HXN50254
Product Specifications
HXN50254
Hisiwell — Discrete & Power Semiconductors
Product Overview
The HXN50254 from Hisiwell is a discrete power semiconductor delivering efficient switching and power conversion. Hisiwell is a recognized semiconductor supplier providing reliable components for global electronics manufacturing.
Superb Automation partners with leading brands including Qorvo, TI, ADI, XILINX, NVIDIA, Intel, ST, MACOM, Renesas, and Coilcraft to provide complete BOM and supply-chain solutions.
Key Features
- Low Rds(on) for minimal conduction losses in power switching
- Fast switching speed with low gate charge for high-frequency operation
- Avalanche-rated for robust operation under inductive load conditions
- High junction temperature rating for demanding environments
- Logic-level gate drive compatible for simplified design
- RoHS and AEC-Q101 qualified options available
Key Specifications
| Vds / Vce | V |
| Id / Ic | A |
| Rds(on) / Vce(sat) | mΩ |
| Qg | nC |
| Technology | — |
| Package | — |
| Configuration | — |
| Mounting | — |
Applications
Request Quote — In Stock, 100% New & Original, Full Warranty
HXN50254: sixty watts peak on 0.7 to 2.7 GHz and a band that the two documents disagree on
The HXN50254 is the middle member of the Hisiwell 0.7 to 2.7 GHz group, an unmatched gallium nitride HEMT on a 28 V rail. The product prose calls it a 60 W part for the 700 to 3000 MHz window and marks it especially for LTE, LTE-A and LTE-U. The family index prints 60 W in the peak column and closes the window at 2.7 GHz. The top of the band is therefore 2700 MHz in one document and 3000 MHz in the other, and that difference is 10 per cent of the band.
Everything else in the row is blank in the same way as its two siblings: no saturation figure, no gain, no efficiency, no package, no thermal resistance. The mode column reads continuous wave while the only filled power column is the peak column, which is the same contradiction carried by the whole group.
Sixty watts is 47.78 dBm. No gain is published, so the drive cannot be derived from a vendor figure. At a typical 15 to 18 dB for this class the drive sits between 0.6 W and 2.7 W. The heat at 45 per cent drain efficiency is 73.3 W, at 50 per cent it is 60.0 W, and at 55 per cent it is 49.1 W, with a drain current of 4.76 A, 4.29 A and 3.90 A respectively at 28 V. The spread is 24 W of heat between the best and the worst assumption, which is the difference between a fan cooled extrusion and a liquid plate.
The step from the 40 W member of the group is 1.76 dB. The step up to the 120 W member is 3.01 dB. Those are the numbers a level plan uses when it decides which of the three to buy, and they are the only rows on which the three can be compared.
The parameter rows behind the HXN50254
| Part class | Unmatched gallium nitride HEMT power transistor |
| Frequency range, index row | 0.7 to 2.7 GHz |
| Frequency range, product prose | 700 to 3000 MHz |
| Band edge difference, calculated | 300 MHz, which is 11 per cent of the index window |
| Power, index row | 60 W, published in the peak column |
| Power, product prose | 60 W |
| Power in dBm, calculated | 47.78 dBm |
| Step above the 40 W member, calculated | 1.76 dB |
| Step below the 120 W member, calculated | 3.01 dB |
| Saturation output | Not published |
| Power gain | Not published |
| Drive power | Not published, cannot be derived |
| Drain efficiency | Not published |
| DC input at 45 per cent, calculated | 133.3 W, 4.76 A, 73.3 W of heat |
| DC input at 50 per cent, calculated | 120.0 W, 4.29 A, 60.0 W of heat |
| DC input at 55 per cent, calculated | 109.1 W, 3.90 A, 49.1 W of heat |
| Drain rail | 28 V |
| Operating mode, index row | Continuous wave |
| Load voltage at 50 ohm, calculated | 54.8 V RMS, 77.5 V peak |
| Matching | Unmatched, external network on the customer board |
| Body and package | Not published |
| Thermal resistance, junction to case | Not published |
| Input and output impedance | Not published in text |
| Gate bias point and quiescent current | Not published |
| Orderable addendum, marking and packing | Not published |
Four amps on the drain and seventy three watts in the sink at the worst case
The worst case assumption is the one that matters here. At 45 per cent efficiency the part leaves 73.3 W of heat in a package that the vendor does not describe, and 73 W is past the point where a small extrusion works with natural convection. At 55 per cent the same output leaves 49.1 W, which a well finned extrusion can hold at a reasonable rise. The 24 W between the two figures is the entire thermal budget of the project, and the only way to close it is a measurement.
