R
Member
EVM is the number everyone asks for, but a spectrum analyzer is not a vector signal analyzer, and pretending otherwise is how bad boards ship. Here is how I use a mid-range analyzer for EVM work without fooling myself.
First, what it can do: with a digital demodulation option, an analyzer can give you a useful EVM reading on a known modulation, and it is good at showing you whether EVM degrades with output power, which usually points at PA compression or supply sag. Sweep the power and watch the EVM curve — a knee in that curve is worth more than a single spot number.
Second, what it cannot do well: it will not separate transmitter EVM from channel effects the way a proper VSA with equalization does, and its own phase noise floor sets a floor on the EVM you can measure. If your measured EVM stops improving as you clean up the DUT, you are probably measuring the analyzer, not the device. Check this by measuring a known-good reference source first.
Third, the setup traps: wrong RBW and averaging will smear the result, and forgetting to disable the analyzer's own preamp when measuring a hot signal will give you a flattering number that disappears on a real receiver. I always record the analyzer settings alongside the EVM number, otherwise the number is not reproducible.
My rule of thumb: use the analyzer for trend and comparison work, and reserve pass/fail claims for a proper VSA or a receiver-based test. What is your cutoff — at what point do you stop trusting an analyzer-based EVM number and reach for the VSA?
First, what it can do: with a digital demodulation option, an analyzer can give you a useful EVM reading on a known modulation, and it is good at showing you whether EVM degrades with output power, which usually points at PA compression or supply sag. Sweep the power and watch the EVM curve — a knee in that curve is worth more than a single spot number.
Second, what it cannot do well: it will not separate transmitter EVM from channel effects the way a proper VSA with equalization does, and its own phase noise floor sets a floor on the EVM you can measure. If your measured EVM stops improving as you clean up the DUT, you are probably measuring the analyzer, not the device. Check this by measuring a known-good reference source first.
Third, the setup traps: wrong RBW and averaging will smear the result, and forgetting to disable the analyzer's own preamp when measuring a hot signal will give you a flattering number that disappears on a real receiver. I always record the analyzer settings alongside the EVM number, otherwise the number is not reproducible.
My rule of thumb: use the analyzer for trend and comparison work, and reserve pass/fail claims for a proper VSA or a receiver-based test. What is your cutoff — at what point do you stop trusting an analyzer-based EVM number and reach for the VSA?
