G/T Versus RF Loss in Phased Arrays and SATCOM Receive Hardware

In a cold-sky Ka-band receiver, 1 dB of pre-LNA RF loss can cost approximately 2.1 dB of G/T.

In one published SATCOM architecture, reducing RF loss from 3 dB to 0.5 dB reduced the required array from roughly 1,500 to 500 elements. [Ref. 1]

That is not merely an antenna-efficiency improvement. It is an array architecture transformation.

Fewer elements can mean fewer beamformer channels.

Fewer channels can mean lower DC power, heat, calibration burden, control complexity, PCB area, and manufacturing cost.

The central engineering question therefore becomes:

Should we increase aperture, or stop losing the signal before amplification?

G/T Is the Receive Metric That Connects Antenna and Receiver Physics

For a receiving terminal:

G/T [dB/K] = Grealised [dBi] − 10log₁₀(Tsystem [K])

ITU defines earth-station G/T as antenna gain in the received-signal direction divided by system noise temperature. It also states that G/T plays a key role in determining satellite-link carrier-to-noise performance and required satellite resources. [Ref. 2]

This means receive performance can improve in only two fundamental ways: increase G or reduce Tsystem.

Most array discussions focus heavily on the first. The second can be equally powerful.

Free-Space Path Loss Is Not Part of G/T

This distinction matters.

G/T characterises the receiving terminal.

Propagation belongs elsewhere in the link budget.

ITU treats G/T as an earth-station figure of merit and separately relates it to the satellite-link carrier-to-noise calculation. [Ref. 2]

So when evaluating a phased-array terminal, do not mix path loss into the intrinsic G/T number.

The propagation channel and receiver architecture answer different questions.

The Real Damage Starts Before the First LNA

Let:

L = linear RF loss before the LNA

Tp = physical temperature of that passive loss

TA = antenna noise temperature

TR = equivalent receiver noise temperature after that loss

Then the aperture-referred system temperature is approximately:

Tsystem = TA + (L − 1)Tp + LTR

This expression exposes the double penalty. The passive structure attenuates the wanted signal. It also generates thermal noise.

It also multiplies the downstream receiver-noise contribution when referred back to the antenna.

Analog Devices explicitly includes pre-DUT loss and its physical temperature when correcting receiver noise temperature. [Ref. 3]

Convert Receiver Noise Figure Into Temperature

Noise factor and equivalent noise temperature are related by:

F = 1 + TR/T₀

where:

T₀ = 290 K

Analog Devices uses this standard 290 K reference and the same relationship between noise factor and equivalent noise temperature. [Ref. 3]

Therefore:

TR = T₀(F − 1)

For NF = 2 dB, we obtain: F = 10^(2/10) ≈ 1.585 and TR ≈ 170 K.

That number becomes extremely important once passive loss precedes the receiver.

The Special Case at 290 K Is Extremely Useful

If Tp = T₀ = 290 K, then:

Tsystem = TA + 290(LF − 1)

Now notice the product, L × F.

In logarithmic terms:

pre-LNA insertion loss and receiver noise figure effectively add.

This creates a powerful architecture insight.

At approximately room temperature, 1 dB less loss before the LNA can have nearly the same receiver-noise value as reducing LNA NF by 1 dB.

This is why feed architecture and semiconductor selection should not be optimized independently.

Friis Explains the Same Physics

For cascaded receiver stages:

Ftotal = F₁ + (F₂ − 1)/G₁ + (F₃ − 1)/(G₁G₂) + …

Keysight identifies noise figure as a critical receiver-sensitivity metric and treats cascade noise as central to receiver characterization. [Ref. 4]

The practical lesson is simple: loss before gain is expensive.

Loss after sufficient low-noise gain becomes much less significant.

A 3 dB Loss Before the LNA Is Not the Same as 3 dB After It

Consider:

LNA NF = 2 dB, and:

LNA gain = 20 dB

Then place 3 dB passive loss before the LNA.

At approximately 290 K, the front-end noise figure becomes approximately: 5 dB.

Now move the same 3 dB attenuator after 20 dB of LNA gain.

Using Friis, its added receiver-NF penalty falls to only a few hundredths of a decibel.

