Yet Another Simple Dummy Load Circuit (with problems)

[brainpowered]

Background
i've been wanting to build my own very simple dummy load for a few yeas now. And now i've finally put some parts on a breadboard. At first (or third) try, it all seemed to work. However occasianlly, something wierd goes on and then the circuit kinda behaves funny.

(see third attachment)

So,
1. a Potentiometer sets the Current (1V/V basis)
2. first opamp 'follows' set current voltage and drives the mosfet
3. a 0.1Ohm sense resistor provides feedback to the 'feedback opamp' with a gain of 10x

In theory the circuit should work as a simple voltage follower, right....?

(see first attachment)

Well, sometimes... So it managed to work on a breadboard, somehow, at some point, i thought it worked, but now that i decided to build a Protoboard version, it really didn't work at all!

The Current floiwing thought the Mosfet it either waaay too high, or other times it's waaay too low, it seems to behave differently each time i built a new version of it on a breaboard, the exact same circuit....! Gremlins? Or just a misunderstanding in Opamp theory.... whats wrong here guys, .... i'm at my wits end.

I thought my circuit theory was on my side, but alas, here I am, asking for help.

What do you guy think? (ignore the 'Fan Controller' block for now, it actually works great!)

My schematic may be slightly oversimplified, but EasyEDA seems to lack standard symbols, or i'm just completely amateur.
I've also included a Fritzing diagram for how i plan on wiring it a breadboard.

(see second attachment)

[opampsmoker]

You need to set the error amplifier (opamp) up as an integrator.
Attached is the kind of thing..its in the free download LTspice

[ledtester]

Here's a recent forum thread on designing an electronic load that you may find helpful:

https://www.eevblog.com/forum/projects/constant-current-dummy-load-ran-a-gutter/

[magic]

That's not unity gain, its 0.1x gain. The output moves 100mV for 1V change in input voltage.
Furthermore, the transistor adds some phase delay which may destabilize a unity gain stable opamp even at unity gain (without the 10x feedback amplifier).
The regulation amplifier (call it A1) likely oscillates in this configuration and goes nuts.

Common solution: resistor between A1 input and A2 output and a small capacitor from A1 output straight to A1 input.
There was a dozen threads about it, which may give you some starting points for experimentation.

BTW, A2 can be dropped altogether and input voltage to A1 divided 10x instead. Barely any difference in accuracy, possibly more bandwidth achievable that way(?)