HXN40060
Product Specifications
HXN40060
Hisiwell — Discrete & Power Semiconductors
Product Overview
The HXN40060 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
HXN40060: two hundred watts pulsed at 5.0 to 5.3 GHz
The HXN40060 is the pulsed member of the Hisiwell pair that covers 5.0 to 5.3 GHz. The vendor index row reads 5.0 to 5.3 GHz, saturated output at least 53 dBm, gain at least 10 dB, drain efficiency at least 45 per cent, 28 V drain rail, plus or minus 0.8 dB of flatness, and a mode of 100 microsecond pulses at 10 per cent duty. That is 199.53 W of peak radio frequency from one packaged transistor on a 28 V rail.
The peak figures are large and the average figures are not, which is the whole point of a pulsed part. At 45 per cent efficiency the peak DC input is 443.4 W, which is 15.84 A at 28 V, and the peak dissipation is 243.9 W. Averaged over a 10 per cent duty cycle those become 44.3 W of DC input, 1.58 A and 24.4 W of heat, which is a thermal problem of a completely different size.
Two numbers decide whether a system can use it. The first is the drive: with a 10 dB gain floor the input for rated output is 43 dBm, which is 19.95 W. A 20 W driver in front of a 200 W stage is a normal architecture for radar and for high power test equipment, and it is not a normal architecture for a communications link, which is the practical filter on this part.
The second is the pulse energy. Fifteen point eight amps for 100 microseconds is 1.58 millicoulombs of charge per pulse, and the bulk capacitor bank has to deliver it without letting the rail sag into the gain of the stage. That is the constraint that decides the size of the power supply, not the average current.
The parameter rows Hisiwell publishes for HXN40060
| Part class | Gallium nitride power transistor, internally matched |
| Frequency range | 5.0 to 5.3 GHz |
| Bandwidth, calculated | 300 MHz, 5.83 per cent fractional at a 5.15 GHz centre |
| Saturation output power as published | at least 53 dBm |
| Saturation output power in watts, calculated | 199.53 W peak |
| Gain as published | at least 10 dB |
| Drain efficiency as published | at least 45 per cent |
| Drain voltage | 28 V |
| Flatness across the band | plus or minus 0.8 dB |
| Operating mode | pulsed, 100 microsecond pulse width, 10 per cent duty |
| Output at 1 dB compression | not published, the index prints a slash in that column |
| Peak DC input at saturation, calculated | 443.4 W |
| Peak drain current at saturation, calculated | 15.84 A at 28 V |
| Average DC input at 10 per cent duty, calculated | 44.3 W |
| Peak dissipation at saturation, calculated | 243.9 W |
| Average dissipation at 10 per cent duty, calculated | 24.4 W |
| Dissipation if efficiency falls to 40 per cent, peak, calculated | 299.3 W |
| Drive power for the rated output, calculated | 43 dBm, 19.95 W |
| Charge delivered per pulse, calculated | 1.58 mC at 15.84 A for 100 microseconds |
| Bulk capacitance for 1 V of droop, calculated | 1580 microfarad |
| Load voltage at peak output into 50 ohm, calculated | 99.9 V RMS, 141.3 V peak |
| Output per 100 MHz of the band, calculated | 66.51 W |
| Thermal resistance, junction to case | not published |
| Package outline and dimensions | not published in text |
| Scattering parameters | not published |
| Pulse rise and fall time | not published |
| Duty cycle limit above 10 per cent | not published |
| Orderable addendum and packing | not published |
| Neighbour on the same band at 47 dBm, continuous wave | HXN40057, same rail, 11 dB gain |
Why the average dissipation and the peak dissipation are both true
24.4 W of average heat is the number the heatsink sees, and it is a modest load for a flange the size that a 200 W part normally carries. 243.9 W of peak heat is the number the die sees within the pulse, and it is the number that sets the junction temperature swing rather than the junction temperature. The two are not interchangeable and neither one is the whole story.
At 100 microseconds the thermal time constant of the die is short enough that the junction follows the pulse, so the peak junction temperature sits above the average by an amount set by the die thermal capacitance. No transient thermal impedance curve is published, so that swing cannot be calculated from the vendor data. It has to be measured, and the measurement is a pulse run with a calibrated forward voltage reading on the gate or the drain.
What can be calculated is the energy budget. Each pulse delivers 19.95 mJ of radio frequency and dissipates 24.39 mJ of heat, so 44.3 mJ leaves the rail per pulse at a 1 kHz repetition rate. A sink that removes 24.4 W continuously holds the flange steady, and the pulse train then rides on top of a stable base rather than on a drifting one.
The rail is where most first articles go wrong. The 1580 microfarad figure for 1 V of droop assumes the whole charge comes from the capacitor bank and the supply delivers nothing during the pulse, which is the worst case. A real supply with a fast loop does better, and a supply sized on average current alone does much worse, because it lets the rail fall through the whole pulse and the gain falls with it.
