Contact Us

HXC50104

Brand: Hisiwell
Hisiwell HXC50104 is a discrete semiconductors component designed for DC-DC Converters & Power Supplies and related applications. In stock at Superb Automation — request a quote for competitive pricing and fast delivery.
quote now

Product Specifications

HXC50104 — Discrete & Power Semiconductors | Superb Automation

HXC50104

Hisiwell — Discrete & Power Semiconductors

Product Overview

The HXC50104 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 / VceV
Id / IcA
Rds(on) / Vce(sat)mΩ
QgnC
Technology—
Package—
Configuration—
Mounting—

Applications

Switch-Mode Power Supplies Motor Drives & Inverters Battery Management Systems DC-DC Converters Load Switching Automotive Power Distribution

Request Quote — In Stock, 100% New & Original, Full Warranty

HXC50104: 650 W single ended with no ears, for plasma, MRI and radar loads

The HXC50104 is a Hisiwell LDMOS RF power transistor at 650 W in a single ended configuration, unmatched, with a band edge at 500 MHz. The vendor page calls it a 650 watt capable, high performance, unmatched single ended and earless LDMOS FET usable at any frequency up to 0.5 GHz, in CW or pulsed. The word earless is the interesting one: the flange has no mounting ears, so the device is clamped rather than screwed down, and that changes the hardware, the thermal interface and the way the part is replaced in the field.

The applications the page lists are narrower than on the neighbouring rows. Where the 900 W and 1000 W pairs name industrial, scientific and medical work, VHF communication, UHF TV and aerospace in general terms, this row names a laser exciter, a synchrotron, an MRI, a plasma generator and a weather radar. That is a list of loads that are either pulsed, highly reactive, or both, and it is a better guide to what the part is for than the power column is.

The page and the row agree on 650 W and on the 500 MHz edge. Neither publishes gain, efficiency, pulse width, duty cycle, thermal resistance, load pull or an orderable addendum. What follows keeps the published figures, adds arithmetic that a buyer can repeat, and marks every assumed efficiency as an assumption.

The HXC50104 rows as the vendor publishes them

Part classLDMOS RF power transistor, unmatched, single ended, earless flange
Frequency range, index rowHF to 0.5 GHz
Frequency range, vendor pageAny frequency up to 0.5 GHz
Output, index row650 W
Output in dBm, calculated58.13 dBm
Load voltage at 650 W into 50 ohm, calculated180.3 V rms, 255.0 V peak
Peak against the 50 V rail, calculated5.10 times
Peak at a 2 to 1 mismatch, calculated509.9 V
Peak at a 3 to 1 mismatch, calculated441.6 V
DC input at 55 per cent, calculated1181.8 W
Drain current at 50 V and 55 per cent, calculated23.64 A
Heat to remove at 55 per cent, calculated531.8 W
DC input at 65 per cent, calculated1000.0 W
Drain current at 50 V and 65 per cent, calculated20.00 A
Heat to remove at 65 per cent, calculated350.0 W
Output per gigahertz of band, calculated1300 W per GHz, the window being 0.5 GHz wide
Mode, index rowCW
Modes named on the pageCW or pulsed
Drain rail, both sources50 V
Gain, index rowNot published
Efficiency, index rowNot published
Thermal resistance, junction to caseNot published
Load impedance and load pullNot published
PackageEarless flange, no drawing and no body style published
Pulse width and duty cycleNot published
Orderable addendum and packingNone published
Applications listed on the pageLaser exciter, synchrotron, MRI, plasma generator, weather radar
Data sheetNo document linked from the page

What 650 W single ended asks of the rail and the load

650 W into 50 ohm is 180.3 V rms and 255.0 V peak. The peak is 5.10 times the 50 V rail, a lower multiple than the 900 W pair at 6.00 times, because the power is lower. The lower multiple does not make the device easier to use: it means the load line sits closer to the rail and the network has to transform 50 ohm down further, which raises the circulating current in the output match and the current rating of the capacitors.

At a 2 to 1 mismatch the peak at the load reaches 509.9 V, and at 3 to 1 it is 441.6 V through the vector sum. A plasma generator and an MRI coil are both loads that change impedance during a pulse or during warm up, so the mismatch case here is not a fault case but an operating case. The protection loop has to be built for a changing load rather than for a constant one with an occasional fault.

The DC side at 55 per cent is 1181.8 W, 23.64 A at 50 V, with 531.8 W to remove. At 65 per cent it is exactly 1000.0 W, 20.00 A, with 350.0 W to remove, the round number being a coincidence of the arithmetic rather than a specification. Neither efficiency is published, so the supply and the cold plate should be sized on the higher heat figure until the customer's own fixture produces a measurement.

A single ended stage at 650 W has one advantage over a pair that no datasheet column shows: there is no combiner, so there is no reject load to cool, no phase balance to hold and no path by which one device failure takes the full drive into a 3 dB pad. Where the system can accept the single device, it is the smaller and the more robust assembly.

The earless flange and what it changes in the build

An earless flange has no screw holes at the ends. The package is held by a clamp bar that presses the flange against the cold plate across its full width. That is a better thermal joint when it is done properly, because the pressure is distributed rather than concentrated at two screws, and a worse joint when it is done casually, because a clamp that touches only the middle third of the flange leaves the die ends hot and the failure appears months later.

