Peak downlink data rate
0 Gbps
Spectral efficiency
—
Aggregated bandwidth
—
Other direction
—
10 GB download
—
How you compare
Your config vs classic builds
Under the hood
The exact TS 38.306 formula
data rate = 10−6 × Σj=1…J [ vLayers(j) × Qm(j) × f(j) × Rmax × ( NPRBBW(j),μ × 12 ) / Tsμ × ( 1 − OH(j) ) ] Mbps
Resource grid per second. Tsμ = 10⁻³ / (14 × 2μ) is the average OFDM symbol time. At μ=1 (30 kHz SCS) a 100 MHz carrier has 273 PRBs × 12 = 3276 subcarriers, giving 3276 / 35.71 µs ≈ 91.7 million resource elements per second.
Bits per resource element. Multiply by modulation order Qm (8 for 256-QAM), MIMO layers v (4), the UE scaling factor f (1) and the peak code rate Rmax = 948/1024 ≈ 0.9258 → ≈ 2717 Mbps of raw coded capacity.
Subtract overhead. FR1 DL reserves 14% for DM-RS, CSI-RS, PDCCH and SSB: 2717 × 0.86 ≈ 2337 Mbps — the famous 2.34 Gbps figure quoted for 100 MHz 4×4 256-QAM.
TDD shares time. A DDDSU pattern gives DL 3 slots + 10/14 of the S slot every 5 slots → ×0.743 ≈ 1736 Mbps. The same math with the UL share (×0.229) prices the uplink. Carrier aggregation just sums this per component carrier.
Reference data
The tables behind the math
Maximum transmission bandwidth NPRB from TS 38.101-1/-2 Table 5.3.2-1, and the TS 38.306 overhead factors.
| BW (MHz) | 15 kHz | 30 kHz | 60 kHz |
|---|
| BW (MHz) | 60 kHz | 120 kHz |
|---|
| Item | Value |
|---|---|
| OH — FR1 DL / UL | 0.14 / 0.08 |
| OH — FR2 DL / UL | 0.18 / 0.10 |
| Rmax (peak code rate) | 948 / 1024 ≈ 0.9258 |
| Qm — QPSK…1024-QAM | 2 · 4 · 6 · 8 · 10 |
| Layers — DL / UL max | 8 / 4 |
FAQ
Common questions
How is 5G throughput calculated?
3GPP TS 38.306 §4.1.2 defines the peak data rate as a sum over component carriers: MIMO layers × modulation order Qm × scaling factor f × Rmax (948/1024) × resource elements per second × (1 − overhead). This calculator implements that formula exactly, then applies your TDD slot pattern's time share — which the spec formula leaves to deployment.
What's the maximum speed on 100 MHz of FR1?
30 kHz SCS → 273 PRBs. With 4 layers and 256-QAM: ≈ 2.34 Gbps on FDD or full-buffer TDD, ≈ 1.74 Gbps with a DDDSU pattern. With Rel-17 1024-QAM the FDD figure rises to ≈ 2.92 Gbps.
Why is my drive-test speed lower than this?
This is the PHY-layer theoretical peak: rank-4 the whole time, top MCS, zero HARQ retransmissions, one UE owning all PRBs. Real cells schedule many users, run adaptive MCS, and lose ~10% to TCP/IP above PHY. Hitting 30–70% of peak in the field is normal and healthy.
How do TDD patterns change the result?
TDD splits one carrier in time. DDDSU = 3 DL slots, 1 special, 1 UL per 5 slots; counting 10 of the S slot's 14 symbols as DL gives DL ≈ 74.3% and UL ≈ 22.9% of the time. The calculator shows both directions priced with the same pattern, and you can set a custom split.
What does the scaling factor f do?
f ∈ {1, 0.8, 0.75, 0.4} is a UE capability signalled per band combination — it lets a UE support a combination without the full baseband budget on every carrier. For network peak figures keep f = 1.
Does carrier aggregation double the speed?
CA sums the per-carrier rates, so 2 × 100 MHz FR1 ≈ 4.67 Gbps FDD-equivalent. In practice the carriers often differ — e.g. n78 100 MHz TDD + n28 20 MHz FDD — and this calculator prices each CC with its own bandwidth, SCS, MIMO and TDD pattern before summing.