A phased array can meet its gain, sidelobe, and scan-angle requirements on a bench, then lose margin in service because junction temperatures rise unevenly across the aperture. The resulting change in amplifier gain, phase-shifter loss, bias behavior, and receiver noise figure is not merely a packaging issue. It is an electromagnetic performance issue with direct consequences for EIRP, beam shape, calibration stability, reliability, and qualification risk.

Key Takeaways

  • Thermal design starts with a credible RF and DC power budget for every active stage, not one average dissipation value. A power amplifier delivering 10 W with 10 dB gain at 35% power-added efficiency draws 25.7 W of DC power and turns 16.7 W into heat.
  • Junction temperature follows the series thermal path from junction to coolant: Tj = Tref + Pdiss Rθ. In the worked example, 3.80 °C/W puts that 16.7 W amplifier near 104 °C on 40 °C coolant.
  • The thermal interface is often the least controlled layer. A voided 250 µm bond line in place of a 100 µm layer raises the same junction by about 15 °C, and module-to-module differences like this become element gain and phase errors.
  • Uniform heating mostly costs EIRP and a common phase shift that calibration can remove. Uneven heating raises the sidelobe floor: in a 256-element array, 10° RMS phase error sets an average error sidelobe level near −38 dB.
  • Validate with instrumented tests at representative scan states, duty cycles and enclosure conditions, measuring DC power, RF output, gain, phase and temperature together.

Why does thermal management determine phased-array performance?

Thermal management determines phased-array performance because active-device temperature changes the electrical behavior of the array's individual transmit/receive modules. In transmit mode, power amplifiers generally exhibit reduced output power and efficiency as junction temperature increases. In receive mode, low-noise amplifier performance, gain distribution, and compression margin can shift. Phase shifters, vector modulators, attenuators, and frequency-conversion stages also have temperature coefficients that can disturb a previously calibrated beam.

The effect is spatial as well as temporal. Center modules may run hotter than edge modules, while modules near power-entry regions, mechanical seams, or restricted airflow may have different thermal environments. A common temperature rise can often be corrected with global calibration. A temperature gradient across the aperture produces differential errors that change array factor and can elevate sidelobes or shift the main beam.

For a first-order estimate, the junction temperature is the reference temperature plus the dissipated power times the thermal resistance of each layer in series:

Tj = Tref + Pdiss Rθ,j-ref

Rθ,j-ref = Rjc + Rcarrier + RTIM + Rconv

RTIM = t / (k A),  Rconv = 1 / (h Aw)

ΔTcoolant = Ptotal / (ṁ cp)

Where:

  • Tj = junction (or channel) temperature (°C); Tref = reference temperature: case, board, cold plate, coolant or ambient (°C)
  • Pdiss = power dissipated as heat in the device (W)
  • Rθ,j-ref = total thermal resistance from junction to the reference (°C/W)
  • Rjc = junction-to-case resistance from the device data sheet; Rcarrier = die attach, carrier and cold-plate wall (°C/W)
  • RTIM = thermal interface resistance; t = bond-line thickness (m); k = effective conductivity (W/m·K); A = contact area (m²)
  • Rconv = convection resistance; h = heat-transfer coefficient (W/m²·K); Aw = wetted or finned area (m²)
  • ΔTcoolant = coolant inlet-to-outlet rise (°C); Ptotal = heat into the coolant (W); ṁ = mass flow (kg/s); cp = specific heat (J/kg·K)

Here, Tref may be case, board, cold-plate, or ambient temperature, depending on the thermal-resistance definition, and the JEDEC JESD51 series defines how those reference points are measured. This expression is useful for budget development, but it cannot substitute for a full path analysis when spreading resistance, transient load, interface variation, and airflow or coolant distribution are significant.

Worked example: junction temperature through the thermal stack

One transmit amplifier dissipates 16.7 W (the 10 W, 35% PAE amplifier worked out in the next section) and is mounted on a liquid cold plate with 40 °C coolant. All layer values below are assumed, illustrative numbers; use data-sheet and measured values for real hardware.

