The thing is that the Siglent SDG1000 is a toy, and while the general hardware quality seems to be ok the design is flawed (jitter issue). It may be cheap but considering that most modes suffer from terrible jitter it still is rather expensive.
What terrible jitter?
It is true that square have jitter due to how just square is produced. But also it is in specs. (0.1% and so on) It is becouse it is derived from sinewave using comparators. (this I do not know why but it is not only Siglent idea.)
If need less timing jitter there can use pulse mode for square wave and it have 8ns peak to peak not cumulative jitter. This is becouse 125MSa/s and there is not littlebit complex methods used for reduce jitter.
This machine works just as it principle is. This is well explained example in some Agilent app note.
This have normal jitter for this kind of equipment and jitter is also normal, not any special terrible if we keep looking this kind of equipments.
It is in my opinion good in its price class if look how it works and what all features it have if look modulations, sweeps, sync, etc. Signal quality is not at all bad in its class.
Of course old middle and high-end or even state of art arb and other generators have lot of better specs. Also I have some these. But if somewhere is terrible jitter it is example HP8161A programmable pulse generator. If people need very low jitter it is good to know what need and do not buy example this Siglent. But also do not buy same class Rigol etc. Of course also Siglent and Owon have lot of better models and more expensive also. Before buy it is good to understand what is buying. Example these jitter read in specs. It is nice Siglent tell peak to peak jitter instead of rms jitter as many do for more nice looking numbers.
But yes. 8ns jitter is of course terrible.
If there is 10 second period 50% square wave (why hell use this mode if need less timing jitter - perhaps user do not know what he is doing)
there may be (typical) 10ms random jitter... 10000000ns! jitter and itv reads in specs, it is designed so. Why user do not use pulse mode and go with around 8ns jitter (exept that with long times there come also system clock "walking" around and timing accuracy is reduced by this. Then can use external good freq reference. )
But then also some manufacturer tell jitter as rms. Some tell peak to peak. This may be big difference. If jitter time is gaussian random it may mean in practice near 10 times difference.
Example Agilent 33220A and square: Jitter (RMS) 1 ns + 100 ppm of period. Terrible or perfect?
How much is peak-peak jitter time in worst case. 1kHz square period time is 1000000ns. 100ppm is 100ns and +1 so specs tell that 101ns jitter RMS. Siglent for Square: Jitter 0.1% of period with 1kHz (typical) and it do not tell if this is peak-peak or RMS but with my measurements it looks more like peak-peak.
For Arb Agilent tell Jitter (RMS) 6 ns + 30 ppm and agen RMS. Siglent for Arb: peak-peak 8ns (typical).
Lets look this:
http://www.sitime.com/support2/documents/AN10007-Jitter-and-measurement.pdf for thinking about rms vs p-p.
Here some tiny tests (designed is more tests but just total lack of time)
http://siglent.freeforums.org/tests-siglent-sdg1000-function-arbitrary-waveform-generator-t5.htmlThis is for random hobby use and for professional use IF user need meet specs. This is NOT high end lab equipment. This is well under 0.5k$ "multifunction" apparatus. Building quality is also good in its class.
(in some old HW there can littlebit improve square wave quality by doing tiny modification as Siglent have told. (Comparators Hys setting resistors))