External Input Details:I have been testing the analog input used for external modulation input and have figured out its characteristics.
From the Specification Sheet:Input Impedance: 10 k Ohm min
Frequency: 0 to 50 kHz
100% Modulation: 11/12/13 min/nom/max
Based On Tests:A/D range: -6 V to +6V nominal
Bits: 9 (lsb = 12V/512 = 23.4 mV)
Sampling Frequency: 300/512 MHz = 585.9375 kHz
Low Pass Filter: >10 MHz/none
Input to Output Delay: 1 sample (1.707 uSec)
DC offset: -27 mV or 1 lsb on the particular test instrument
Note: Output offset adjustment is after modulation. Setting wave amplitude to 0 and offset to 1 V with modulation on gives a dc 1V output and not 1V modulated.
The test I ran used a second generator to drive the SDG2042X external input and set the SDG2042X to DSB AM with the carrier set to 1 mHz 10 Vpp input 50 Ohm load. Using this setup I could directly look at the external modulation input to CH1 output path with a constant gain factor while the carrier was high and/or low.
When I input sine waves below 250 kHz there was no aliasing. Input frequency of 585938 Hz aliased to almost zero Hz. Given the 300 MHz DDS sub-system clock that probably means the SDG2042X is using that clock divided by 512 for the A/D. I also measured the quantization step time length and that period coincided.
I then input an external square wave to look at the rise time. Basically it jumped up the full step amplitude in 1 and a small fraction sample times hence the conclusion there is no anti-alias low pass filter before the A/D. I also put in 2.92 MHz sine and the aliased output of 9.687 kHz was only half amplitude and I expect an issue with the A/D sampling being not fast enough to see 2.92 MHz rather than an actual low pass filter.
I used the scope in sweep averaging mode with a trigger from the external modulation input. The output showed a full A/D sample frequency of delay followed by a linear slope from the many sweep averages over another sample period. See attached scope plot. So, the net phase delay from input to output is 1.5 A/D samples periods or 2.56 uSec.
Lastly, I wondered about the A/D being 9 bits. I am familiar with 8, 10, 12, 14, and 16 bits converters but not 9. So, I wondered if the Siglent SDG2042X might being using an 8 bit unipolar A/D and adding input analog processing to rectify the input signal to get the 9th bit. Looking carefully at small amplitude (500 mVpp) signals around zero there was clearly a huge flat spot in the output around 0 implying that the input analog rectifier being used to convert the 8 bit unipolar A/D into 8 bit plus sign converter is not the best. In fact, for quality waveforms you really need the external modulation input to be above about 1.5 Vpp. If the slew rate through 0 is fast enough from either higher frequencies or through larger amplitude it seems to work much better. For example, 10 kHz input small amplitude sine looked better than 2 kHz input sine same amplitude.
Given the poor response through 0 on this converter I took at picture to show how bad it can be. 500 mV p-p input sinewave (scope not showing 10x probe atten on sine input) with very low distortion yielding (5V 0-p)/(6 Vdc)*(500 mV) = 417 mV p-p ideally.