As a reaction to
@jeraymond's finding on the SDG2000X and the obvious similarities between the instruments, I tried to verify the mentioned problems on my "6000". And here's what I found (both channels set to 10MHz Sine wave, CH1 "left alone" and connected to a scope, external REF supplied and connected to the triggered channel of a scope -- but the AWG shows identical bahaviour if the internal reference is used --, CH2 Burst Mode enabled/disabled and burst parameters changed for testing):
The SDG6000X in "Phase Mode: Locked" performs even worse than the '2000X in this regards: It doesn't matter if Burst is enabled or not, as soon as certain burst periods are entered, the frequency of the "unaffected" channel (10MHz sine) changes by a maximum (as yet figured out at a burst period of 99.999991ms ) of ~0.8Hz. What makes it even more funny: If the burst period is set to 100ms via the keypad, some frequency drift is present (~0.2Hz deviation). If I now change the period via the encoder by 1ns down and up again (so the diplay shows 100ms burst period just like before), the frequency drift is gone! I've got to increase the period by another 1ns to make it reappear. Seems like the internal resolution of the burst period register / math engine is less accurate than the displayed value (1ns).
If "Phase Mode: Independent" is selected, the situation is much better -- means: No frequency deviation whatsoever if burst mode is selected or changed for the "other" channel, but still, when pressing the the "Burst" menu button, the "should-be" unaffected channel stops shortly and restarts at a new phase angle vs. the reference clock. Siglent, WHY?
I probably don't need to tell, my current "go-to" generator for stuff that wouldn't require the high frequency performance or high(er) frequency level accuracy of the SDG6000X, Rigol's dg811(+++++++)
, shows none of this "magic". It works like two separate AWGs, which is what the waveform engines in the FPGA probably are...