Three things follow. The cold plate needs a flatness and a finish that a clamp can use, typically better than 25 micrometres across the contact area with a surface that will not tear the interface material. The clamp needs a torque figure and a torque sequence from the factory, and if the factory has none, the figure has to be established on a first article with a pressure film rather than guessed. And the replacement procedure in the field needs a tool, because a clamped device cannot be swapped with a screwdriver.

The upside is mechanical. An earless package is shorter than a flanged one with ears, which matters when two devices have to sit close together in a push pull or a balanced pair, and it removes the stress that two screws at the ends put into a ceramic body when the cold plate and the flange expand at different rates.

Pulsed loads, plasma and the band edge at 500 MHz

A weather radar and a plasma generator both pulse, and neither condition is published for this part. Without a pulse width and a duty cycle there is no way to argue a junction temperature for a pulsed application from the row alone, because the thermal impedance at a 100 microsecond pulse is not the DC resistance and the difference can be a factor of two or more. Ask the factory for the transient thermal curve, and if it does not exist, measure the case temperature on the first article under the actual pulse condition.

The 500 MHz edge is where the network stops being forgiving. A shunt capacitor of 22 pF is 14.5 ohm at 500 MHz and 32 ohm at 220 MHz, so the same part value does two very different jobs across the band. A single network cannot hold 650 W everywhere in the window, and the row states one power for the whole range. Cut the network for the sub band the application occupies, and if the application is a plasma generator that sweeps, accept the power variation and state it in the specification rather than discovering it at the customer site.

For a synchrotron or a laser exciter the load is often a cavity or a lamp head with a reactance that varies with temperature. That is the case where a 5.10 times rail ratio and a 2 to 1 mismatch figure of 509.9 V matter less than the phase of the reflection, because a load that is reactive at the wrong phase puts the peak voltage and the peak current in the same place in the device.

The single ended rows around the HXC50104

HXC50104, 650 W single endedThe row this page covers, 58.13 dBm, earless flange
HXC50101, 200 W single ended450 W lower, 5.12 dB below, published data sheet
HXC50119, 250 W single ended400 W lower, 4.15 dB below
HXC50118, 170 W single ended480 W lower, 5.82 dB below
HXC50113, 450 W per device pair900 W, 1.41 dB above, two devices and a combiner
HXC50114, 500 W per device pair1000 W, 1.87 dB above, two devices and a combiner
HXC50111, 300 W per device pair600 W, 0.35 dB below, 900 MHz edge
HXC50109, 130 W per device pair260 W, 3.98 dB below, thermally enhanced

The step from 650 W to the 600 W pair is only 0.35 dB, and the two parts cover different band edges. A buyer who needs 650 W at 400 MHz and a buyer who needs 600 W at 850 MHz are not choosing between these rows on power; they are choosing on frequency, and the frequency decision is made first.

Against the 900 W pair the single device is 1.41 dB lower and removes a combiner, a reject load and a second flange. In a plasma or MRI system where the enclosure is already crowded, that trade is usually worth taking, and the 1.41 dB can be recovered in the transducer or the coil rather than in the amplifier.

Incoming inspection on a clamped device

A clamped device cannot be checked for mounting the way a flanged one is, so the incoming plan has to add a mechanical step. Measure the flange flatness on a surface plate, check that the clamp bar is straight, and confirm the interface material is the one specified rather than a substitute. Then run the electrical checks in the usual order: gate and drain junction on a curve tracer, marking and date code against the certificate, quiescent current at the stated gate voltage, and gain at the band centre.

The incoming inspection plan page carries the electrical sequence in full. The two additions for this part are the flatness check and a first article thermal run, where the case temperature is measured at three points across the flange rather than at one, because a clamp that is not seating shows up as a spread between those three readings long before it shows up as a failure.

Where the part is bought for a pulsed application, the first article should be run at the actual pulse width and duty cycle rather than in CW. A part that survives CW at 650 W will usually survive a pulse, but the reverse is not true, and a supplier who has only been tested in CW has not been tested for the application the page lists first.

Questions buyers ask about the HXC50104

What does earless mean? No mounting ears on the flange, so the device is held by a clamp bar rather than by screws. It is a mechanical difference with a thermal consequence, not a different die.

Is 650 W the CW rating? The row says CW and the page says CW or pulsed, so the class answer is yes. No pulse width, duty cycle or thermal resistance is published, so the rating for a pulsed application has to be confirmed by the factory.

How much heat is there? 531.8 W at 55 per cent and 350.0 W at 65 per cent, both calculated from an assumed efficiency. The cold plate should be sized on the higher figure.

What supply current does it draw? 23.64 A at 55 per cent and 20.00 A at 65 per cent, at 50 V. The supply should be rated above the higher figure with margin for the drive stage.

Can it drive a mismatched plasma load? The page lists plasma generation, and the 2 to 1 case puts 509.9 V at the load. That is survivable with a protection loop and not survivable with a fixed drive and no reflected power limit.

Is there a data sheet? None is linked for this row. On the same shelf the 200 W single ended row has a full document with maximum ratings, thermal resistance and a fixture drawing, which is the format to request here.

For price and lead time on the HXC50104 with the 200 W row as a lower power alternate, send the sub band, the pulse condition and the quantity to the RF component sourcing desk. Where the part replaces an incumbent device in an existing cavity or lamp driver, start from the cross reference page so the comparison is made at the same drive and the same load.