Four point three amps at 28 V is still an easy DC problem. A 150 mil trace at one ounce carries it, and the choke carries the full current with a saturation rating above five amps. The bulk capacitance on the rail wants a low ESR bank sized for the pulse current of the envelope rather than for the average current.
The matching network is where the effort sits. Two gigahertz of band at 60 W needs a broadband transformer on the output and a feedback network on the input, and the output transformer has to handle 77.5 V peak at 60 W and about 110 V peak under a two to one mismatch. That is a ferrite loaded transmission line transformer or a broadside coupled structure on a high frequency laminate, not a lumped network on FR-4.
The heatsink interface follows from the package. With no published junction to case resistance and no published body outline, the first article is also the first thermal measurement, and the case temperature at rated output is the only figure that closes the loop. A vendor that publishes no package drawing cannot be held to a thermal number, so the measurement is the contract.
HXN50254 against the 0.7 to 2.7 GHz group and the wider shelf
| HXN50253 | 0.7 to 2.7 GHz | 40 W peak | Not published | Not published | 1.76 dB below |
| HXN50254 | 0.7 to 2.7 GHz | 60 W peak | Not published | Not published | Reference point |
| HXN50256 | 0.7 to 2.7 GHz | 120 W peak | Not published | Not published | 3.01 dB above |
| HX1020120P2 | 1 to 2 GHz | 140 W | 20 dB | 80 per cent | 3.68 dB above, narrower band |
| HX446030F2 | 4.4 to 6 GHz | 30 W | 14 dB | 55 per cent | 3.01 dB below, different band |
| HX2040185F | 2 to 4 GHz | 185 W | 18 dB | 70 per cent | 4.89 dB above, starts at 2 GHz |
The three 0.7 to 2.7 GHz members are separated by 1.76 dB and 3.01 dB and by nothing else that the vendor publishes. A customer choosing between them is choosing a power level and a price, and a customer who needs gain or efficiency numbers needs a different shelf.
The narrow band HX1020120P2 gives 3.7 dB more output plus a published 20 dB of gain on one gigahertz of band. If the product only needs 1 to 2 GHz, that part is cheaper to build because the match is narrower, and the published gain removes one round of measurement.
For a multi band product the group is bought together, and the one stop RF supply page covers a kit of three stages from one vendor. When one member is short, the obsolete RF part replacement page covers what can be substituted and what has to be re-qualified rather than substituted.
Board notes and the band edge question
The band edge question comes first because it costs money. If the product has to work to 3 GHz, the index row says the part is not guaranteed there and the prose says it is. The honest route is a measurement at 2.8 and 3.0 GHz on the first article, taken before the board is released rather than after, because a match that is correct at 2.7 GHz can be a stub away from correct at 3 GHz.
The bias sequence is the second item. No gate voltage and no quiescent current is published, so the operating point is set on the bench and the factory confirms the window for the lot. Negative gate before drain, drain off before gate, clamp at the gate, and a bleeder on the gate line so the part cannot be left with the gate floating.
The match is the third item. With no scattering parameters in text, the input network comes from a measured set or a load pull. The output network has to carry 77.5 V peak matched and about 110 V peak under a two to one mismatch, and the capacitors in that network have to be rated for the second number.
The fourth item is the DC feed. The choke carries 4.3 A and has to present a high impedance across the band, which at 700 MHz means a substantial inductance and at 2.7 GHz means a low self capacitance. Two chokes in series with a capacitor between them is the usual answer, and the resonance of that pair sits inside the band and has to be checked.
Questions buyers ask about HXN50254
Does it cover 3 GHz? The index row stops at 2.7 GHz and the prose says 3000 MHz. We quote the band as 0.7 to 2.7 GHz and state the prose figure, and a 3 GHz requirement is confirmed in writing with the factory before a schedule is set.
What is the gain? Not published, so the drive is measured rather than read. A line up sized on an assumed 15 dB should carry one more stage of headroom at the top of the band.
Is 60 W peak or average? It sits in the peak column on a row marked continuous wave. On a continuous wave carrier the two are the same; on a modulated signal the crest factor has to be added before the figure is compared with an average requirement.
Do you supply a reference match? That is the question we put to the factory on every order for this group, because the vendor publishes no impedance data and a match built from a load pull is a different schedule from a match built from a supplied set.
How do we start? Send the band edges including the top edge as the product needs it, the output at the connector, the signal and its crest factor, the ambient and the rail. We will confirm the band and the gain in writing, quote from the current schedule and ship a sample measured at both edges.