DACx760, how to use external Rset

Steve,

The DACx760 current output stage uses a two stage high-side voltage to current converter. The first stage is a current sinking stage that is mirrored and gained by the second stage. Inaccuracies of the current setting resistor that is connected to ISET-R or the internal resistor contribute gain error. For the 4-20mA output span, the current setting resistor can also impact the offset error.

Page 4 and 5 in the DACx760 datasheet describe the characterized gain error performance with the internal resistor as well as an external resistor for all of the current output spans. Page 5 also includes characterized offset error performance with both resistor configurations. The characterization was done using 0.1% tolerance resistors with 10ppm/C drift.

You'll notice that the typical values look very similar in all cases but the maximum values and performance over temperature favors the external current setting resistor. As it relates to total unadjusted error, using the external resistor carries more impact in the 4-20mA output setting since the resistor tolerance impacts both offset and gain error terms.

Using a resistor with better tolerance and/or better drift would make using the external resistor option more effective. Alternatively, you could consider implementing analog or digital calibration techniques but, of course, that requires that each unit is calibrated at production.

Steve,

Happy to help.

steve xu said:

1). What do you think I use a 0.01% (instead of 0.1%) tolerance resistor?

As I hinted at, there are other mechanics at work inside the device that will contribute error to the system. For instance, the current gain between the first stage and the second stage is controlled by the ratiometric matching of two resistors inside the device. So even if you had a 100% perfectly accurate 15kOhm resistor with 0ppm/C drift there would still be some offset and gain error. Unfortunately, I do not have data from characterization that expresses the magnitude of that error so it's hard for me to say how much better a 0.01% resistor would be than a 0.1%. All I can say is that it will be better based on intuition and what I know about our process tolerance in general.

As we talked about, a lot of the error here comes from drift. Just as with the case above - I'm not sure how much of the drift came from the current setting resistor (whether internal or external) versus the rest of the signal chain inside the device because I don't have that data. Again, though, through intuition I would anticipate some improvements from a super low drift external resistor - I think I've seen as low as 0.2ppm/C resistors in the 15k variety you'd need here.

steve xu said:

2). How to enable the "External current resistor"? I didn't see a setting of it in the registers. I like to make some experiments.

You can see this on page 32 of the datasheet. Use of the internal versus external current setting resistor is controlled by the "REXT" bit (bit 13) in the Control Register (address 0x55). A logic 0 in this bit means the internal resistor will be used. A logic 1 in this bit means the external resistor will be used.

steve xu said:

3). Do you have any other ideas?

Without implementing some kind of calibration our options to enhance the device's performance from the outside are sort of limited. One other suggestion or comment I will give you is to consider using an external reference. You'll notice that the conditional statements above the electrical characteristics table on each page of the datasheet indicate that a 5V external reference was used to generate the results in the table. If you do not do the same, you will face additional error contributed by the reference accuracy. We have a variety of high accuracy low drift shunt and series reference solutions. I would suggest a series reference since they are generally more accurate.