Stacked bar chart of temperature rise from 40 degree C coolant to the junction of a 16.7 W phased-array power amplifier for three cases: liquid cold plate with a good thermal interface reaching 104 degrees C, liquid cold plate with a voided interface reaching 119 degrees C, and finned forced air reaching 117 degrees C, with the 50.1 degree junction-to-case rise dominating each bar
Figure 1. Temperature rise through each layer of the thermal path for one 16.7 W amplifier. A voided thermal interface adds about 15 °C at the junction, and replacing the liquid cold plate with finned forced air adds about 13 °C. Assumed, illustrative values.
  1. Junction to case: Rjc = 3.0 °C/W, so the rise is 16.7 × 3.0 = 50.1 °C. The device itself is the largest single resistance, which is why device selection and die attach are thermal decisions.
  2. Carrier, attach and cold-plate wall: 0.35 °C/W, a 5.8 °C rise.
  3. Thermal interface: t = 100 µm, k = 3 W/m·K over a 12 × 12 mm footprint gives RTIM = 100 × 10⁻⁶ / (3 × 1.44 × 10⁻⁴) = 0.23 °C/W, a 3.9 °C rise.
  4. Convection: h = 5,000 W/m²·K over a 30 × 30 mm wetted area gives Rconv = 1 / (5,000 × 9 × 10⁻⁴) = 0.22 °C/W, a 3.7 °C rise.
  5. Total: Rθ = 3.80 °C/W and Tj = 40 + 16.7 × 3.80 ≈ 104 °C.
  6. Voided interface: if the bond line grows to 250 µm and voids lower the effective conductivity to 1.5 W/m·K, RTIM = 1.16 °C/W, the total becomes 4.73 °C/W and Tj ≈ 119 °C. That one module now runs about 15 °C hotter than its neighbors.
  7. Coolant rise: 64 modules at about 19.7 W each (16.7 W from the amplifier plus an assumed 3 W for driver, control and DC-DC losses) put 1.26 kW into the coolant during continuous transmit. At 4 L/min of water-glycol (ρ ≈ 1,060 kg/m³, cp ≈ 3,500 J/kg·K), ṁ = 0.0707 kg/s and ΔTcoolant = 1,262 / (0.0707 × 3,500) ≈ 5.1 °C. Modules at the outlet end run about 5 °C hotter than those at the inlet, a built-in gradient across the aperture.
Thermal stack for one 16.7 W phased-array amplifier on a liquid cold plate (assumed, illustrative values)
LayerAssumed valuesRθ (°C/W)Rise at 16.7 W (°C)
Junction to caseDevice data-sheet value3.0050.1
Attach, carrier and cold-plate wallSolder attach, metal carrier0.355.8
Thermal interface, good100 µm, k = 3 W/m·K, 12 × 12 mm0.233.9
Thermal interface, voided250 µm, k = 1.5 W/m·K effective1.1619.3
Convection to coolanth = 5,000 W/m²·K, 30 × 30 mm wetted0.223.7
Total, good interface40 °C coolant3.8063.6 (Tj ≈ 104 °C)
Total, voided interface40 °C coolant4.7379.1 (Tj ≈ 119 °C)

Calculated by SiberSci from R = t/(kA) and R = 1/(hA) with the assumed values shown. Series-resistance method per Bergman, Lavine, Incropera and DeWitt, Fundamentals of Heat and Mass Transfer; spreading resistance is neglected (see References).

Typical convection coefficients and what they mean for a 30 × 30 mm module footprint
Cooling methodTypical h (W/m²·K)Rconv over 9 cm², no fins (°C/W)Typical phased-array use
Free convection, gases2–25556–44Low-power receive panels, with large finned or chassis area
Forced convection, gases25–25044–4.4Fan-cooled arrays; practical only with fins that multiply the area
Free convection, liquids50–1,00022–1.1Rare as the primary path in arrays
Forced convection, liquids100–20,00011–0.056Cold plates for dense transmit apertures
Boiling or condensation2,500–100,0000.44–0.011Two-phase loops, heat pipes and vapor chambers

h ranges from Bergman, Lavine, Incropera and DeWitt, Fundamentals of Heat and Mass Transfer, Table 1.1; Rconv = 1/(hA) calculated by SiberSci for A = 9 cm². The finned forced-air case in Figure 1 assumes h = 100 W/m²·K over about 100 cm² of fin area (see References).