The component did not change, its position in the RF chain changed everything.

Apply This Physics to a Ka-Band SATCOM Study

Pla et al. published a Ka-band phased-array G/T study using [Ref. 1]:

5 dBi unit-cell directivity

0.2 dB mismatch loss

290 K antenna physical temperature

20 K antenna noise temperature, and

2 dB receiver noise figure excluding antenna loss.

Their target was approximately:

G/T = 8 dB/K

These parameters allow the architecture to be reconstructed quantitatively.

Start With No Additional Antenna RF Loss

Using:

TA = 20 K, and

TR ≈ 170 K

we obtain approximately:

Tsystem ≈ 190 K

Now insert only 1 dB of RF loss before the receiver.

The linear loss becomes:

L ≈ 1.259

Therefore:

Tsystem ≈ 20 + (1.259 − 1) × 290 + 1.259 × 170 ≈ 309 K.

The published antenna and receiver-temperature assumptions come directly from Pla et al. [Ref. 1]

One RF Decibel Became More Than One G/T Decibel

Compare 309 K against 190 K.

The G/T penalty associated with noise temperature becomes:

10log₁₀(309/190)

giving approximately: 2.1 dB.

So under this cold-antenna scenario, 1 dB physical RF loss produces roughly 2.1 dB G/T degradation.

That is the hidden penalty. The receive system does not merely lose signal. It becomes noisier.

Why Cold-Sky SATCOM Systems Are Especially Sensitive

The published Ka-band model assumes TA = 20 K. [Ref. 1]

Now place a lossy structure physically near 290 K.

The lossy structure is thermally much hotter than the incoming antenna-noise environment.

That is why small front-end losses can produce disproportionate noise-temperature penalties.

This is one reason low-loss waveguide and antenna-in-package architectures remain attractive in high-performance SATCOM receivers.

At 1.5 dB RF Loss

For: L = 10^(1.5/10) ≈ 1.413,

the same model gives approximately: Tsystem ≈ 379 K.

Compare that with the idealized zero-additional-loss result, 190 K.

The system noise temperature has nearly doubled.

Therefore, 1.5 dB pre-LNA RF loss produces approximately 3 dB G/T degradation under these assumptions.

The underlying antenna and receiver parameters are those reported by Pla et al. [Ref. 1]

That is the kind of result that should influence architecture before PCB layout begins.

RF Loss Can Be Converted Directly Into Element Count

For an ideal equal-element array:

Garray ≈ Gelement + 10log₁₀(N)

Therefore, for fixed G/Ttarget, the required element count is approximately proportional to Tsystem.

Double system temperature, you need roughly double the number of equally performing elements.

This is why receiver loss quickly becomes aperture size.

A Closed-Form Element-Count Estimate

Using the published model:

G/Ttarget = Gelement + 10log₁₀(N) − Lmismatch − 10log₁₀(Tsystem)

Therefore:

N = 10^[(G/Ttarget − Gelement + Lmismatch + 10log₁₀Tsystem)/10]

Using Pla et al.'s [Ref. 1]:

8 dB/K target

5 dBi element directivity, and

0.2 dB mismatch loss,

we can reproduce the published architecture trend.

The Numbers Are Remarkably Consistent

0.5 dB RF loss

Calculated Tsystem ≈ 246 K.,

First-order element requirement, N ≈ 513,

The published paper reports approximately 500 elements.

1.5 dB RF loss

CalculatedTsystem ≈ 379 K,

First-order element requirement, N ≈ 792,

The published benchmark is approximately 800 elements.

3 dB RF loss

Calculated Tsystem ≈ 647 K,

First-order element requirement, N ≈ 1,350,

The published study reports roughly 1,500 elements.

The agreement is strong enough to reveal the governing physics.

The 0.5 dB Versus 3 dB Comparison Is the Real Architectural Shock

Physical loss changes by 2.5 dB, but calculated system temperature changes from approximately 246 K to 647 K.

That corresponds to approximately: 4.2 dB G/T difference, for the same antenna gain.

The source parameters and published element-count comparison come from Pla et al. [Ref. 1]

This explains why only 2.5 dB of RF loss can produce an approximately threefold element-count change.

