1. I found a few firmware issues with it, but the price is very compelling, and otherwise it was pretty good.
2. I think the main drawback is the 1GS/s sample rate. Not good enough when two channels are enabled on the 200MHz model.
3. Got to be the cheapest way to get a 500uV/div scope as well.
1. There is many FW issues (also I have found many) but most of these are not severe class. But of course also less severe need fix. Some more severe need repair asap. But not at all this nearly catasrof what was case when SDS2000 was first time launched and there was finaölly so much that it was more wise to not fix and fix over fix, they write new whole revision. I do not believe after looking SDS1000X that this is case. I believe this is level where normal bug fix sequence is enough.
One bug is example in "SPO engine". (example: CH1 on) If set example to 50ns/div (all fast timebases) and measure wfm/s speed. It may show average (average over one TFT update cycle) 61kwfm/s and burst speed over 70kwfm/s. Now, if turn example vertical shift or adjust V/div, wfm/s drops markable. If you return vertical settings and signal etc all are agen same, it do not return to original speed at all. If V/div and vertical shift keep changed or not but cycle CH2 on and off it returns normal speed or cycle CH1 off and back on same result, until you agen touch vertical settings. (speed drop is 10-20% or something like this...depends perhaps combination of other settings)
This is just one example of bug what need immediately repair because reason may be hidden in some important place in SPO engine and
perhaps somehow connected to some other bug (I have least one supect)
2. True. 500MSa/s Nyq. is 250MHz and 250/200 is 1.25.
It is barely enough in theory IF: There is perfect true Sin(x)/x math and there is perfect "brick wall" corner at 200MHz. There is not so much problems with Nyquist Shannon rule 2 due to fact that there is just one ADC for 500MSa/s (not interleaved what may lead to more or less phase errors between true samples). But in practice it is not well enough in practice. Minimum in practicde is somewhere 1.33 - 2.5 For 200MHz sinewave BW 600MSa/s (and- "more better") and for other waveforms if analog front end before ADC BW shape is tight brick wall (what it is not in any mid or low end scopes what I know)
I can tell that Siglent Sinx/x is good and it do not destroy true sampled data but not mathematically ideal perfect. But also it do not make it good for 2 channel full 200MHz.
If look Rigol DS1000Z 100MHz (hacked or original) all 4 channel in use it have 250MSa/s. Nyquist 125MHz.
125/100 is 1.25. Same as in Siglent with all 2 channels in use with 200MHz. But here is very poor fake Sin(x)/x.
It can not say it is good 200MHz if both channels are in use simultaneusly. 1.25 is not enough. 1.35 ... 1.5 and more is of course better. But as all, it also depends many things.
It need always tightly keep in mind that if talk "200 MHz or 20MHz" scope it is stated for pure sinewave.
200MHz: IF take 20MHz ideal square wave (it do not exist) to scope input it have harmonics 60, 100, 140, 180, 220...etc. 9th harmonic is inside and even this is some amount attenuated and 11th go over.
THis is if there is not any limitinmg factor coming from sample rates. When there still exist harmonixs what go near or over nyquist limit and ADC itself see these it start produce aliases. Corners start wobbling etc.
3. It (AFAIK) is true analog side 500uV/div. Example in Rigol DS1000Z most low is 5mV/div. After then down to 1mV/div is digital side multiply... Also Rig DS2000 do not have analog side 500uV/div.
(I have not measured yet Siglent true BW using 500uV/div. If it have true analog 200MHz BW then I can accept its natural noise in this scope class. Also there is amount of "1/f" like noise. If it is BW rejected, then it is bit noisy. But still not true bad, as we understand all realities about noise when analog front end is still quite cheap circuit solution. (good state of art or even high-end analog input stage easy cost over this whole scope price)