What heat sources belong in a phased array thermal budget?

The thermal budget must include all dissipative states, with the worst credible combination of transmit duty cycle, scan angle, bandwidth, supply voltage, and ambient condition. It is inadequate to multiply nominal PA output power by a catalog efficiency number and call the result the array heat load.

Phased-array heat sources, why each is underestimated, and the design implication
Heat sourceWhy it is commonly underestimatedDesign implication
Power amplifiersEfficiency changes with output backoff, mismatch, pulse conditions, and temperatureEvaluate dissipated power at required EIRP and load mismatch limits
Driver and gain stagesLower individual power can conceal substantial cumulative aperture dissipationInclude every active stage in each module
Phase shifters, attenuators, and control ICsDC bias and insertion loss become heat, especially in dense tile architecturesAccount for mode-dependent control states
DC-DC converters and bias networksConversion losses and copper losses concentrate near power entrySeparate power-distribution hot spots from RF hot spots
Digital processing and data linksHigh-speed serializers, FPGAs, and clocking circuits can dominate in receive arraysModel their thermal coupling to sensitive RF circuits
Radome and solar loadingExternal heating raises the starting temperature for the entire thermal stackDefine environmental boundary conditions explicitly

SiberSci engineering summary, consistent with the transmit/receive module treatment in Brown, Active Electronically Scanned Arrays, and the power-budget framework in Pozar, Microwave Engineering (see References).

Transmit array calculations should distinguish RF output power from DC input power. For example, a PA delivering 10 W RF at 35% power-added efficiency dissipates considerably more than a 10 W thermal load, once quiescent power and driver contributions are included. The appropriate calculation depends on the PA architecture and how PAE, drain efficiency, and module-level conversion efficiency are specified. The worked example below shows the arithmetic for the power-added-efficiency definition.

How much heat does a 10 W phased-array amplifier produce?

A power amplifier turns DC power into RF power, and everything that does not leave as RF output becomes heat. With power-added efficiency defined in the usual way, the dissipation follows directly from the output power, the gain and the efficiency:

PAE = (Pout − Pin) / PDC

Pdiss = PDC + Pin − Pout = (Pout − Pin)(1/PAE − 1)

Pavg = D Pdiss,on + (1 − D) Pidle

Where:

  • PAE = power-added efficiency (fraction); Pout = RF output power (W); Pin = RF input drive power (W)
  • PDC = DC supply power drawn by the amplifier (W); Pdiss = heat dissipated in the amplifier (W)
  • D = transmit duty cycle (fraction); Pdiss,on = dissipation while transmitting (W); Pidle = quiescent dissipation between pulses (W); Pavg = time-averaged dissipation (W)

Worked example: one transmit channel. The amplifier delivers 10 W with 10 dB gain at 35% PAE.

  1. Drive power: 10 dB gain means Pin = 10 / 10 = 1 W.
  2. DC power: PDC = (10 − 1) / 0.35 = 25.7 W.
  3. Heat: Pdiss = 25.7 + 1 − 10 = 16.7 W, which is 1.67 W of heat for every watt of RF output.
  4. Pulsed operation: at 20% duty with 1 W of quiescent dissipation between pulses, Pavg = 0.2 × 16.7 + 0.8 × 1 ≈ 4.1 W. The average sets the cold-plate and coolant load, but the junction can still follow the peak when pulses are long compared with the die's thermal time constant, so check pulsed cases with transient thermal impedance data.
  5. Aperture total: adding an assumed 3 W per module for driver, control and DC-DC losses gives about 19.7 W per module, or about 5.0 kW for a 256-element array in continuous transmit.
DC power and heat for a 10 W, 10 dB gain amplifier at different power-added efficiencies
PAE (%)DC power PDC (W)Heat Pdiss (W)Heat per watt of RF output
2536.027.02.70
3525.716.71.67
4520.011.01.10
5516.47.40.74
6513.84.80.48

Calculated by SiberSci from PDC = (Pout − Pin)/PAE and Pdiss = PDC + Pin − Pout with Pout = 10 W and Pin = 1 W. PAE definition per Pozar, Microwave Engineering (see References). Efficiency falls in backoff and under load mismatch, so evaluate the operating point, not the data-sheet peak.

