Duy,
[quote user="Huynh Anh Duy Nguyen"]1) The input pins (J2A) and switch S1:[/quote]
If you want to control LDAC through LDAC2 you must switch JP1 to position 1-2. Currently it is being controlled by LDAC1.
If you want LDAC to be grounded at all times, allowing for transparent DAC updates, S1.1 must be a 1. This will connect LDAC to GND.
S1.2 is UNI/BIPB.
S1.3 is /RST, which is currently HIGH. Allowing the device to operate normally.
S1.4 is UNI/BIPA.
[quote user="Huynh Anh Duy Nguyen"]1) How can we know the shift register get the sequence of pulse correctly??? How can we debug this issue???[/quote]
You can readback the DAC registers as explained in page 24 of the DA7816 datasheet.
The readback command is 1000 ADDR XXXX XXXX XXXX XXXX XXXX XXXX XXXX XXXX
For VOUT-0 ADDR = 0100
I took a plot for your reference.
[quote user="Huynh Anh Duy Nguyen"]2) When I give the supply to the input SDI, I observed the SDO pin on oscilloscope, the pulse is similar but lag 1 or 2 bits (means SDI: 0000 0100 0111 1111 1111 0000 ->> SDO: 0000 0010 0011 1111 1111 1000)? Is it normal???[/quote]
No, I don't think this shift is normal. As you can see on the plot above, SDO updates during the rising edge of SCLK and is valid during the falling edge of SCLK. A plot of this behavior may help me figure out what is going on.
[quote user="Huynh Anh Duy Nguyen"]3) In page 22 of datasheet, the Power-Supply Sequencing; is it important for a simple conversion???[/quote]
Yes, this sequencing is important to the start-up of the device. The device may latch the wrong values at start-up if the sequence is not used. Keep in mind that there are product variations. Some devices may miss-behave while others might not.
[quote user="Huynh Anh Duy Nguyen"]4) Do we need to set up anything about Zero or Gain calibration???[/quote]
The zero and gain calibration are optional. The zero and gain calibrations are in place in order to reduce the offset and gain errors that change from one device to another. The maximum offset and gain errors before user calibration are specified to be ±0.5LSB. If you wanted to calibrate the device to an even smaller error, you would have to measure the offset and gain errors of each device and program the device's registers accordingly. The easiest way to do this is using a two point line. Take two points, (usually from 1/4 and 3/4 of full scale) and find the equation of the line( y = m*x+b) m being the slope (gain) and b being the offset.
If you need me to expand more on any of my answers, please feel free to ask.