This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DAC81408: Internal voltage reference deviates more than specified

Part Number: DAC81408

Hi,

I have a question regarding the accuracy of the DAC81408RHA's internal voltage reference.

According to the datasheet the initial accuracy should be +/- 2.5mV with an ambient temperature of 25°C. However, I have measured the VREFOUT on a few devices and got results ranging from +2.496765V (-3.235mV) to +2.494524V (-5.476mV) at room temperatue. Measurements were done with a calibrated benchtop multimeter, Agilent 34470A).

Even when I take into account the output drift (over time) and typical thermal hysteresis, I don't expect to get these big deviations. 

The VAA /VDD supply voltage is +4.9865V and stable, no noticeable ripple. The REF and REFCMP pins are decoupled as per datasheet, 150nF and 330pF, respectively. 

Have these big voltage reference deviations been seen before? What could cause this big deviations? 

Thanks,
Jeroen

  • Hi Jeroen,

    Could you share the DAC portion schematic for our review? Is there anything else connected to the reference pins?

    As an extra check, does another meter correlate this measurement? 

    You mentioned two systems have a lower than expected reference - how many units have you looked at? 

    Thanks,

    Paul

  • Hi Paul,

    Thank you for your quick response. 

    Please find the schematic below. Note that C124 and U16 are unplaced, so all what is connected to REF is a 150nF cap and a 2 pins header.

    Measurements with another, less accurate meter (34401A) show slightly better results, but still bad. Please see the results below of the 3 DAC devices on my board. I have 4 other boards that I will look at later this week.

    Regards,
    Jeroen

    * DAC1

    - Internal Voltage Reference: +2.496765V (34470A 7.5 digits DMM) => Error = -3.235 mV

    - Internal Voltage Reference: +2.49700V (34401A 6.5 digits DMM) => Error = -3.000 mV

    - Internal Voltage Reference: +2.49704V (34401A 6.5 digits DMM) => Error = -2.960 mV (after warming up)

    * DAC2

    - Internal Voltage Reference: +2.494524V (34470A 7.5 digits DMM) => Error = -5.476 mV

    - Internal Voltage Reference: +2.49482V (34401A 6.5 digits DMM) => Error = -5.180 mV 

    - Internal Voltage Reference: +2.49486V (34401A 6.5 digits DMM) => Error = -5.140 mV (after warming up)

    * DAC3

    - Internal Voltage Reference: +2.496315V (34470A 7.5 digits DMM) => Error = -3.685 mV

    - Internal Voltage Reference: +2.49642V (34401A 6.5 digits DMM) => Error = -3.580 mV

    - Internal Voltage Reference: +2.49643V (34401A 6.5 digits DMM) => Error = -3.570 mV (after warming up)

    Thank you for your quick response
  • Hi,

    I have couple questions regarding measurements.

    1. whats the impedance on the meter set to?

    2. Status of digital signals while you are measuring?

    Also can you try to repeat the measurements with adding 0.1uF capacitor on the REFOUT pin?

    Regards,

    AK

  • Hi AK,

    Thanks for your reply.

    The impedance of both multimeters was set to >10G Ohm.

    All digital signals were idle during the measurement. The clock is only present when a command is send.

    I will repeat the measurement with an additional 0.1uF cap connected on the REFOUT pin. However, due to COVID, access to the lab is limited, so it may take a day or two before I can try this.

    Regards,
    Jeroen

  • Hi AK,

    I have repeated the measurement with an additional 0.1uF cap soldered over the existing 0.15uF cap. Please find below the results for each DAC device. Unfortunately, it's not looking any better, rather worse.

    DAC1 (34401A 6.5 digits DMM, >10G Ohm)

    - Internal Voltage Reference: +2.4969399V => Error = -3.060 mV
    - Internal Voltage Reference: +2.4969512V => Error = -3.049 mV
    - Internal Voltage Reference: +2.4969461V => Error = -3.054 mV

    DAC2 (34401A 6.5 digits DMM, >10G Ohm)

    - Internal Voltage Reference: +2.4946933V => Error = -5.307 mV
    - Internal Voltage Reference: +2.4947056V => Error = -5.294 mV
    - Internal Voltage Reference: +2.4947015V => Error = -5.299 mV

    DAC3 (34401A 6.5 digits DMM, >10G Ohm)

    - Internal Voltage Reference: +2.4963606V => Error = -3.639 mV
    - Internal Voltage Reference: +2.4963545V => Error = -3.646 mV
    - Internal Voltage Reference: +2.4963524V => Error = -3.648 mV

    The components were soldered at a typical temperature of 240 degrees Celsius (Vapour Phase). Please advise if that's OK?

    Regards,
    Jeroen

  • Hi,

    I don't think this has anything to with solder shifts, our characterization data shows that solder shift can be some where in 100's of microvolts range. So at this point in time, I don't suspect Vapour phase soldering.

