A lot of times, the test limits used during manufacturing test are identical to the limits in the product's requirements. I strongly advocate against blindly doing this. The manufacturing test limits need to be tied (at least partially) to the actual design intent.
Let's say that marketing wants a circuit board with a resistor divider that divides V_in by two for V_out. Picking numbers out of the air, marketing puts down 5% as the V_out accuracy. Therefore, if V_in is 2V, V_out can be 0.95-1.05V.
The design engineer looks at the specifications and decides to pick resistors that are 1% accurate. If one resistor is +1% and the other resistor is -1% (I'm ignoring temperature effects), then with a 2V input, the output voltage can range from 0.99-1.01V.
And then in manufacturing, what if a 5% resistor gets used? Perhaps, there's a counterfeit part. Perhaps, the wrong reel is loaded. (Don't laugh.) If we had blindly chosen the original 5% tolerance, we would still be within the specification, but likely outside of the design intent of 1%. Even if that original specification is met, we would now be releasing a product that does not match what was used during design verification testing.
In a future post, I'll talk more about taking in the manufacturing test limits further.