Scan state also changes the picture. Beam steering can alter element excitation, supply loading, active impedance, and mismatch. In tightly coupled arrays, scan-dependent embedded element impedance can change PA loading enough to affect both RF output and heat dissipation. This is one reason electromagnetic and thermal models should not be developed as independent compliance exercises.

How should a phased array thermal path be designed?

A phased array thermal path should be designed as a series of measurable interfaces from semiconductor junction to the final heat sink. The lowest apparent resistance in one layer cannot compensate for a poor interface, inadequate spreading plane, or coolant plate with a large temperature gradient.

At module level, the package attach and thermal ground structure should provide a low-resistance route without compromising RF grounding or introducing mechanical stress. Via fields beneath high-power devices, metal-backed substrates, copper-molybdenum carriers, vapor chambers, and heat spreaders can all be appropriate. The correct choice depends on frequency, module pitch, allowable mass, CTE compatibility, manufacturing method, and repairability.

At the array level, liquid cooling offers high heat-flux capability and tighter control of plate temperature, but it imposes pump, plumbing, corrosion, leak, maintenance, and qualification requirements. Forced-air cooling may be simpler and lighter for moderate heat loads, yet it is sensitive to altitude, contamination, fan reliability, recirculation, and nonuniform flow. Conduction to a chassis may be attractive in sealed equipment, but chassis temperature must then be treated as a real boundary condition rather than an assumed heat sink.

Thermal interface materials deserve particular scrutiny. Their published bulk conductivity does not directly establish assembly performance. Bond-line thickness, surface flatness, mounting pressure, pump-out, dry-out, voiding, and aging can dominate contact resistance. A design that depends on an unusually thin or tightly controlled interface layer should state that requirement in production drawings and verify it through process controls. The thermal-stack worked example shows how much one voided interface can raise a junction temperature.

Which analyses reveal coupled RF and thermal problems?

Coupled RF and thermal analysis reveals problems that separate simulations often miss. Begin with an electrical power map that identifies dissipation for each module and operational state. Apply that map to a three-dimensional conduction and fluid model, then feed predicted temperatures back into device models, gain-and-phase data, and calibration algorithms.

The model should report more than maximum junction temperature. Useful outputs include module-to-module temperature spread, cold-plate inlet-to-outlet rise, board warpage risk, interface pressure distribution, and transient response after pulse or mode changes. If the array is installed behind a radome or in an enclosure, include the relevant radiative and convective boundary conditions. Solar loading, conductive mounting paths, and neighboring electronics can materially change results.

For RF consequences, use measured or modeled temperature coefficients for PA output power, small-signal gain, phase, noise figure, and control-circuit behavior. A practical engineering question is not simply, “Will the device survive?” It is, “At the specified temperature distribution, does the array still satisfy EIRP, beam-pointing, sidelobe, spurious, and receive-sensitivity requirements?”

How much do temperature gradients degrade the beam?

Temperature gradients degrade the beam by giving each element its own gain and phase error. A uniform temperature rise shifts every element by the same amount, so it mainly lowers EIRP and adds a common phase that does not steer the beam. A linear gradient tilts the phase front and moves the beam slightly. A random spread from interface variation, coolant rise and local hot spots adds an error sidelobe floor that limits how low the sidelobes can be, regardless of the designed taper.