    How is the thermal pad connected in the layout? Can you please share the same?

    Also I want to know how clean is your VAA, VCC & VSS supplies? In multimeter what NPLC setting you used to measure these readings? if its set to 1, can you repeat the measurements with NPLC = 10 ? I just want to eliminate the noise aspect from the measurements due to power line induced AC noise

    Regards,

    AK

  • Hi AK,

    The thermal pad is connected with the internal ground layers by means of 9 vias (0.25 mm), see figure below (red = solder paste).

    All measurements were conducted with the NPLC set to 10. 

    I also measured the VDD, VAA, VCC and VSS supplies with a scope (didn't save the plots) and didn't see anything suspicious. Ripple and noise are the same as when I connect the probe tip with ground (< 10mVpp). With the BW limiter enabled it's less than 2mVpp.

    Regards,
    Jeroen

  • Hi,

    Can you please share the plots of the same? Just want to verify

    Also I assume VREF is not connected anywhere other than the meter.

    Regards,

    AK

  • Hi AK,

    The REF pin is only connected to the meter. There's option to fit an external voltage reference, but those parts are currently not populated on the board.

    I have measured the internal reference of the 3 DAC chips on two other boards. They seem to have the same issue. 

    Board #2

    DAC1: Internal Voltage Reference: +2.4964896V (34401A 6.5 digits DMM) => Error = -3.5104 mV

    DAC2: Internal Voltage Reference: +2.4958939V (34401A 6.5 digits DMM) => Error = -4.1061 mV

    DAC3: Internal Voltage Reference: +2.4970576V (34401A 6.5 digits DMM) => Error = -2.9424 mV

    Board #3

    DAC1: Internal Voltage Reference: +2.4950648V (34401A 6.5 digits DMM) => Error = -4.9352 mV

    DAC2: Internal Voltage Reference: +2.4967137V (34401A 6.5 digits DMM) => Error = -3.2863 mV

    DAC3: Internal Voltage Reference: +2.4960986V (34401A 6.5 digits DMM) => Error = -3.9014 mV

    Could it be a wrong batch? 

    Please find below the plots of VDD, VAA, VCC and VSS, all AC coupled (I do have some DC coupled plots too if you want).

    The scope used is an Agilent MSO-X 3054T, 500MHz, 200MSa/s. 

    Reference measurement (AC coupled, persistence mode, w/o BW limiter):

     

    +5V VDD (AC coupled, persistence mode, w/o BW limiter):

    +5V VAA (AC coupled, persistence mode, w/o BW limiter):

    +11.5V VCC (AC coupled, persistence mode, w/o BW limiter):

     

    -11.5V VSS (AC coupled, persistence mode, w/o BW limiter):

     

    Regards,
    Jeroen

  • Hi Jeroen,

    Looks like these are very high frequency noise components. Our reference's PSRR will be limited at those high frequencies which may find a way through the sensitive analog circuitry to outputs. Do you have any other sample device to verify this is issue with this batch?

    Do you need samples for this device to test out? I can arrange the same or I can do the tests myself, but it may take a week's time as I am out of office.

    Regards,

    AK

  • Hi AK,

    I know it doesn't look very clean, but if you compare the 'reference measurement' plot (probe tip connected to GND) with the plots of the supply rails, you don't see additive noise, they even look better. Also note I used 10x scope probes (for convenience) where I should have used a direct BNC connection or 1x scope probe. I believe the actual noise is much less than what is shown on the plots. I will try to make new plots later this week (with and without 20MHz BW limiter enabled). 

    We don't have other sample devices available. But even if we had these, we can't (re)place the device in-house because of the ground pad. For that we have to send it back to our EMS, what can take a while before we got the PCB back. 

    Probably later this week we are going to populate one of the PCBs with external reference devices. If results are well, we probably take that route as time is becoming an issue.

    Regards,
    Jeroen

  • Hi AK,

    I just had an internal discussion about replacing one or more DAC chips to verify if there's an issue with this batch. 

    Could you please provide me with three samples, or at least one?

    Also, is the batch number part of the device marking? The DAC chips on all our PCBs have the device marking as shown in the figure, does that mean they come form the same batch?

    Regards,
    Jeroen

  • Hi,

    I will check the package marking with my team.

    Meanwhile if you can send me your address details, I can arrange samples or you can order samples from our store free of cost.

    Regards,

    AK

  • Hi AK,

    I tried to order the samples for the online store but I don't have access, see message below.

    It would be appreciated if you could send me the samples. However, our procurement department would like to know if there're shipping costs attached? 

    Regards,
    Jeroen 

  • Hi,

    I can ship the samples from Ti store to your address.

    There won't be any shipping costs. Can you pleases send me your complete address to a.k@ti.com ?

    Regards,

    AK