To be honest, I never understood this letter grading system used in the US (and maybe elsewhere)... and if you already have scores (like 40/100 vs 85/100), why even bother with another system?
A = 1 standard dev above the mean. B = mean. C = 1 std dev below the mean, but still performed remotely adequately. D = "If your major is theater and the class is math, I won't stop you from graduating."
Interestingly, in many of my middle- and high-school classes, grades were not based on standard deviations from the mean of actual class performance; they were a fixed scale: 90% or better was an "A", 80 - 89% was a B, 70 - 79% was a "C", etc. So you really could have the entire class be "better than average" (by the more usual method of computing a "C" grade) if the teacher made an assignment too easy.
My son's school still does this, though they've slid the scale up a bit, you need a 93% or better now to get an "A", and they've added a safety net: if you turn in any kind of reasonable attempt at all work assigned you will not get lower than a D- in the class.
There's no requirement that you curve the scores to any real distribution after the fact--you can design your tests with the intention that the average student gets 70-79% of the questions correct and students in other quintiles get higher or lower scores, or with the intention that the average student gets 80-89% of the questions correct, or with some other mix of intentions and related test scores. With some experience and fiddling I bet you could get a grading system like you describe which fell along a standard distribution.
It can even vary within schools. At my university, the classes in my major (Chemistry) were graded as you describe but other classes I took (like Psychology) used the inflated scale.
If I slack off and you study hard, the professor is comparing my 40/100 to your 85/100. This results in me getting a lower letter grade.