Δφn = αφ ΔTn,  ΔGn = αG ΔTn

Δθ ≈ Φ λ / [2π (Nx − 1) d cos θ0]

σ² ≈ [(1 − P) + δ² + P φ²] / (P η N)

Where:

  • Δφn, ΔGn = phase (degrees) and gain (dB) change of element n; ΔTn = its temperature change (°C)
  • αφ = phase temperature coefficient of the element's RF chain (°/°C); αG = gain temperature coefficient (dB/°C)
  • Δθ = beam-pointing shift (rad); Φ = end-to-end phase tilt across the aperture from a linear gradient (rad)
  • λ = free-space wavelength (m); Nx = elements along the gradient; d = element spacing (m); θ0 = scan angle
  • σ² = average error sidelobe level relative to the main-beam peak (convert with 10 log10)
  • P = fraction of elements working; δ = RMS fractional amplitude error; φ = RMS phase error (rad)
  • η = aperture taper efficiency (1.0 for uniform illumination, lower for low-sidelobe tapers); N = total number of elements

Worked example: a 16 × 16 array at 10 GHz. The array has 256 elements at d = 15 mm (half a wavelength) and a 0.8 taper efficiency. The temperature coefficients, αφ = 0.3°/°C and αG = −0.015 dB/°C, are assumed for illustration; real values depend on the device, frequency and chain, and should be measured.

  1. Uniform 20 °C rise: every element shifts 6° and loses 0.3 dB. The common phase does not move the beam, and a temperature-indexed calibration table can remove it. Transmit EIRP falls by about 0.3 dB unless the amplifiers have headroom to restore it.
  2. Linear 15 °C gradient across the 16 columns (for example, along the coolant flow): the phase tilt is Φ = 4.5°, and Δθ ≈ 0.0785 × 0.03 / (2π × 15 × 0.015) ≈ 0.10° at broadside. That is small against a beamwidth of about 6.3°, but it matters for precision pointing and monopulse tracking.
  3. Random 10 °C RMS spread: φ = 3° RMS and δ ≈ 0.15 dB RMS give σ² ≈ −48 dB, comfortably below a −40 dB sidelobe requirement.
  4. The same spread with a chain coefficient of 1°/°C gives 10° RMS phase error and σ² ≈ −38 dB, which no longer meets a −40 dB requirement. The coefficient matters as much as the gradient.
  5. Failed elements: if 2% of the 256 elements fail from overtemperature (P = 0.98), the failures alone set σ² ≈ −40 dB.
Chart of average error sidelobe level in dB versus RMS element phase error from 1 to 20 degrees for phased arrays of 64, 256 and 1,024 elements with 0.8 taper efficiency, with minus 30 and minus 40 dB requirement lines, marking minus 48.7 dB at 3 degrees and minus 38.3 dB at 10 degrees for 256 elements
Figure 2. Average error sidelobe level from random phase error alone, φ²/(ηN), for three array sizes. For 256 elements, 3° RMS (10 °C spread at 0.3°/°C) gives −48.7 dB; 10° RMS (10 °C at 1.0°/°C) gives −38.3 dB. Peak sidelobes sit several dB above this average floor.
Average error sidelobe level (dB) versus RMS phase error and number of elements (η = 0.8, phase error only)
RMS phase error (°)N = 64N = 256N = 1,024
2−46.2−52.3−58.3
3−42.7−48.7−54.8
5−38.3−44.3−50.3
10−32.3−38.3−44.3
15−28.7−34.8−40.8
20−26.2−32.3−38.3

Calculated by SiberSci from σ² ≈ φ²/(ηN) with P = 1, δ = 0 and η = 0.8, the random-error expression in Mailloux, Phased Array Antenna Handbook, Chapter 7 (see References). This is an average level; peak sidelobes are higher, and correlated errors, such as a gradient repeated on every tile, behave differently from random errors.

The table shows why larger apertures tolerate more error: the error floor drops about 6 dB each time the element count quadruples. It also shows why small subarrays and tiles, which are often tested on their own, can fail a sidelobe test that the full array would pass.

