Output voltage accuracy of Digital Pot vs DAC?

[treez]

Which would be the most accurate for giving a voltage output, the MCP4013 as a digital pot, or the MCP4716 DAC?

We need to give a control voltage to a led driver IC (LT3756) between 0v and 1.2v. Whatever voltage we set it to, we need it to be accurate to within 50mV…eg, if we set it for 0.98v, then we don’t want it being higher than 1.03v, as the led driver components would overheat.

We don’t mind that the mcp4013 only allows 64 output voltages steps, -as long as it gives these accurately then that is fine.

MCP4716 DATASHEET
https://ww1.microchip.com/downloads/en/DeviceDoc/22272C.pdf


MCP4013 DATASHEET
https://ww1.microchip.com/downloads/en/DeviceDoc/21978c.pdf

Looking at the INL parameter, (0.5LSB for mcp4716, and 0.25LSB for mcp4013, you would think that the MCP4013 digital pot was more accurate than the MCP4716 DAC)?

 

[FvM]

As you can easily calculate, 6 Bit resolution is already fullfilling your requirements (1.2V can be set with 18 mV stepsize +/- 4.5 mV INL error). The 10 bit DAC has higher resolution (1.2 mV) and absolute accuracy (+/- 0.6 mV), but won't be required.

 

[treez]

thanks, though the thing is I suspect that the error in the digital pot is cumulative, and every step can be 25% bigger in resistance than nominal?

- - - Updated - - -

Page 5 of the following says that
http://www.onsemi.com/pub_link/Collateral/AND8414-D.PDF

The linearity specifications of the digital POT are
similar to a DAC, one-half to one LSB

..So from this I presume that the digital pot is pretty good and does not have a dreaded chance of having say 30 consecutive steps which comprise individual resistances which are all 25% above the nominal st

 

[D.A.(Tony)Stewart]

Specs say full scale gain error is 5.2 LSbits and offset error 0.75 so worst case in theory is 6 LSbits, full scale.

In practise the analog ground must not shift due to load current at the control point.