Siglent SDG1032X Harmonic Distorsion

[mawyatt]

Now it looks better.

But as previously told first image can not accumulate any jitter related to waiting time. It can not see in image but I assume you have again used Trigger Holdoff and of course this Holdoff time do not accumulate any jitter due to zero delay when next trigger event after Holdoff time elapsed.  Result looks same if Holdoff or not, exept delay next displayed acquisition.

If you recall, the original intent of these screen captures was to show that the DSO was not introducing significant jitter because the Holdoff is retriggered and the Zoom is not.

DPO zoom have nearly nothing to do with analog scope zoom even when result may looks like somehow same kind of image.
This is also small Achilles heel in conventional DSO and more modern DPO (or SPO like Siglent name it) but mostly there is way to walk over this. But dual independent beam dual time base dual delayed trig  -  just forget with these digitals.
This, and most other simple ones,  DPO can only do one whole sweep based to one trigger engine trig and nothing else.
For Zoom window we take only more or less long part from this one acquisition and from time position how user set zoom window position. There can not generate example other delayed trig for zoomed detail. There is not second "Timebase" where scope sweep part of trace more fast or do sequentially slower sweep and then for zoom faster sweep for adjusted position. No, DPO do not at all have other than just one "timebase" so it is even bit wrong to name horizontal axis adjustment as timebase. It is only time scale. With Zoom there is just this same original one "sweep" aka acquisition in memory. When we zoom, we only take part of this acquisition in memory and show it using different time scale on display.  And in this case now here, time jitter come only visible using this zoom so that we look other memory position than trigger position. There we can see if signal detail exist sooner or later. After then we know that oscilloscope timebase have jitter or signal under test have time jitter or both have time jitter. For know more we need know scope timebase jitter. If we know it is far less than signal under test we can name this observed jitter is signal under test jitter + error from scope timebase jitter. If we Know signal jitter is example "zero" after then we can tell that jitter we have seen is oscilloscope own jitter in timing.

The core fundamental difference between the analog scope and the DSO regarding triggering is, the DSO must decide the trigger after the signal has been captured by the ADC, the analog scope can trigger essentially instantly on the signal and in the case of the dual timebase can have two entirely different timebase sweeps with continuous resolution between them, whereas the DSO is limited by the prior sampled signal by the ADC. So in effect the ADC clock becomes the controlling single timebase in the DSO, and the memory depth becomes the limiting factor between using delayed functions like Zoom to try and emulate the analog dual timebase. The analog scope has limits also since with very large differences in the dual timebases the CRT update rate limits one's ability to "see" the signal, and one reason (the main reason was to capture single or infrequent events like "glitches") why Tek developed the long persistence phosphor CRTs and later the special Micro-Channel Plate CRT.

In your images (zoomed ones) can see quite small time jitter.

Yes the jitter is very small indeed, so I can feel confident when observing jitter from other sources

Now small problem there is that slope is very slow for this purpose.
Decimated samplerate is 100MSa/s
It mean that there is decimated sample interval 10ns.  In Zoomed window there is only 5 samples in one acquisition and rest visible things are interpolation. Due to lack of knowledge about just this Siglent model I can not estimate what is perhaps possible trigger/fine interpolation/display positioning jitter amount and what is timebase related jitter and signal jitter. I believe trigger engine just after ADC use full non decimated samplerate but this is only now hope and believe.  But so or so, it can say --  scope timebase jitter is small!  Better than I expect. And it is good to know due to fact it do not have ExtRef input.

Yes the jitter is very small indeed, so I can feel confident when observing jitter from other sources

The sample rate and memory depth place limits on these, so higher sample rates require deeper memory or faster main timebase. I think this SDS2102X Plus DSO achieves a good compromise considering the cost.

Maybe Siglent will consider an ext ref for this level scope in the future, and the ability to tune the internal reference with a DAC like the SDG2042X AWG. These additions shouldn't cost much to implement.

Oh well, example Hethkit IO-12 and perhaps even one bit older.. is still deeply "burned" in my memory, I can even remember smell of it and my first old Collins Rx from 40's.

Have fond memories of that period of time including my Heathkit VTVM. At 8 years old I knew what I wanted to be, an EE, and have had a fun career which is winding down now, faster than I would like tho

Best,

[rf-loop]

Just for fun.

