22,23,24,25MHz square waves with SWP=off (yellow) and SWP=on (pink).

https://postimg.org/image/3y5qq3kef/

Evidently in the current implementation you can have sharp edges or the much-ballyhooed smooth control over width, but not both.

EDIT: the "quantum" of distance between the purple edges is t?3.38ns, 1/t?300MHz.

EDIT2: I see some ripple on the top of the square wave when I make fine adjustments to the frequency with SWP=off. I also see smoother width increments than I would expect from the 1.2GS/s DAC frequency. Seems they've implemented their constant-width square waves with a digital filter that trades bandwidth for slope positioning rather than by playing with the DAC clock. It also seems that for SWP=on they use their usual arb path rather than a counter that bypasses the interpolater. Ah well.

[pascal_sweden]

Was that measurement made on a Rigol DS4000 series?

The menu structure looks nicer than the Rigol DS1000Z series

[jjoonathan]

Yep, it's an old (dec 2011) DS4014 hacked up with decodes and 500MHz. I do like the menus and other creature comforts it has over the DS1054Z I got started with (slightly faster UI compositing, dedicated menu hide & scroll buttons on both sides, separate channel offset/scale knobs, auto-scaling probes, 50ohm/ac/bw lights, segmented memory, 50% trig button) but unfortunately the firmware is notably crustier. For the longest time, decodes didn't work in segmented memory, now they *only* work in segmented memory. Older versions had freezing issues, newer versions have herky-jerky scrolling at 1ns/div scale. If I had payed Rigol's asking price for any of the extras, I'd be pissed: Agilent price at a fraction of Agilent quality. Unhacked, it's not really competitive in its price bracket, especially with Agisight getting serious about discount refurbs. My guess is that Rigol wrote it off and is focusing on their next iteration of professional scopes, only keeping it in their catalog so that they can claim experience in the price bracket while selling remaining inventory to the captive audience of the Chinese defense industry.

Hacked, it's pretty good, at least until the fine fellows over in the DSOX thread figure out how to hack a DSOX3014 to a DSOX3104 

EDIT: oh, and ADC3 (pink) is dying -- rail-to-rail noise after a cold startup that goes away as it heats up. I'm not sure whether it's due to a bad bond wire, 3 years of the previous owner running the poor ADCs at 85C without heatsinks, or both, but Rigol wants $1k for the repair. That's fine, I got it for $1k off the sticker price on ebay, and I might even be able to make do with a $130 replacement ADC I got off taobao. It's been an adventure, and that suits me just fine, but it's not what most people are looking for in a professional scope. Look out for a post when ADC3 finally kicks the bucket and I make the swap; regardless of the outcome it's bound to be entertaining.

[alank2]

Rigol once told me that their out of warranty repair was:

>Our out-of-warranty repair prices are quite reasonable. It is $250 or 10% of the purchase price, whichever is greater..
>plus shipping to/from our repair office in Cleveland, OH.

[gby]

Hi all,
     finally got some time to properly try out most of the features of my new SDG2042X last night and came across something strange:

If I choose a square wave at 25MHz I get a 25MHz wave with a rise time of 8.2ns (pretty much what the datasheet implies - 9ns). However, if I now press the sweep button and leave the parameters at "sweep start" 25MHz and "sweep end" at 25MHz, the unit is now outputing a much nicer 25MHz square wave with a 3.0ns rise time??

Can anyone else confirm this on their unit?

McBryce.

This is a known quirk of this signal generator.   Please my post #94 from January this year that explains how this difference in square wave rise time occurs.
https://www.eevblog.com/forum/testgear/the-siglent-sdg2042x-thread/msg834806/#msg834806

The real question is what is the logic behind Siglent's choices for when the Analog Devices D/A internal interpolator is turned on (lower rise times) and when it is turned off (output steps to new sample for faster rise time)??  As I noted in my earlier post it might be nice if Siglent added to the FW to be able to turn the D/A chip interpolator on/off.

[jjoonathan]

Rigol once told me that their out of warranty repair was:

>Our out-of-warranty repair prices are quite reasonable. It is $250 or 10% of the purchase price, whichever is greater..
>plus shipping to/from our repair office in Cleveland, OH.

That's what I thought. I even mentioned that tidbit to them, although at the time I was writing the ticket I thought I was remembering the $250 figure from a webpage somewhere rather than from a post of yours.

I have an old DS4014, serial DS4A1350----, purchased used from testbox, which I suspect is out of warranty. Channel 3 has rail-to-rail noise for 10-60 seconds after a cold startup. It isn't actually dead but I anticipate it dying soon and want to figure out what my options are so that I can plan ahead. From what I can find on your warranty page, this out-of-warranty repair would cost $250+parts+shipping, is that correct?

Notice how I sneak in "+parts". $250+shipping is around the cost of a replacement ADC but with less risk and is therefore preferable. $250+parts+shipping is not. I'm trying to force the question.

Hello Jon,

My name is ------------ and I am an Applications Engineer at RIGOL Technologies USA. Thank you for writing in.

Yes. That is out-of-warranty.