Published Result: About 1,000 Elements Avoided

Pla et al. conclude that an approximately 8 dB/K target can be achieved with roughly 500 elements using a 0.5 dB-loss waveguide aperture. Their higher-loss comparison requires roughly 1,500 elements at approximately 3 dB RF loss. [Ref. 1]

So the difference is approximately 1,000 radiating elements for the analysed system. That is not an academic detail.

It changes the product.

More Elements Propagate Through the Entire Hardware Architecture

Pla et al. explicitly frame efficiency, aperture complexity, and element count as major limitations in practical SATCOM AESAs. [Ref. 1]

More elements can imply more:

• beamformer channels

• LNAs

• bias networks

• calibration coefficients

• PCB interconnects

• thermal interfaces

• mechanical structure

• digital-control resources

So an electromagnetic loss improvement can become a system cost reduction.

Here Is an Even More Powerful Trade

Use: 1.5 dB RF loss + 2 dB receiver NF. This gives approximately: Tsystem ≈ 379 K.

Now leave the 1.5 dB loss unchanged. Improve receiver NF from 2 dB to 1 dB. The resulting system temperature becomes approximately 246 K.

Now instead keep the original 2 dB receiver NF and reduce the RF loss 1.5 dB → 0.5 dB. The result is again approximately 246 K.

This Creates a Very Useful Engineering Equivalence

Under the 290 K passive-loss assumption, 1 dB less pre-LNA loss ≈ 1 dB better receiver NF, from the perspective of the active receiver-noise term.

So before paying for a substantially better semiconductor process, ask whether the same benefit can be obtained electromagnetically;

Could you shorten the feed?

Remove a switch?

Improve a transition?

Move the LNA?

Change the antenna architecture?

The Low-Loss Element Still Had to Scan

Pla et al. proposed a tri-ridge waveguide aperture specifically to improve radiation efficiency while preserving wide scanning and circular polarization. The structure was fabricated monolithically using metal additive manufacturing.

The experimentally estimated radiating-element loss was below approximately 0.2 dB. That is the kind of loss improvement that changes G/T architecture.

The fabricated demonstrator operates approximately from 10.7 to 12.7 GHz. [Ref. 1]

The LEO scenario considers an Earth field of view around ±60° and user coverage around ±50°.

Cross-polar discrimination is approximately 15 dB across most user-coverage angles.

So this is not merely a low-loss broadside radiator.

Wide-angle performance remains part of the solution.

Key Point: Broadside G/T Is Still Incomplete

ITU defines usable G/T as including pointing and polarization mismatch. It also includes the increase in system noise temperature associated with atmospheric attenuation. [Ref. 2]

That means:

G/Tscan ≠ simply Gboresight − scan loss

The T term can change as well.

This is especially important for LEO terminals operating near low elevations.

G/T Must Therefore Become a Function

A credible terminal should be analysed as:

G/T = f(frequency, azimuth, elevation, polarization, temperature)

not simply:

G/T = one broadside number

This is the receive-side equivalent of reporting EIRP across scan.

Current ESA Targets Show How Serious This Has Become

ESA's ongoing SPADE-25 project targets a Ka-band phased-array demonstrator for LEO/MEO multi constellation operation. [Ref. 5]

Its target receive G/T is 18 dB/K at 19.45 GHz. The project targets up to four full-duplex beams. Its intended elevation range is approximately 10° to 90°.

ESA lists power dissipation and calibration among the programme challenges. That is exactly why receive loss, beam count, scan performance, power, and calibration cannot be optimized independently.

A 1,024-Element Commercial SATCOM Benchmark

A published K-band commercial terminal uses 1,024 array elements. It achieves approximately 10.2 dB/K broadside G/T between 19.5 and 20.2 GHz. [Ref. 6]

It supports approximately ±70° scanning in all planes without grating lobes. It also reports more than 20 dB cross-polar discrimination across scan angles.

That is a proper receive-array benchmark. Not merely element count. Not merely LNA NF. Not merely broadside gain.

Full-Duplex Integration Can Consume Receive Performance

A published hybrid K/Ka phased-array module contains 64 transmit elements and 32 receive elements. [Ref. 7]

Its receive-only G/T is approximately −10.2 dB/K.