In this thread was FeelTech FY6600 told and also some words about more new FY6900.

I was bit interest to see what it is.This is not after DIY modification project. This is out from box, as it is. I like instruments are ok out from box. I have other things to do than playing with DIY boxes.

Here first old image about SDS1032X. CH1 10MHz sine CW out.
Also SDG1032X is some off from 10MHz but I remember it is much better than specifified. But in this image I believe it is connected to ext reference from SA.  This is not so important here.
Yes in home I have LOT of better images and data.


But then this FY6900

I have not often seen this dirty output.
Totally shit. And I mean this AM modulation what must not be there at all or so small that with this system I use here can not detect it. But this modulation is high... it is totally so bad that I try find if there is some external reason what disturb now things but with many kind of cross check my conclusion is that it come from FY without any doubt. I have also tried full isolation (and signal from antenna to antenna... just same result. )

Other thing is that it is out from box initially 30ppm off. Specs max limit is 20ppm. In one hour it drift roughly 5ppm.

These freq accuracy things are not so severe but this AM modulation is terrible. And of course it continue more wide that just this 1kHz span... bit less amplitude but 5kHz offset and still high. What a fuck they have done or forget to do.

But it looks that it have roughly 50ohm internal output resistance (yes I say output resistance and not impedance because I do not know enough due to total lack of test instruments in my situation just now). But something weird there is in output amplifier... when it drive some load it is perhaps warming and level start drift down..
Still if look price, under 80 Euro... even there is missing lot of features and performance what example SDS1032X have it still can do some things quite well. This frequency full resolution adjust I like.

But this really dirty sine CW.... bhuuhhh...  for what it can use?
Who knows if there is some simple trick inside what cause this and is easy repairable.


1.9Hz RBW   as can see 30ppm off (304Hz)

[mawyatt]

This looks horrible 

Is the Feeltech AWG the one that uses an R2R ladder network that's directly driven from the FPGA outputs, rather than using a proper DAC?

Quality high resolution, high frequency DAC's are costly, but the use of the mentioned design practice to replace the DAC would help explain this. The FPGA supply noise levels and non-linearity introduced by using the direct FPGA drive for the R2R network are going to make creating a quality waveform almost impossible 

Best, 

[rf-loop]

This looks horrible 

Is the Feeltech AWG the one that uses an R2R ladder network that's directly driven from the FPGA outputs, rather than using a proper DAC?

Quality high resolution, high frequency DAC's are costly, but the use of the mentioned design practice to replace the DAC would help explain this. The FPGA supply noise levels and non-linearity introduced by using the direct FPGA drive for the R2R network are going to make creating a quality waveform almost impossible 

Best,

Afaik 2x 14 bit DAC.  I believe FeelElec is just overall this bit "better" manufacturer.
What I have here is just for fun because it was so dirty cheap and I was bit interesting to look it after I have read many peoples talk about these 6600, 6800 and most new 6900 what may have some advantages over these others/olders. Model is FY6900-40M  FW V 1.3.1  Later some day when I carry it to my homeland I will look more carefully and deeply some "performance" things. I think there is something bad in internal power supply and power sharing. Perhaps HW design error or least poor desing...

[radiolistener]

PSG9080 10 MHz sine:

[Fretec]

Wow, that FY6900 output signal is really very bad.

[Electro Fan]

Is it the conclusion of this thread that a 1032x has too much/unacceptable jitter and that the next solution up to bat is a 2042x, or is there some other conclusion/recommendation (maybe a Rigol DG1022Z generator, or something else) in the $300 - $500 range? Or is this thread possibly a bit tough on the the 1032x ok (ie, it's "fit enough for use" in the ~$300 neighborhood)?

[rf-loop]

Is it the conclusion of this thread that a 1032x has too much/unacceptable jitter and that the next solution up to bat is a 2042x, or is there some other conclusion/recommendation (maybe a Rigol generator, or something else) in the $300 - $500 range? Or is this thread possibly a bit tough on the the 1032x ok (ie, it's "fit enough for use" in the ~$300 neighborhood)?

This have so many times handled that I do not explain so much images exept if someone ask some detail.
Of course if we compare clock jitter to other clock what is DIY assembled example in some Feeltech generator we can say Siglent internal reference have more jitter.
Out from box example FY6900 freq accuracy is with one word:Horrible and sinewave is so dirty that it can not use for any kind of serious work as also shown here previously. Of course if someone do same DIY work for Siglent he can also install there even best possible high end DOCXO. But when need better frequency accuracy and stability it have also input for external 10MHZ reference.