Pozar's Microwave Engineering provides the circuit and power framework needed to establish realistic RF operating conditions, while Balanis's Antenna Theory addresses array behavior and pattern sensitivity. Mailloux’s Phased Array Antenna Handbook quantifies how random amplitude and phase errors raise the sidelobe floor. For heat-transfer fundamentals, Bergman, Lavine, Incropera, and DeWitt cover conduction, convection, and contact resistance, and for microwave-heating context, Metaxas and Meredith provide useful treatment of electromagnetic power deposition and temperature-dependent material behavior. Relevant IEEE publications on active electronically scanned arrays and electronic packaging can support detailed architecture-specific assumptions.

How should engineers validate the thermal design?

Validation should measure the same conditions that the product must survive and perform under. Infrared imaging is useful for locating unexpected hot regions, but emissivity uncertainty, reflected radiation, and obstructed views limit its use as the sole temperature measurement method. Calibrated thermocouples, RTDs, embedded sensors, and device telemetry provide complementary data.

A disciplined test sequence starts with low-power functional checks, then proceeds to representative transmit and receive states, environmental temperature extremes, thermal soak, and repeated cycling. Measure DC power, RF output, gain, phase, array pattern metrics, coolant or air temperatures, and module temperatures together. Correlating these measurements makes it possible to distinguish a thermal bottleneck from a control, bias, mismatch, or calibration defect.

Particular attention is warranted after assembly changes. A revised torque specification, different interface material batch, altered solder profile, or substituted cold plate can shift thermal performance without producing an obvious RF failure during room-temperature acceptance testing. Production thermal verification therefore belongs in the manufacturing plan, not only in prototype qualification.

What design decisions create the best thermal margin?

The best thermal margin comes from allocating heat early, reducing dissipation where practical, and preserving a predictable heat path through production. Higher PA efficiency can reduce cooling burden, but it may require a different linearity strategy, bias condition, device technology, or operating backoff. A thicker heat spreader can reduce gradients but add mass and complicate module pitch. Liquid cooling can support high aperture power density but may be unacceptable for systems with strict maintenance or contamination limits.

The appropriate design is therefore specification-dependent. Establish allowable junction temperature, temperature spread across the aperture, environmental range, duty cycle, calibration approach, service life, and manufacturing tolerances before selecting cooling hardware. For high-consequence RF systems, thermal requirements should be traceable to measurable RF performance limits rather than treated as a separate mechanical constraint.

FAQ

What is the most useful thermal metric for a phased array?

Maximum junction temperature is necessary, but not sufficient. The most useful set includes peak junction temperature, module-to-module temperature variation, thermal transient response, and the resulting gain-and-phase error across the aperture.

Can calibration correct thermal beam errors?

Calibration can correct repeatable temperature-related errors when sufficient sensing and correction authority exist. It cannot fully compensate for rapid transients, unmeasured gradients, PA compression, or permanent degradation caused by excessive temperature.

When should thermal analysis begin?

It should begin during architecture selection, when module pitch, PA technology, power distribution, enclosure constraints, and cooling concepts can still be changed economically. A defensible thermal model at that stage protects both technical margin and schedule later in development.

How much heat does a phased-array power amplifier dissipate?

Dissipation equals DC input power plus RF input power minus RF output power. An amplifier delivering 10 W with 10 dB gain at 35% power-added efficiency draws about 25.7 W of DC power and dissipates about 16.7 W, so the heat load is larger than the RF output. Drivers, control ICs and DC-DC conversion add to that total.

How do temperature gradients raise sidelobes?

Each element's gain and phase drift with its own temperature. A uniform rise shifts every element equally and mainly lowers EIRP, but uneven temperatures create element-to-element errors that add an average error sidelobe floor of roughly (δ² + φ²)/(ηN) relative to the main beam. For a 256-element array, 10° RMS phase error gives a floor near −38 dB.

Why does the thermal interface material matter so much?

The interface layer is often the least controlled resistance in the thermal path. Its resistance scales with bond-line thickness divided by conductivity, and voids, flatness and mounting pressure can change it several times over. In the worked example, a voided 250 µm layer in place of a 100 µm layer raises one amplifier's junction temperature by about 15 °C.

