R
Member
A mixer never gives you just sum and difference — it gives you the whole m*f_RF ± n*f_LO zoo. When a spur lands in your IF band, here's how I hunt it down:
1. Map it first, guess later
Measure the spur frequency precisely, then solve m*f_RF ± n*f_LO = f_spur for small integers m, n. A spreadsheet with m,n from -5..5 covers 95% of real cases. Knowing it's a 2x2 or 3x1 product tells you exactly which port is the culprit.
2. Vary LO drive and watch
Real mixer products follow predictable slopes: a 2x1 product moves 2x with RF and 1x with LO. Nudge the LO by 1 MHz — if the spur moves 2 MHz, you've confirmed the order. This beats guessing every time.
3. Check your drive levels
Overdriving the LO pushes the mixer deeper into nonlinearity and the higher-order products grow fast. Back the LO off 1-2 dB from the datasheet nominal and re-measure; I've seen 10+ dB spur improvement from that alone. Underdriving is just as bad — conversion loss rises and the noise floor eats you.
4. Filter at the right port
Once you know the product order, filter the offending frequency at its source port, not at the IF output. A low-pass on the LO port kills LO harmonics before they ever mix; an IF bandpass alone just hides the symptom.
5. Don't forget the test setup
Cable leakage and poor shielding create "spurs" that aren't in your DUT at all. Swap in a through line and confirm the spur disappears before blaming the mixer.
What's the nastiest spur you've chased — and what order turned out to be the culprit?
1. Map it first, guess later
Measure the spur frequency precisely, then solve m*f_RF ± n*f_LO = f_spur for small integers m, n. A spreadsheet with m,n from -5..5 covers 95% of real cases. Knowing it's a 2x2 or 3x1 product tells you exactly which port is the culprit.
2. Vary LO drive and watch
Real mixer products follow predictable slopes: a 2x1 product moves 2x with RF and 1x with LO. Nudge the LO by 1 MHz — if the spur moves 2 MHz, you've confirmed the order. This beats guessing every time.
3. Check your drive levels
Overdriving the LO pushes the mixer deeper into nonlinearity and the higher-order products grow fast. Back the LO off 1-2 dB from the datasheet nominal and re-measure; I've seen 10+ dB spur improvement from that alone. Underdriving is just as bad — conversion loss rises and the noise floor eats you.
4. Filter at the right port
Once you know the product order, filter the offending frequency at its source port, not at the IF output. A low-pass on the LO port kills LO harmonics before they ever mix; an IF bandpass alone just hides the symptom.
5. Don't forget the test setup
Cable leakage and poor shielding create "spurs" that aren't in your DUT at all. Swap in a through line and confirm the spur disappears before blaming the mixer.
What's the nastiest spur you've chased — and what order turned out to be the culprit?
