Post Go back to editing

AD9523-1 Power-up error

Category: Hardware
Product Number: AD9523-1


We use the AD9523-1 clock generator in our design. But sometimes I get an error. We are using the MSP430 to configure the AD9523-1. After the system is turned on, this error occurs, we read all registers as FF.  When this error comes the REFA clock is ok but the output clocks are missing. I checked  all powers of the AD9523-1 , it's  ok. We used the another card but  this problem occurs again. 

  • Hi,

    I have moved your question to the clock and timing support forum where they can better assist you.  Do you have a schematic drawing of how your AD9523-1 is connected?  Can you share the code on how you are initializing it?  Are you using the no-OS driver from Analog Devices? no-OS/drivers/frequency/ad9523 at master · analogdevicesinc/no-OS (  Do you have any scope shots of the REFA input?

    The more information you can provide the clock and timing team the better they will be able to troubleshoot your problem.



  • Thank you Brandon.

    Hi Gabay,

    please tell me what serial communication the MSP430 uses to manage the AD9523-1.

    Then, let me see if I understand the sequence of the operations:

    - you power up the AD9523-1

    - the MSP430 writes all AD9523-1 registers

    - the MSP430 calibrates the PLL2 VCO

    Then you say, sometimes the controller reads all registers as 0xFF. And this happens on multiple boards. Is this board now being  developed or has it been in production for some time and now this behavior began to appear?

    I cannot think but something is going on with the power supplies of the chip. I suggest you start with replacing the power supplies of the AD9523-1 with voltages being provided from regular dc power supply sources. See if do this if the issue reappears. 

    Please send me the stp file you used to obtain the register values (is created by the AD9523-1 evaluation software) and also the schematic of the AD9523-1.


  • Hi,

    The cards are in mass production stage. We access the chip with SPI. I've attached screenshots of the error and its absence.(CH1-REFA, CH2:REFB , CH3: AD9523 OUT8 ). 

    I measured the inputs with a frequency counter and it was fine. It's not always a problem we face.

  • HI,

    I offered you a debug method. I do not know how to help you.

    From the figures you sent, I see in the first two, a certain phase offset between REFA and REFB. Then, in the 3rd photo, I see a different one. So something is happening on the board for REFA and REFB to change offset.

    Then REFA, which seems to be ac coupled, has an amplitude of around 6Vp-p (if that 1V represents 1V/div), which is greater than the 1.8V, the max voltage a clock can have at the REFx pins. The common mode seems to be enormous, 10V?. Could you please put the scope probe after the ac coupling capacitor, on the side of the AD9523-1?

    REFB seems to be a 3.3V CMOS, which is correct.


  • Hi,

    It's my mistake. I made a measurement mistake. We configure AD9523 with MSP430F using SPI interface. We saw the error at the beginning. Sometimes the controller reads all registers as 0xFF. I attached the related schematic pages.

    AD9523 powers controlled by MSP430F.  For example , we turn off the AD9523 power supply for 10 seconds after that we turn on supplies and configure the AD9523 registers. But we readback the registers its all 0xFF. When we saw the error, we turn off the all systems for 10 seconds and the reading error disappeared.

    We are kindly waiting your return. Thank you.

  • Hi,

    " we turn off the AD9523 power supply for 10 seconds after that we turn on supplies and configure the AD9523 registers. But we readback the registers its all 0xFF."

    The register 0x0000 configures the SPI. What value do you write in it?

    If you need to use a non zero value for this register (meaning you use the 4-wire SPI), you need to write it first. Then read it back to make sure the value has been captured correctly.

    if the value is not read correctly, you may have accessed the AD9523-1 too early and you need to try again. Then you can write all the other registers.

    If the register 0x0000 must be left at the default value (meaning you use 3-wire SPI), read a register that has a non zero default value, like register 0x018, which must have a default value 0xC. Only after you read this value correctly, meaning the AD9523-1 powered up correctly and it's ready to be configured, you start writing all the the registers.

    Then you need to execute the IO Update instruction in order for the buffered values to move into the live registers. Then you can read the registers back and verify the desired values have been captured.


  • Hi,

    First of all , we write 0x95 and 0xAD  values into 0x05 and 0x06 registers. And readback these registers to check SPI is ok or not. If there is a error , we don't continue the initializing the device

    We configure the AD9523 at the beginning according to datasheet. But when this error occurred the chip stuck at a condition like SPI writing/reading error. Do you have any suggest to us ?

  • HI,

    I already recommended to read back the AD9523-1 registers to see they were captured by the chip correctly. Did you do it?