Help choosing OpAmp

[AlxDroidDev]

I am building a small electronic load, based on an Atmega1284P and using a IRLZ14.

The Atmega1284P will be controlling a DAC121C085CIMM (12-bit DAC from TI). The OpAmp will sit between the DAC and the IRLZ14, and this is what I need help choosing. I will also be using a MAX6126 (5.0V) Voltage Reference connected to both the DAC and the Atmega1284.

In principle, a LM358 will work fine, since I'll have a 12V supply on the board (used to feed the V_REF, the fan and a relay). On the final project, however, I am wondering if a faster rail-to-rail OpAmp, with better characteristics won't be a better choice. I do have a few R2R opamps, and I undecided on how I should choose one that better fits the project.

This decision is exactly what I need your help with: what characteristics would be desired on an OpAmp for what I have in mind?

The attachment below is what I have so far, and uses the LM358D. It is still showing an Arduino Nano, but the final project will use the Atmega1284P, since the sketch I've developed so far already uses exceeds the 32kb of flash memory of the Atmega328.

[Brutte]

So you are using ADC to measure current, then you process the value by an AVR and then the control value is sent to a DAC and from there that goes to MOS transistor??

If so - that seems overengineered because the uC is in the loop (what bandwidth do you expect from 2 x i2c + avr?).

I'd make an analog voltage-controlled "resistor":
1V -> 10 ohms
1.1V ->  11 ohms
...
4.0V -> 40 ohms

that is driven in open loop with an RC filtered PWM (12-bit timer in CTC mode)

[Kleinstein]

There seems to be an ADC to independently measure the current. So no need for a really low offset, as this could be corrected. The LM358 is relatively slow and has a nasty dead zone. So a first improvement would be just a resistor to GND at the output of the LM358. If you wand it faster something like an TLC272 or higher specked TLC277 would be an improvement. Depending on the FET used one might need more than 5 V at the gate. Though one usually does not need to turn on the MOSFET that hard anyway.

To get a real resistance simulation, the DAC would get an ref voltage derived from the output voltage. Having a constant reference is making it a constant current sink.

The IRLZ44 is not that well suited for linear operation at more than about 15-20 V. A slightly better option would be an IRFP250 or similar (higher voltage and larger case).

[AlxDroidDev]

So you are using ADC to measure current, then you process the value by an AVR and then the control value is sent to a DAC and from there that goes to MOS transistor??

If so - that seems overengineered because the uC is in the loop (what bandwidth do you expect from 2 x i2c + avr?).

I've already built an analog dummy load, using a SUP85N03, a LM358, a potentiometer and a couple big ass 0R1 resistors!

The SUP85N03 is not exactly a logic FET, but it's V_GS curve is close enough for it to work as such for currents below 4A.

What I want with this project - and the reason it looks overengineered - is that I want a LCD display (thus the Nokia 5110 display) with a graphical UI, menus, logging to internal memory, logging to an internal EEPROM or serial port, and most of all, I want to be able to put all those things together in a single project.

Also, and another reason for the uC, I'd like the device to work at constant current, constant voltage or constant power, and except for constant current, the rest has to be done through software.