L
Admin
A comb (impulse / harmonic) generator is one of the cheapest honesty checks on an EMC or antenna bench — and one of the easiest to misread. Used well, it catches chamber, cable, and analyser lies before you blame the DUT. Used poorly, it rubber-stamps a broken setup.
**1. What a comb is good for**
- **Repeatability:** same comb, same geometry, same analyser settings → spectrum shape should overlay day-to-day within your noise + positioning budget.
- **Cable / adapter / preamp health:** a sudden tilt, a notch that moves when you flex a jumper, or a whole band dropping a few dB is usually interconnect or bias — not the DUT.
- **Chamber / site smoke test:** place the comb at a marked spot, antennas and turntable at marked positions; compare to a stored reference trace. Absorber left on the floor, a door ajar, or a mast cable re-routed shows up here.
It is a **transfer** check, not a primary calibration of absolute field strength unless the comb itself is characterised and you control geometry tightly.
**2. What a “flat-looking” comb does not prove**
- Absolute V/m or antenna gain — unless you have a calibrated path and a known comb effective output into that path.
- That your **DUT** measurement is correct — only that the receive chain and site did not jump since the reference.
- That analyser preselector / image responses are innocent at every harmonic — combs light up many frequencies; an image or mixer spur can sit on a tooth and look “real.”
Habit: label comb plots as **relative site+chain check** unless you have actually tied them to a calibrated field or power reference.
**3. Geometry discipline (same story as any transfer)**
- Mark comb height, orientation (most combs are not isotropic), distance to receive antenna, and cable dress.
- Same analyser: RBW/VBW, attenuator, preamp on/off, detector, sweep. A “reference” taken with preamp on and today’s run with preamp off is not a drift — it is a different measurement.
- If you change anything (adapter, cable, mast), take a **new** reference; do not compare across a hardware change and call it chamber drift.
**4. A short weekly / pre-campaign ritual**
1. Comb at marked position → save trace (or peak table at agreed teeth).
2. Flex the receive jumper and re-seat the adapter once; peaks should not jump more than your agreed tolerance (often ~1 dB at the teeth you care about, setup-dependent).
3. Compare to last week’s reference. If the whole comb dropped, chase cable/adapter/preamp before the absorber layout. If one tooth moved, chase that band’s antenna factor path, mixer, or a bad contact.
**5. Pair with a system metric when the DUT is the product**
For desense / RE work, a comb that still matches last week only says the *site* is stable. Still gate the product on the **system number** (RE limit, throughput, NF, spur) — the comb is the “did we break the lab?” check, not the certification.
**Discussion**
Do you keep a dated comb reference trace for your chamber or desk setup — and when a tooth moves, do you usually find cable/adapter, analyser settings, or real site geometry change first?
— rfforum 编辑部 / Editorial Team
**1. What a comb is good for**
- **Repeatability:** same comb, same geometry, same analyser settings → spectrum shape should overlay day-to-day within your noise + positioning budget.
- **Cable / adapter / preamp health:** a sudden tilt, a notch that moves when you flex a jumper, or a whole band dropping a few dB is usually interconnect or bias — not the DUT.
- **Chamber / site smoke test:** place the comb at a marked spot, antennas and turntable at marked positions; compare to a stored reference trace. Absorber left on the floor, a door ajar, or a mast cable re-routed shows up here.
It is a **transfer** check, not a primary calibration of absolute field strength unless the comb itself is characterised and you control geometry tightly.
**2. What a “flat-looking” comb does not prove**
- Absolute V/m or antenna gain — unless you have a calibrated path and a known comb effective output into that path.
- That your **DUT** measurement is correct — only that the receive chain and site did not jump since the reference.
- That analyser preselector / image responses are innocent at every harmonic — combs light up many frequencies; an image or mixer spur can sit on a tooth and look “real.”
Habit: label comb plots as **relative site+chain check** unless you have actually tied them to a calibrated field or power reference.
**3. Geometry discipline (same story as any transfer)**
- Mark comb height, orientation (most combs are not isotropic), distance to receive antenna, and cable dress.
- Same analyser: RBW/VBW, attenuator, preamp on/off, detector, sweep. A “reference” taken with preamp on and today’s run with preamp off is not a drift — it is a different measurement.
- If you change anything (adapter, cable, mast), take a **new** reference; do not compare across a hardware change and call it chamber drift.
**4. A short weekly / pre-campaign ritual**
1. Comb at marked position → save trace (or peak table at agreed teeth).
2. Flex the receive jumper and re-seat the adapter once; peaks should not jump more than your agreed tolerance (often ~1 dB at the teeth you care about, setup-dependent).
3. Compare to last week’s reference. If the whole comb dropped, chase cable/adapter/preamp before the absorber layout. If one tooth moved, chase that band’s antenna factor path, mixer, or a bad contact.
**5. Pair with a system metric when the DUT is the product**
For desense / RE work, a comb that still matches last week only says the *site* is stable. Still gate the product on the **system number** (RE limit, throughput, NF, spur) — the comb is the “did we break the lab?” check, not the certification.
**Discussion**
Do you keep a dated comb reference trace for your chamber or desk setup — and when a tooth moves, do you usually find cable/adapter, analyser settings, or real site geometry change first?
— rfforum 编辑部 / Editorial Team
