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.

DAC8750 Zero Drift Question

Other Parts Discussed in Thread: DAC8760, LM7705, DAC8750

Hi.

I'm working on new design including analog output, and I have preferred DAC8760 but I met problem about zero drift especially voltage output of DAC8760 (approx 9.5mV at Zero).

Power on my design 12V unipolar as expected. Actually, I know at least -500mV supply rail should be VADD- and GND. But I did not expect offset drift this much. I can use LM7705 and provides negative bias of ~230mV, but pcb space restrict me.

My question about DAC8750. I have checked the datasheet and I saw DAC8750 operates with a unipolar power, this suitable for me (my supply voltage changes 12-36V unipolar). Is there any possibility encounter with the same problem if I used DAC8750 with unipolar power?

Thank you in advance.

  • Hi Comme,

    comme_il_faut said:
    Is there any possibility encounter with the same problem if I used DAC8750 with unipolar power?

    The DAC8750 is current output only. The current output won't display a zero scale error, but it will display an offset error of 0.17 %FSR across temperature.

    comme_il_faut said:
    Power on my design 12V unipolar as expected. Actually, I know at least -500mV supply rail should be VADD- and GND. But I did not expect offset drift this much. I can use LM7705 and provides negative bias of ~230mV, but pcb space restrict me.

    Unfortunately in order to guarantee the specs in the electrical characteristics table the 0.5 V footroom is required. Without it you can observe devices that display a  50 mV zero scale error.

    Do you require both a voltage output and a current output?

  • Not necessary both voltage and current output, but It would be nice.
    I decide to use DAC8750, thank you.
  • I have another question.
    On DAC8760 datasheet and DACx760EMC-EVM User Guide, TVS diodes used on analog and and voltage output. TVS diodes is bipolar, thats OK, because recommended supply power is bipolar.
    Bipolar TVS diode recommendation also available on DAC8750 datasheet, but power is 24 V and recommendation power is unipolar. What is the issue if I use unipolar 36V unipolar tvs diode on Iout for DAC8750?
  • Hi Eugenio.

    I am using a DAC8760. I noticed an 8mV zero error factor too and now I know that is necessary a negative voltage to Correct it. (Right?) But what I can do in order to compensate the FS voltage not 10V (9.94V in fact)? Can I regular it with Software gain or over range? The same question for the current (20mA is 19.94mA real)... Please tell me ASAP. Thank you very much for your reply.

  • Gio,

    The Electrical Characteristics Tables in the Product Datasheet indicates that the data was collected assuming -500mV footroom for all unipolar data sets. Under such conditions, typically at room temperature the zero-scale error is 100uV or +/-600uV maximum. Data I have collected outside the datasheet suggests closer to 8mV typical and maybe 10mV max at room temperature, however I do not really have a statistically significant / large enough database for the maximum value to be consider a reliable 6-sigma value.

    With that in mind, the answer to your first question is that the zero-scale error will improve from the ~8mV value you are currently observing to closer to 100uV typical or +/-600uV maximum if you provide a -500mV footroom or more.

    Your second question concerning full-scale error is a bit more involved. The full-scale error can be a product of output voltage swing to rail limitations, gain error, and offset error. If you're using a supply that also provides at least 500mV of headroom for VOUT and provides adequate compensation for compliance voltage for the IOUT, then swing to rail limitations are not likely to be what is causing the errors you see at full-scale. Instead it is probably a product of offset and gain error. Offset error is a positive or negative offset that moves the entire linear region of the transfer function up or down. For VOUT in the DAC8760 datasheet for example there is offset error again of 100uV typical and +/-600uV maximum (this is the same as zero-scale error because the PDS is assuming negative footroom is provided and therefore basically the entire transfer function is in the linear region of operation). By itself this means that you may fall short of the FS value by as much as +/-600uV. The gain error specification adds a similar component where you may be off by as much as 0.04% FSR at full-scale (along with the offset spec). All together offset and gain errors at 25C could contribute as much as 4.6mV error at full-scale, and perhaps a small signature of INL errors are creating what you've observed.

    The offset and gain calibration registers can be used to correct these errors over the linear region of operation, but this is purely digital calibration. That means that the behavior near the end-points could remain the same depending on the direction of these error sources because it is not possible to generate extra-codes. To reduce error near the rails would require analog calibration or both analog and digital calibration (for example using the 10% over-range mode and then using digital calibration with a negative gain calibration coefficient).

    For IOUT there's not necessarily an over-range feature directly but you might use one of the wider ranges to achieve a similar result.

    I hope this helps.