LTC2664 Maximum Output Voltage can not be reached?

Dear Sirs!

Currently I’m working on a LTC2664-16 DAC and I’m wondering why I cannot get the maximum output voltage out of the Part. The chip is integrated on a own Board (see SCH). When I try to get the maximum positive output Voltage out of a channel I would assume 10Volts as the datasheet shows. Of course Span and Reference should be configured correctly and headroom from the power supply has to be there.

I get very close to 10Volts but not exactly. On my design I get +9.9833 Volts positive and -9.9841 Volts negative. Where do I lose the 0,0167Volts / 0,0159Volts. There is no current down from the output, the load is a 100k resistor.

Due to the fact that I’m not sure if I made everything all right in my design, I tried trouble shooting the problem and bought the following Boards. Linduino ONE Board in combination with LT3471EDD for power supply (±12Volts) and the demo circuit 2376A-A. The same thing happened on this approach. I cannot get the ±10Volts at the output.

Can you give me some advice? Do I overlook something? Do I have a general mistake in my thoughts?

What else could I check? Any other ideas are welcome.



  • 0
    •  Analog Employees 
    on May 21, 2018 7:32 PM

    A couple questions, can you also provide a layout of your board?  What voltmeter are you using to take this measurement?  Are you using the internal or external reference?  You would get better gain accuracy with an external reference.  

  • Hello Clarence,

    I will try to answer your questions.



    Yes I can provide the layout of the Board (see PCB). In my thoughts I wanted to eliminate my design, by testing the reference Board 2376A.



    To do the voltage measurement I used a 6 ½ Digit HP 34401A (now Keysight). The maximum error should count like this (90-Day accuracy, 10 Volt Range, 10 V Input Level, 23°C ± 5°C):

    Total Error          = ± (% error reading + % of error range) =

    = ± ( 0.0020% * 10V + 0.0005% * 10V) =

    = ± ( 0.0002 V + 0,00005 V) = ± 250 µV = ± 0,0025% of 10V = ± 25ppm of 10V


    With both references the output does not reach the 10Volt.

    What else could cause this beahavior?

    Thanks for your help,


  • 0
    •  Analog Employees 
    on May 21, 2018 10:13 PM

    I did some calculations and the error you are seeing is ~54 counts off.  It is is a huge error for a 16 bit DAC.  Are you certain you are programming at 16 bits?  Can you put a scope on your SPI lines to make sure all of the bits are getting to the part?  

    The gain error I would expect would be 4mV (or 16mV over temperature)  so you are in that ball park, butif you are just at room temp, it shouldn't be 16mV.

    How many parts are have you tested?

  • Hi Clarence,

    Sorry for the delay, I was busy with other things…now to your questions...

    Yes I’m sure writing the 16 Bits, or better the full 32 Bits (don't care, command word, adress word, data word) . The pictures of the SPI-Lines are attached. You can see the previous 32 Bit Input Word too.
    I tested four prototype boards and the Linduino with the Demo Circuit 2376A-A. All of them behave in the same way.

    At the Code 0xFFFF à 9,981 Volt

    At the Code 0x0000 à -9,981 Volt

    At the Code 0x7FFF à -0,6 mVolt

    Are there any other ideas?

    Actually I have to mention that I’m going on holiday from tomorrow night until end of next week.

    But then I will return to our discussion. I’m very thankful about your help!

    Thanks Georg

  • Hello Clarence,

    back again...did you have a chance to look at the SPI-Signals. Is there something wrong? Could you get a Demo Board to check if the Problem can be reproduced?

    Thanks Georg