See these resources:
I posted the latter form on my door, and referred students to it when they came grumbling. They were furious the first couple of times, but then learned that it was a moot point to argue with me. (Although there was a case when I screwed up the final grade computation, and had to submit a few dozen grade change slips).
On a more serious side, I stopped doing the 60-70-80-90-100 cutoffs in the latest courses at all. Think about this: you are spending 70% of the test material and volume to test for a C, and then another 10% for a B, and then another 10% for an A. A much better use of your test problems is to have 1/3 of the problems address the grade of C, then another 1/3 of more complex problems to address the grade of B, and finally the most difficult 1/3 of problems address the grade of A. So in my exams, I would state something like:
There are 12 problems on this test marked at level C; you need to get 10 of them right to get the grade of C. There are 10 problems on this test marked at level B; you need to get at least 8 problems at level B or higher, and at least 15 problems total, to get the grade of B. There are 8 problems on this test marked at level A; you need to get at least 6 problems at level A, and at least 20 problems overall, to get a grade of A.
That is, out of 12+10+8 = 30 problems, you need to do two thirds, pick and choose, to get a full A, and only a third to get a C. You got 4 A problems, 4 B problems, and 8 C problems? That's a B+ in my books; you are a far shot for the required 20 problems for A, but you fulfilled the B requirements, and slightly exceeded them by having completed a bunch of A problems. The problems would be clustered, on most occasions: a given "big" problem would start with two-three C points, progress to one or two B points, and then culminate in an A point. Some A problems would be stand alone ones. So there would be 8 to 10 "big" problems and items within it. The smartest students would work on 6 problems, get everything right, and leave early. Not so bright students would attempt everything and fail at everything, and walk away with a C. I've seen all of the different ways that students approached it, which showed their learning styles and testing strategies -- pretty interesting per se.
While initially confusing (I train students to it by giving like three quizzes based on this system in the first three weeks), this system works very well in the end. Before each test, I also give students rubrics stating what they need to know at C/B/A level (C level: know the formula; B: determine which formula to use in a given simple context; A level: know where the formula breaks down, and how), so there was very little arguing about grades: either you've done the problem to my liking, or you haven't; and then you just count the completes up.