My suggestion is to use the weight of the wheel that feels best to you… but to try to make sure you are getting fill telemetry… Which as far as I know which the AC model is to run less than 100% in game and probably significantly less if you are running LM type cars based on what I understand is how they set up the output, I haven’t done research into it so someone else may know better as to when a 100% output signal from AC/ACC occurs…
Regarding the forces at the wheel as I said those are for the most part controlled by the game as if the Game outputs a 100% signal then if you have your wheel set to give you 20Nm it will give you 20Nm if you have it set to give you 5Nm then it will give you 5Nm at that 100%.
iRacing works things differently as when in Linear/Nm mode You are the one that sets where the 100% point for the game falls in the telemetry spectrum so if you set the slider to 75Nm then 75Nm of telemetry = 100% if you set the slider to 30Nm then 30Nm of telemetry =100% to the game.
Now to if certain outputs are realistic… They are BUT and this is a big BUT… what we feel at the wheel also has to do with acceleration or the wheel, timing, yes the accuracy of the cars digital model (including the tire model)… FFB from games is not always timed exactly correctly whether at 60Hz or 360Hz or 2000Hz which is part of why we end up with oscillations as the wheel is always going to be slightly behind the game at the minimum. as well these wheels SPIN way faster than the wheel of a real car which means that the implementation of force will make the wheel move quicker which causes us as humans to feel that force more pronounced than if everything else were equal. This is the area where getting proper Filter settings really helps… i.e. if you are using a Small Mige with no filtering then you have it at its most active state which can get VERY VERY brutal at higher powers… it needs to be tamed back to reality through the drive filtering, so just turning up the power and having it feel perfect is NOT really an option (you have to play with settings). Then there is the Tire and Car models used in game. Unless everything is perfect (including the timing) everywhere they will not mimic the real thing exactly… In talking with the guys developing at iRacing one of their biggest issues is the Tires as NO tire manufacturer offers up any data regarding the performance of their tires (they are very secretive). Therefore iRacing (and other game companies) have to choose a tire model that they feel will work… thing is Tires are one of the Biggest factors in a cars handling (ie you can take a car that will only corner at .7G’s in real life well over 1G JUST by changing the tires). In addition different compounds and construction of tires can greatly affect whether you have really good off-center response or whether the car just feels sluggish. That tire that feels sluggish could have 4x more grip mid corner than the responsive tire but it doesn’t feel good when turning into the corner. This is why your Daily driver car feels different if you get different brands or levels of tires when you replace them.
So basically the simulation will never be EXACTLY like real life… they get as close as they can in dynamics but there will always be differences. So in the end just try to maximize the Range of feedback that the wheel receives and then once that is done adjust the power at the wheel. Then if you feel you need more power at the wheel you will have to compromise the Range.
For iRacing the Maximum Range is having the iRacing slider at about 60 - 65 Nm
Personally I run 47Nm in iRacing with a Large Mige at 25A 100% - which equates to just under .6:1 Specific Output (power out : telemetry in)