Square cycle to cycle jitter. Note frequency, it is selected so that it is not "golden frequency" for lowest jitter.
Here in this image jitter is roughly 170ps Peak (340ps Peak to Peak) specs limit is 300ps RMS what is far higher what is peak depending jitter ditribution and how much data (how long time) we collect.
Previous rising edge (trigger) is left side out from image.


Square cycle to cycle + width jitter. Note frequency, it is selected so that it is not golden frequency for lowes jitter.


DG1032Z


SDG1032X










https://siglent.fi/pic/SDG1000X/SDG1000X-1Hz-Pulse-duty0-001-fall-zoom.png
Pulse width jitter. 1s period pulse. Pulse width 10000ns falling edge slope zoomed. Infinite persistence. Jitter roughly in 200ps peak to peak. 10/90 Ft 2.9ns.




5.1kHz  Note. 1st harmonic (fundamental) level is higher. So level distange to 2ns harmonic is not accurate and perhaps more than displayed 66dBc. This is due to fact that 5.1kHz is out from SSA frequency range and it start attenuate more and more when go under specified 9kHz.


3.7MHz


30MHz



In this last image pulse period is continuously 1s
All time scope have been with infinite persistence.
First there have been 50ns pulse. Sfter bthen pulse width is adjusted slowly with very small increments to 100ns. After then adjusted for 300ns. After then 200ns width and then started to ajust slowly with very small increments falling edge slope to 150ns fall time.



In this image continuous 1kHz freq and 100ns width. After then width have adjusted 8 steps using 1ns steps and finally 10 steps using 0.1ns steps.
There can see small jitter, roughly p-p 200ps.
If look shape in zoom window top and bottom there can see this "EasyPulse" technology produced some kind of risetimejitter or wobbling or undulating how it need name I do not know. But partially reason is also when I "wrong use" oscilloscope. Analog input stage is highly overdriven because I want better image for show these pulse width steps and if whole signal top and bottom is in image this falling edge slope looks very different in this time scale 2ns/div. This is for show these steps, not for show whole edge shape.

Continue in next msg due to forum image limits

[rf-loop]

Continue...



I have lot of more tests but these tests data are not with me here far from my homeland.

And example if use for some under 30 or 60MHz works with some radio things, example for filters adjustment or other things. Siglent AM modulation is made like real RF generators.  And you can even use two independent channel AM with independent modulations. Also same for dual sideband DSB. Sad it do not have single sideband SSB.

In SDG1000X channels can also combine.

scope Ch2 is 400kHz sine out from generator Ch2. 400kHz sine is also AM modulated with some audio as can see level change due to persistence. It is trigged to this 400kHz HF.
Generator CH1 have other higher frequency with lower level
Oscilloscope Ch1 is coming from generator CH1 what is inside generator combined CH1 and CH2 signal. Freq. are selected so that no sync what can also seen due to scope persistence.

But, of course all depends what are user needs related to jitter and other things.
SDG2000X series internal reference is better than SDG1000X.


Do this tell enormous jitter. Imho, not.

[Johnny B Good]

Is it the conclusion of this thread that a 1032x has too much/unacceptable jitter and that the next solution up to bat is a 2042x, or is there some other conclusion/recommendation (maybe a Rigol DG1022Z generator, or something else) in the $300 - $500 range? Or is this thread possibly a bit tough on the the 1032x ok (ie, it's "fit enough for use" in the ~$300 neighborhood)?

 Since it was me that had brought up the issue of 'jitter' with a recently purchased SDG1032X which I've now returned. I can tell you that I finally concluded that the problem had been down to either a faulty XO chip or a fault in the supporting circuitry or just a dry joint, broken capacitor or whatever.

 Aside from that, courtesy of the use of an external 10MHz reference clock, there doesn't seem to any jitter issue other than what is entirely expected of DDS and sharp edged arbitrary waves. This model uses a special circuit for square waves to eliminate the DAC clock jitter issue that would otherwise afflict them in cheaper designs such as those Feeltech/FeelElec FY66/68/6900 models.

 If jitter is your only concern, then worry not. 

John