It was manufactured in 2011. The warranty likely expired in 2014 or 2015. Without a receipt or invoice, I'd have to go off the serial number for the active warranty dates.

Out-of-warranty repair for the DS4014 is $920 plus shipping.

If you would like to continue with the repair, please provide the following information:

- Return shipping address

- Phone number

We also have a clearance section on our website. This area has demo and refurbished equipment that come with 90 day warranties.

https://www.rigolna.com/clearance/

J

I especially like the last bit. "Yup, your device's knockoff chips are dying from years of suffering at 85C under our crappy thermal design, we want to charge you almost as much as the device costs for a repair, would you please consider buying a new one with 1/12th our standard warranty?"

 

In your dreams, Rigol. In your dreams.

Thanks for the prompt reply.

Ouch, $920 is steep, but at least it makes the repair/replace decision easy. And frees me to take a swing at the issue myself. Any chance you guys would be willing to sell me a MXT2001 ADC? I'm not going to be in China for the next 5 months or so and they don't stock them on DigiKey.

Cheers,
Jon

No prompt reply (or any reply) this time, of course. Hardly surprising, but they had a second chance to impress me ($250/10% out-of-warranty repairs would have been impressive in my book) and they did not.

[rf-loop]

Topic: The Siglent SDG2042X Thread

[pherdie]

Old timer getting back into electronics after 20 years, so please be patient.

I have a 2042X and I have a PE4302 attenuator board from China. I am trying to test drive the serial input with the 2042X and learn a little about both. I am aware the parallel input can also be used to input.

A clocked 10MHZ and a 10MHZ six bit word is required to set the various attenuation levels. I want to be able to generate all six bit binary input permutations required to fully test the device. The best I have done is generate a 10MHZ clock on channel 1 and a six bit 10MHZ pulse train, manually triggered, synced to channel 1 and of sufficient duty cycle to insure a good write when clocked. There appears to be no way to easily modify the individual data bits. Did I miss something here? Is there another method with the 2042x? Does the supplied software have a methodology (Mac here, no Windows and I do know Windows can run on a Mac). It would seem to me (for what it's worth) that creating a data pulse train would be just an ideal use for this wonderful device. I have to be missing something.

I will probably later drive it with something like an Arduino.

[tautech]

Old timer getting back into electronics after 20 years, so please be patient.

I have a 2042X and I have a PE4302 attenuator board from China. I am trying to test drive the serial input with the 2042X and learn a little about both. I am aware the parallel input can also be used to input.

A clocked 10MHZ and a 10MHZ six bit word is required to set the various attenuation levels. I want to be able to generate all six bit binary input permutations required to fully test the device. The best I have done is generate a 10MHZ clock on channel 1 and a six bit 10MHZ pulse train, manually triggered, synced to channel 1 and of sufficient duty cycle to insure a good write when clocked. There appears to be no way to easily modify the individual data bits. Did I miss something here? Is there another method with the 2042x? Does the supplied software have a methodology (Mac here, no Windows and I do know Windows can run on a Mac). It would seem to me (for what it's worth) that creating a data pulse train would be just an ideal use for this wonderful device. I have to be missing something.

I will probably later drive it with something like an Arduino.

Here's some threads with varying clues for you to proceed with:
https://www.eevblog.com/forum/testgear/video-review-comparison-and-teardown-of-siglent-sdg5162-rigol-dg4162-arbs/
https://www.eevblog.com/forum/testgear/digital-signal-generators/
https://www.eevblog.com/forum/testgear/siglent-sdg1025-matlab-programming-arb-waveform/

Hope those help.

[tautech]

An FFT of a SDG2042X @ 40 MHz FYI

Image borrowed from this thread:
https://www.eevblog.com/forum/testgear/siglent-ssa3000x-spectrum-analyzers/msg989007/#msg989007



Pretty good internal reference IMO. 

[rf-loop]

An FFT of a SDG2042X @ 40 MHz FYI

Image borrowed from this thread:
https://www.eevblog.com/forum/testgear/siglent-ssa3000x-spectrum-analyzers/msg989007/#msg989007



Pretty good internal reference IMO. 

It can think true freq is nearly sure inside  40MHz +/- 40Hz. If image was taken after SSA 30 minute temp stabilization with autocal enabled.

[tautech]

An FFT of a SDG2042X @ 40 MHz FYI

Image borrowed from this thread:
https://www.eevblog.com/forum/testgear/siglent-ssa3000x-spectrum-analyzers/msg989007/#msg989007



Pretty good internal reference IMO. 

It can think true freq is nearly sure inside  40MHz +/- 40Hz. If image was taken after SSA 30 minute temp stabilization with autocal enabled.

It certainly was. 

[PartialDischarge]

I did measure the same, SDG2042x outputting a 40mhz sine using my CTRU which has the OCXO option (0.005ppm drift, last calibrated 2 years ago).....started measuring +3hz and ended at -2hz after some time...

[janekivi]

I was turning too all the buttons one day

[CustomEngineerer]

Did it actually output at that frequency or did the setting just show that it was set?

[janekivi]

It output what you limit in one xml there but let's say, their limits is set fear enough.
It won't output something normal over the limits.