setting reference value using DAC

[electros6]

Hi guys,
           I am working on DC electronic load. I used MCP4921 12bit DAC to set the set value of current,power. when I set the value in the DAC and run the load I found that there is some variation about 3% then the set value. I used 1% resistance for current sense. My question is whether in commercial instruments the set value is set  based on the calculated value or feedback loop is used to set the DAC value. Why am I asking this because , in agilent E3631 power supply SINGLE DAC used to set 7 different values with a reference rate of 13ms , if feedback loop is used how it is possible to do refresh in 13ms, OR if value is set based on the calculation then if the DAC is drif due to aging how it is compensated. 

[tszaboo]

You cannot compensate any (open loop) drift because it is unpredictible. The E3631 uses sample and hold circuits very hard low pass filtering, like R26, C25 U10 (please find it on the schematic).
If you would post a schematic of your circuit, it would be easier to help. My suggestion in the blind> calibrate and use precision parts.

[electros6]

I have attached the power part of my schematics

[Jeroen3]

You should compensate the analog setpoint by running amplified current sensing back to the ADC.
Using an ina225 for example (it has programmable gain).

Making it a closed control loop.

[tszaboo]

Is this running stable? Have you looked at the load current with a scope?
There are a lot of parts on this schematic which can make 3% error.
R33,R37 can divide the voltage to a different factor than 0.5. Is R37 required?
The same way,  the small errors in R13-R16 could have a different amplification than 20. Maybe the common mode rejection is not good enought, creating an offset voltage.
And then there is R10-R11. If you are not using proper 4 wire routing, thatcan lead to errors, or the resistor itself can have big tempco, or it is just not precise enought for your application. If you want anything like more than 1% precision for this circuit, you need to calibrate it, and use components from trusted sources, reputable manufacturers, well defined datasheets.
If you need help choosing components, include the BOM, I'll try to help.

[void_error]

There are a few things that will affect accuracy: opamp input offset voltage, DAC errors (DNL,INL), imbalance between the resistors in U4 and U5 feedback loop (I suggest using an integrated difference amplifier if you don't want to hand pick resistors). Feeding the DAC output back to an ADC will allow you to tweak the output voltage and possibly minimize the DAC related errors.

[electros6]

Hi NANDBlog ,
           Thanks for your reply. I have prototype my design and looks to be works fine expect the issue that I told. If I need to choose components(resistors for amplification)  how to match the resistors. I have some offset errors due to bias current as the current increases. If I need to implement the feedback what type of control strategy ( eg PID ) should I use in this type of application. Also if possible please give me name of the good books in control systems.

[SteveyG]

That is one badly drawn schematic. It would be worth spending a bit of time making it more readable so if you come back to it in a few years you can understand it.

[electros6]

 I know it is bit messy.The schematic is drawn in proteus for simulation purpose. In that schematic the digital part is not shown.