Last Post: April 8, 2021:
This linked article from LitePoint barely scratches the surface of a dirty little secret in the marketing and selling of Wi-Fi chip sets.
Error Vector Magnitude, or EVM, is a straight forward method of measuring and reporting the quality of OFDM signals. In a pinch, it can be used to replace a myriad of other tests. One can argue that this measurement, on its own, can tell you how your radio will transmit well in the real world and be "understood". Obviously power levels are also important, but EVM addresses how well your transmitter works.
Unfortunately, traditional EVM measurements have been bastardized to obscure transmitter problems via the use of NON-Standard "corrections" in the measurement process. While calculating EVM over an entire packet may seem like a valid statistical technique, it usually hides valuable indicators when "something is wrong". Likewise, "channel equalization correction", assumes a flat response across the channel, which also makes unrealistic assumptions. The only thing iy really buys you is shorter tests.
I have conducted thousands of radio tests, and have compared the data made with these "corrections" both enabled and disabled. These results can vary dramatically, sometimes horribly.
When comparing or purchasing radio components, make sure you understand the sellers EVM claims. You really need to compare apples to apples, and not be mislead by bogus EVM reports.
One other warning I would include is that you need to measure EVM over ALL possible rates. Don't trust reports made over merely a subset of the possible rates. It is sometimes the case that faster punctured rates will pass their EVM limits, whereas the slower un-punctured rates do not Intuitively this should seem backwards to you, and if nothing else, is a hint to take a closer look at the tests measurement parameters.