AD9781 BIST

Hi,

I closed the earlier discussion as i thought the doubt is cleared. But still I am not able to figure out some problems.

I am using offset binary now. I am giving the data vector followed by zeros and some of the results that I am getting are -

Input Output
14'h0001 14'hFFFB
14'h0002 14'hFFF7
14'h0008 14'hFFDF
14'h20FA 14'h7C17
14'h1F05 14'h83EB

In the datasheet it is given that the BIST behaves as simple adder. So when giving this data it should have just returned the same data as result.

But in this case it looks like its just inverting the bits or XOR with 1. The two LSBs are 1 might be because of compatibility with AD9783.

Can you please verify that the output result what I am getting is correct or not ?

Thanks,

Nikhil

  • 0
    •  Analog Employees 
    on Apr 29, 2013 6:38 PM

    I'll have to check this with the designer for the AD9781. This will take a few days because the designer is out of the office this week.

  • Hello,

    I'm working with the AD9783 and I'm also having problems the BIST feature in the AD9783. I have run some test with a vector of 1024 zeros with a bunch of consecutive ones and what I got is clearly different from a simple sum:

    ...00100... =>  0xFFFE

    ...001100... =>  0x0803

    ...0011100... =>  0xEFF8

    ...00111100... =>  0x280F

    ...001111100... =>  0xAFE0

    ...0011111100... =>  0xA83F

    ...00111111100... => 0xA781

    ...001111111100... =>  0xB8FD

    I'm already using the BIST in the AD9736 without any problem, for the AD9736 there is an application note with a matlab implementation of the BIST algorithm. It seems the AD9783 has a similar BIST algorithm but not exactly the same. Could you please give us information regarding the AD9783 BIST algorithm: the equations, matlab code or an application note..

    Thank you very much,

    Nestor

  • 0
    •  Analog Employees 
    on May 6, 2013 6:40 PM

    We no longer  have the software  to check  the way you are using the Bist. I did confirm with the desinger  all the unused bits are clamped at the input word coming into the part so that if they were somehow inputted, they would be ignored. This  information  along with the Bist explination in our data sheet  is all the information we can offer at this time.

  • Hi,

    In the datasheet it is given that it is a simple adder and will give sum of rising edge data and falling edge data as result. So if I am adding any word with zeros it should return the same word only the last two LSBs will be clamped. But as I am not getting expected result how can i verify that data is being latched properly by AD9781 or not ? Is there any other method by which I can check the digital interface of DAC ?

    Thanks and Regards,

    Nikhil

  • 0
    •  Analog Employees 
    on May 21, 2013 1:08 AM

    Nikhil,

    Sorry for the delay getting back to you, I've been overloaded last couple weeks.

       I spoke with the designer and he says you need to have the part in unsigned Binary format  for BIST to work properly. This is done setting bit 7 or register 0x02 to a logic 1. The AD9783,81,80 parts default to bit 7=0 which puts the part into twos compliment. Please ask them to try changing the part to unsigned binary and see if the BIST works properly for them.

    The only other method is to send in a known signal like sinewave single tone and measure DAC output with an analyzer. This takes analog equipment.

                   Tom