Hello ADI Team,
I am currently porting the AD-FMCDAQ2-EBZ reference design from the ZCU102 platform to an ALINX AXU15EG (ZU15EG) evaluation board. Since the AXU15EG is not officially supported, I used the ZCU102 design as a reference.
Hardware
-
Carrier board: ALINX AXU15EG (XCZU15EG)
-
FMC board: AD-FMCDAQ2-EBZ
-
Vivado: 2023.2
-
Vitis: 2021.2
-
Software: ADI no-OS (bare-metal) 2023_r2 Release
Step 1 – Vivado Hardware Porting
I first created a Vivado 2023.2 project by starting from the ZCU102 DAQ2 reference design.
The modifications included:
-
Replacing the ZCU102 board with the AXU15EG device.
-
Updating all board constraints (.xdc) according to the AXU15EG FMC HPC connector.
-
Verifying:
-
JESD204 transceiver lane mapping
-
Differential clock pins
-
SYSREF
-
SYNC signals
-
SPI signals
-
GPIO assignments
-
-
The design was implemented successfully and the bitstream was generated without critical errors.
Step 2 – Bare-Metal Software
I then created a Vitis bare-metal application using the ADI no-OS framework.
The software is based on the standard FMCDAQ2 example application without major modifications except for the platform-specific BSP.
The application starts correctly until the initialization of the AD9523 clock generator.
The console output reports:
AD9523: SPI write-verify failed (0x00FF)!
The failure occurs during ad9523_setup().

Debug Information
The SPI initialization completes successfully.
The failure happens when the driver performs a write followed by a read-back verification.
The read-back value is:
0x00FF
instead of the expected value.
From debugging, the failure occurs inside:
ad9523_setup()
↓
ad9523_spi_write()
↓
ad9523_spi_read()
↓
SPI write-verify failed

What I Have Already Checked
-
SPI pin assignments match the AXU15EG schematic.
-
SPI clock, MOSI, MISO and CS are connected correctly.
-
FMC connector routing has been verified.
-
The FPGA bitstream programs correctly.
My Questions
- Has anyone successfully ported the FMCDAQ2 reference design to a non-supported Zynq UltraScale+ board?
- What could cause the AD9523 SPI read-back to always return 0x00FF?
- Are there any recommended debug steps to isolate whether the problem is hardware, HDL, or software?
- Any suggestions would be greatly appreciated.
Thank you.