In full-duplex operation, G/T falls to approximately −12.2 dB/K.

That is an approximately 2 dB system-mode penalty without changing the number of receive elements.

This demonstrates another critical truth.

Receiver architecture can consume G/T even when the aperture stays unchanged.

This Makes TX-RX Isolation a Receive-Sensitivity Problem

The same K/Ka module uses self-diplexing transitions and waveguide distribution to provide transmit-receive isolation.

It also uses an aluminium rectangular-waveguide feed network to address compact packaging and thermal design. [Ref. 7]

Every isolation structure creates an RF trade.

More isolation can help receiver protection.

More insertion loss can hurt G/T.

That is why shared-aperture engineering is a multidimensional optimization problem.

Reference Plane Discipline Is Essential

ITU explicitly states that noise-temperature analysis requires a defined reference plane. It notes that the LNB input is a typical earth-station reference-plane location. [Ref. 2]

Move the reference plane and the bookkeeping changes.

The underlying physical receiver does not.

So every G/T claim should state where the reference plane sits.

Physical Temperature Belongs in the Loss Budget

Analog Devices explicitly includes the physical temperature of pre-DUT loss when correcting noise temperature. [Ref. 3]

Therefore, 1 dB of loss at 300 K is not equivalent to 1 dB at 80 K from a thermal-noise perspective.

That becomes important in:

• spacecraft receivers

• cryogenic systems

• high-temperature arrays

• thermally nonuniform AESAs

RF loss and thermal design are therefore coupled.

The System-Level Trade Should Be Done in Kelvin First

A useful workflow is:

antenna noise temperature

→ passive front-end loss

→ LNA equivalent noise temperature

→ cascade temperature

→ G/T

→ required element count

Only after that should the design return to dB.

This avoids hiding physical mechanisms inside one noise-figure number.

HFSS and CST Should Feed the G/T Model

Full-wave extraction should quantify actual loss from:

• radiator conductors

• Rogers substrates

• copper roughness

• vias

• launches

• transitions

• package interfaces

• waveguide structures

Those losses should then enter the receiver noise-temperature model.

The EM solver should not stop at S₁₁ and realized gain.

Its loss data should influence the system link budget.

Altium and KiCad Geometry Must Match the EM Model

A schematic may say 0.5 dB feed loss.

The fabricated PCB may produce 0.9 dB, because of:

• longer routing

• additional vias

• different copper roughness

• altered stack-up

• connector transitions

At Ka-band, that discrepancy can materially change G/T.

So the RF layout database becomes part of the noise-temperature model.

The Most Important Trade Study Is a Contour Plot

Before architecture freeze, sweep RF loss against receiver NF and element count for constant G/T.

The Pla et al. study demonstrates how revealing this comparison becomes. [Ref. 1]

Then add contours for:

• DC power

• aperture area

• thermal dissipation

• cost

Now the electromagnetic optimization becomes a product decision.

Stop Asking Only How to Add More Elements

More elements certainly increase gain. But they also add architecture.

Sometimes that is justified. Sometimes it is compensating for avoidable RF loss.

The distinction must be quantified before scaling the aperture.

The Summary Takeaway

The receive figure of merit is:

G/T = Grealised − 10log₁₀(Tsystem)

For a passive RF path preceding the receiver:

Tsystem = TA + (L−1)Tp + LTR

Using the published Pla et al. Ka-band assumptions:

The target 8 dB/K G/T required approximately 500 elements at 0.5 dB RF loss. [Ref. 1]

The same target required approximately 800 elements at 1.5 dB RF loss.

At approximately 3 dB RF loss, the requirement rises toward 1,500 elements.

That means:

2.5 dB of front-end-loss improvement can potentially avoid roughly 1,000 array elements in that architecture.

The deeper physics is even more important:

1 dB before the LNA can cost substantially more than 1 dB of G/T.

So when a SATCOM phased array misses its receive target, the first question should not automatically be:

“How many elements must we add?”

Ask instead:

“Where are we losing the signal before the first low-noise gain?”

Because in advanced receive arrays:

The most valuable beamformer channel may be the one low-loss RF design allowed you never to install.