Wider channel? Check! Ultra-wide antenna spec? Check! Yet the link stumbles, range shrinks and EVM climbs. That’s because bandwidth is a budget, open it up and you pay in noise, linearity, synchronization, filtering and integration. “More MHz” can expose weaknesses the narrowband demo politely hid.
-
What Bandwidth Really Buys (and Costs)?
Extra bandwidth enables higher symbol rates, better ranging resolution and lower latency but it also integrates more thermal noise, widens phase-noise skirts and invites more interferers into your front end. Filters get tougher, ADCs must sample faster with the same ENOB and PAs must stay linear over a broader spectrum. Net, capacity may go up in perfect conditions and down in the real world. -
Antennas and Matching Aren’t Free Lunches:
Pushing fractional bandwidth stresses electrically small antennas and matching networks. Efficiency drops as Q rises, radiation patterns warp across frequency and Bode–Fano limits show you can’t get a deep, wide match without paying in loss or size. A “VSWR < 2 from 600 MHz to 6 GHz” line is easy to print and hard to make radiate well everywhere. -
Channels Are Frequency-Selective Not Flat:
Wider signals encounter more multipath delay spread, producing frequency-selective fading. Equalizers and OFDM help but PAPR, synchronization and guard intervals eat into SNR and power headroom. As bandwidth grows past the coherence bandwidth, parts of your spectrum carry while others crash, averaging to disappointment. -
Critical Formulas:
a). Thermal noise power:
→ N = k · T · B · F
b). EVM–SNR relation (AWGN approx):
→ EVM_rms ≈ 1/√SNR_lin
c). Coherence bandwidth vs delay spread:
→ B_c ≈ 1/(2π·τ_rms)
d). Fractional bandwidth vs Q (resonant antennas, approx):
→ FBW ≈ (f₂ − f₁)/f₀ ≈ 1/Q -
Real-World Surprises:
- A 160 MHz Wi-Fi link in a crowded apartment delivered lower throughput than 80 MHz because integrated noise, neighbor overlap and DFS events crushed SNR and airtime.
- A “600 to 3000 MHz” PCB antenna met VSWR spec but averaged <40% efficiency across the band, a narrower 860-960 MHz design doubled range at the target frequency.
- A 77 GHz automotive radar widened its chirp to improve resolution only to raise ghost detections in wet, reflective scenes as sidelobes and phase noise integrated over the sweep.
- A 100 MHz 5G NR channel showed worse cell-edge rates than 40 MHz because PA headroom, UE EVM and base-station phase noise dominated the budget.
Bandwidth is power and penalty. Use as much as your noise, linearity, synchronization and antenna efficiency can actually support and no more. Smart RF isn’t “wider by default”, it’s “wide enough to win”.
LinkedIn: 