Is liquid cooling required for a phased array?

Not always. Receive arrays and low-duty transmit arrays can often use conduction to a chassis or finned forced air. Liquid cold plates become attractive when heat flux, module pitch or ambient temperature leave too little margin, because forced liquid convection coefficients are typically 100 to 20,000 W/m²·K, compared with 25 to 250 W/m²·K for forced gas convection.

In a phased array, thermal margin is electromagnetic margin: the temperature map across the aperture sets EIRP, beam pointing and the sidelobe floor as surely as the RF design does. SiberSci provides antenna engineering and microwave engineering support to build coupled RF and thermal budgets, review thermal paths and plan validation tests. Contact Dr. Scott Best to discuss your array.

References

  1. D. M. Pozar, Microwave Engineering, 4th ed. Hoboken, NJ: Wiley, 2012. https://www.wiley.com/en-us/Microwave+Engineering,+4th+Edition-p-9781118213636
  2. C. A. Balanis, Antenna Theory: Analysis and Design, 4th ed. Hoboken, NJ: Wiley, 2016. https://www.wiley.com/en-us/Antenna+Theory:+Analysis+and+Design,+4th+Edition-p-9781118642061
  3. R. J. Mailloux, Phased Array Antenna Handbook, 3rd ed. Norwood, MA: Artech House, 2017. Chapter 7, “Array Error Effects.” https://uk.artechhouse.com/Phased-Array-Antenna-Handbook-Third-Edition-P1909.aspx
  4. T. L. Bergman, A. S. Lavine, F. P. Incropera and D. P. DeWitt, Fundamentals of Heat and Mass Transfer, 7th ed. Hoboken, NJ: Wiley, 2011. https://bcs.wiley.com/he-bcs/Books?action=index&bcsId=6563&itemId=0470501979
  5. A. C. Metaxas and R. J. Meredith, Industrial Microwave Heating, IET Power and Energy Series 4. London: Peter Peregrinus/IET. https://doi.org/10.1049/PBPO004E
  6. JEDEC, JESD51-1: Integrated Circuit Thermal Measurement Method, Electrical Test Method (Single Semiconductor Device), 1995. https://www.jedec.org/standards-documents/docs/jesd-51-1
  7. Y. Aslan, J. Puskely, J. H. J. Janssen, M. Geurts, A. Roederer and A. Yarovoy, “Thermal-Aware Synthesis of 5G Base Station Antenna Arrays: An Overview and a Sparsity-Based Approach,” IEEE Access, vol. 6, pp. 58868–58882, 2018. https://doi.org/10.1109/ACCESS.2018.2873977
  8. A. D. Brown, “Transmit Receive Modules,” in Active Electronically Scanned Arrays: Fundamentals and Applications. Hoboken, NJ: Wiley-IEEE Press, 2022. https://doi.org/10.1002/9781119749097.ch4
author avatar
Dr. Scott Best Founder and Chief Technology Officer
SiberSci, LLC is built upon the professional experience of Dr. Scott Best in both industry and government. This period of professional experience has witnessed dramatic changes in techniques available for solving Electromagnetic Engineering problems, as well as for performing laboratory measurements and experiments. Engineers must continually study new developments in Electromagnetic Engineering, or they risk becoming obsolete in a short period of time. SiberSci, LLC is fluent working with the latest tools available in the industry, and can also use older techniques to provide fast iterations of prototypes. SiberSci, LLC can solve your Electromagnetic Engineering problems for you, so how do they do this? Please continue reading to learn more. SiberSci, LLC provides clients with comprehensive short-term and long-term Electromagnetic Engineering Services that are customized to satisfy their individual technical requirements and time constraints. To determine if SiberSci, LLC can be of assistance to your company, please provide Dr. Scott Best with a detailed description of the work to be performed along with a schedule, and SiberSci, LLC will provide a budgetary quotation for the services that are required to fulfill the stated technical requirements. A Non-Disclosure Agreement (NDA) is readily available to protect the business sensitive information, intellectual property, and proprietary information for each client to ensure their information is protected and not compromised.