The duty cycle limit and what happens above it
The vendor publishes the mode as 100 microseconds at 10 per cent duty and publishes no derating curve above that. Ten per cent duty at 100 microseconds is a 1 kHz repetition rate. Systems that run at 20 per cent or at 1 per cent are outside the published condition, and the part may well work there, but the thermal answer has to come from measurement rather than from the row.
The pattern across the family is instructive. The 5.8 to 6.7 GHz group publishes parts at 100 microseconds and 10 per cent, the 8.5 to 9.6 GHz group publishes members at 1 millisecond and 20 per cent, and the 13.7 to 14.2 GHz part publishes 1 millisecond and 15 per cent. The vendor chooses a pulse width and duty for each band rather than for each part, so a system that needs a different pulse has to establish its own limit.
Where the duty cycle is fixed by the application, the honest way to specify the part is by average power. 199.53 W at 10 per cent is 19.95 W of average output, which is the same average as a 43 dBm continuous wave part. A system that needs 20 W average in a 10 per cent pulse train is exactly the application for this part, and a system that needs 200 W average is not.
For a first article the test plan should include a long pulse train at the maximum duty the product will ever use, with the flange held at the worst case ambient. That single test answers more than any calculation here, and it is the one test the vendor data cannot replace.
The pulse capacitor bank and the output path
The output voltage at peak is 141.3 V peak into 50 ohm, which is a hard constraint on every component after the drain. Connectors rated for 100 V peak are not enough. Cables used during bench test see the same voltage, and a cable that flashes over at 5 GHz will damage the part and the instrument behind it.
The capacitor bank sits on the other side and has its own constraint. It has to deliver 1.58 mC in 100 microseconds, recover between pulses without ringing into the drain, and present a low impedance at the pulse repetition frequency. Ceramic banks with a mix of values work well because the small values handle the leading edge and the large ones hold the droop. Electrolytics alone ring and age.
A current sense in the source lead is worth including on the first article. Fifteen point eight amps of peak drain current at 5 GHz is not something a clamp meter sees, and a small shunt with a wideband probe gives the pulse shape that the vendor does not publish. The RF procurement route can supply the sense resistors and the pulse capacitors on the same order as the transistor.
The driver stage is the other half of the budget. A 20 W driver at 5 GHz with a flat response over 300 MHz is a part in its own right, and its own compression point has to sit above the 43 dBm the output stage needs. If the driver compresses first, the output stage never reaches its published figure and the fault looks like a transistor problem when it is not.
Choosing between the 5.0 to 5.3 GHz members
The continuous wave member of the same band is HXN40057 at 47 dBm with 11 dB gain. It runs on the same 28 V rail with the same flatness and the same 300 MHz window, and it costs six decibels of output and one decibel of gain in exchange for a steady thermal load. A system that transmits continuously takes the 47 dBm part; a system that transmits in pulses takes this one.
The family also has two broadband members that cover this window from above. HXN43005 spans 5.0 to 5.9 GHz at 47 dBm with 10.5 dB gain and 52.4 per cent efficiency, and HXN43004 spans 4.4 to 6.7 GHz at 51 dBm with 8.5 dB gain. Neither reaches 53 dBm, and both serve a multi band chassis better than a channelized part does.
Cross referencing a pulsed part needs the duty cycle in the comparison table, not only the power. Two parts at 53 dBm are not interchangeable if one is specified at 100 microseconds and 10 per cent and the other at 1 millisecond and 20 per cent. Our RF power transistor cross reference page puts pulse width and duty beside the output column for exactly that reason.
Where the pulse width has to change and no vendor part is published at the new condition, the answer is a bench characterisation rather than a guess. That work sits with the amplifier cross reference route, which covers the driven measurements and the report a customer can put in a project file.
Questions buyers ask about HXN40060
Is 53 dBm peak or average. The vendor publishes 53 dBm in the saturation column with the mode stated as 100 microseconds at 10 per cent, so the figure is peak envelope power within the pulse. Average output at that duty is 19.95 W.
What is the pulse rise time. Not published. A 200 W stage at 5 GHz with a 100 microsecond pulse normally rises in tens of nanoseconds, and the driver and the bias network set that, not the transistor alone.
Can it run continuous wave at reduced power. Nothing on the page says so and nothing forbids it. At continuous wave the 243.9 W of peak dissipation becomes steady state, which needs a carrier that no packaged part of this size can support without a cold plate, so the practical answer is no for most builds.
How is it supplied. There is no orderable addendum, no packing column and no marking code on the vendor page. The purchase order names the band, the rail, the pulse condition and the quantity, and we confirm packing with the factory before a date is quoted. Where a program needs the part held against a build schedule, the one stop RF supply route covers the holding arrangement.