My system is similar, so I'll piggy-back on this. I think it does need to be slightly more complicated, because you are skipping the dynamic of forcing glancing blows and that different weapons do damage differently (assuming we have a system that allows for bludgeoning/piecing/whatever). Let me preface this by saying I understand that complexity is not always preferable, but I think armor systems are generally overly simple as a holdover from pen and paper gaming where simple is more necessary. Basically, rather than using critical hits, you should use a variable DR. That is to say, depending on the armor type, the defender's and attacker's relative skills, and the attack type, DR gets weighted on a statistical curve which the computer rolls against to determine actual damage from 0 to "unarmored."
If I were to design it, I'd give armor four categories: encumbrance or stamina drain, body coverage (corresponding to hit chance, could incorporate defense skill for things like shields), dissipation (bludgeoning defense), and yield strength (pierce defense). Ideally, better armor would have a higher chance of making a strike be:
A) A miss - Pretty obvious. The attacker’s skill (roll) doesn't even connect, resulting in no damage.
B) A glancing blow - the attacker's skill connects but severely fails to beat armor coverage, resulting in very high DR,
C) Bludgeoning - attacker's skill fails to beat coverage by a small margin, resulting in a strength and weapon type combined check against dissipation to determine DR.
D) Penetrating - The attack either beats the yield strength and goes through, are beats the coverage completely resulting in very low DR.
Note this isn't really dissimilar from D&D 3e rules; in 3e option 1 fails at the touch AC, 2 fails to beat AC, 3 hits normally, and 4 is a crit. What I am doing is refiguring 2 and 3 into B and C, saying some damage may be conferred on a hit to armor (like with a mace or magic) but will be mitigated depending on factors. Then critical are refigured into D, where bypassing or penetrating armor is similar to a critical hit, but depends as more on strength and coverage than attack rolls.
Then armor becomes more a rock-paper-scissors game. Lighter armor can afford high-coverage with less encumbrance or a higher dodge, but low yield strength and dissipation, allowing more attacks to be in categories A and D; thus it defends well weapons that have low hit chance but do high damage like war hammers or pikes. Medium armor affords movement and is strong, but does yields under a palpable hit, shifting attacks into B and D; thus it defends well against bludgeoning weapons like maces. Heavier armor would rarely fail, but have high encumbrance if the coverage is good, and could still be bludgeoned pushing most attacks into B or C and thus defends well against attacks that seek to pierce like rapiers or arrows. Active defense (such as parrying or shields) could help shift attacks into category B at the expense of attacking ability. Of course, I would do armor by pieces rather than as a suit, so you can mix and match. As for balance, damage types can be normalized with the expected hit values; that would be mathematically trivial with a computer just matching hit value with price/availability.
Of course this would be too time-consuming in paper-and pen RPGs, but I'd love to see it tried in a video game.
Edited for spelling. "I" before "E"?