HX130450
Product Specifications
HX130450
Hisiwell — Amplifiers & Comparators
Product Overview
The HX130450 from Hisiwell is a precision amplifier / comparator delivering high gain, wide bandwidth, and low distortion for signal conditioning. 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
- High gain-bandwidth product for wideband signal amplification
- Low input offset voltage and drift for precision DC applications
- EMI-hardened design for robust operation in noisy environments
- Rail-to-rail input/output for maximum dynamic range
- Low quiescent current for power-sensitive designs
- Unity-gain stable with capacitive load drive capability
Key Specifications
| Channels | — |
| GBWP | MHz |
| Slew Rate | V/µs |
| Vos | µV |
| Iq per Channel | µA |
| Input Type | — |
| Supply Voltage | V |
| Operating Temp | °C |
| Package | — |
Applications
HX130450: 2 to 20 GHz as bare die, and the half nanohenry that sets the top of the band
The HX130450 is the die version of the 2 to 20 GHz distributed amplifier: 16 dB of gain with plus or minus 1 dB of flatness, 3 dB noise figure, 14 dBm at the 1 dB compression point, 1.3:1 at both ports, 5 V at 60 mA. The packaged HX130450F5 carries the same band and almost the same row. The difference that decides between them is not on the datasheet at all: it is the inductance of the two wires that connect the die to the board, and what that inductance does at 20 GHz.
Published values
| Parameter | Value |
|---|---|
| Frequency range | 2 to 20 GHz |
| Small signal gain | 16 dB, flatness plus or minus 1 dB |
| Noise figure | 3 dB |
| Output P-1dB | 14 dBm |
| Input VSWR | 1.3:1 |
| Output VSWR | 1.3:1 |
| Supply | 5 V at 60 mA |
| Form | Bare die |
Half a nanohenry at 20 GHz is 63 ohm
A 0.5 nH bond wire is 6.28 ohm at 2 GHz and 62.83 ohm at 20 GHz, because reactance rises with frequency. A 62.83 ohm series element in a 50 ohm interface is a mismatch larger than anything the die itself presents, and it appears at exactly the frequency where the part is supposed to still work. The die row quotes 1.3:1 at both ports, which is a measurement made with the vendor's own bond length and launch. On a customer board the match at 20 GHz is a property of the assembly, not of the part, and that is the difference between the die and the QFN.
Where the die is the better buy
The QFN costs 0.5 dB of gain, 15.5 dB against 16 dB, and improves the input match to 1.25:1, 19.08 dB of return loss against 17.69 dB. When a board already runs a die attach process and the wire bonder is qualified, the die removes a package from the BOM and puts the thermal path directly on the spreader. When it does not, the QFN delivers the same band without a wire bond process to qualify, and half a dB is inside the uncertainty of most bench measurements at this frequency.
DC, heat and drive
5 V at 60 mA is 0.300 W in, 25 mW out at compression, 8.4 percent conversion, 0.275 W of heat. The input at compression is minus 2 dBm, so a mixer or a DAC that delivers 0 dBm will push the stage into compression and cost intercept. At 3 dB noise figure and 16 dB gain, the part works as a first stage in a wideband receiver, but only if the level in front of it is controlled.
Two questions buyers ask
Does the die need a negative gate supply? The row states a single 5 V rail with no gate reference, unlike the 8 V wideband parts in the same catalogue which list a negative gate. Confirm the bias with the vendor before assembly, because the gate condition on first power up is not recoverable on a die.
Can the same board take either version? Not without a change. The QFN 5 x 5 footprint and the die attach pad are different, and the wire lengths are different, so the 20 GHz performance has to be re measured after the swap even though the band and the pin function are the same.
SUPERB supplies the HX130450 as bare die and the HX130450F5 packaged, quoted together so the assembly cost can be weighed against the part price. Amplifier cross reference covers what has to match before a distributed stage is replaced, RF component sourcing covers the surrounding BOM, and the family sits in the amplifiers and comparators index.
HX130450: a level plan for a 2 to 20 GHz first stage
Sixteen decibels of gain, 3 dB of noise figure and 14 dBm at compression make this die a receiver first stage, and the level plan is where that claim is won or lost. The input at compression is minus 2 dBm, so a mixer or a converter that delivers 0 dBm pushes the stage 2 dB into compression before any cable loss is counted. On 5 V at 60 mA the part dissipates 0.30 W and produces 25 mW at compression, which is 8.4 percent conversion and 0.275 W of heat, so the thermal side is easy and the level side is not.
| Quantity | Value | What it does to the chain |
|---|---|---|
| Gain | 16 dB, flatness plus or minus 1 dB | A minus 60 dBm signal becomes minus 44 dBm at the output |
| Noise figure | 3 dB | Sets the noise floor of everything behind it divided by the gain |
| Input at compression | minus 2 dBm, from 14 dBm out and 16 dB of gain | A 0 dBm source already costs 2 dB of compression |
| Dissipation | 0.30 W dc, 0.275 W of heat | Trivial for a die on a carrier, and stable in still air |
| Output drive | 25 mW at compression | A 50 ohm load sees 3.54 V peak to peak at that point |
| Noise floor behind it | Second stage noise divided by 16 dB | A 10 dB second stage adds about 0.3 dB to the total |
Minus two dBm is a low ceiling for a wideband stage
A receiver that expects signals up to 0 dBm needs a pad or a limiter in front of this part, and the pad costs the same number of decibels of noise figure that it provides in headroom. That trade is the whole calculation: 3 dB of pad gives 3 dB of input headroom and moves the system noise figure from 3 dB to 6 dB, which is a real cost in a wideband front end. A limiter costs less noise figure and protects the die, which is why the part is normally placed behind one rather than directly at the connector.
The die and the packaged version differ at the top of the band
The packaged HX130450F5 costs 0.5 dB of gain and improves the input match slightly, and those numbers are small enough that the choice is made by the assembly process rather than by the electrical row. Where a die attach and wire bond process is already qualified, the die removes a package and puts the heat path directly on the spreader. Where it is not, the packaged part reaches the same band without a new process to qualify, and 0.5 dB is inside the repeatability of most bench setups at 20 GHz.
What to hold on the drawing
A die with no published gate reference has to be biased in a sequence that the vendor confirms, because the first power up is not recoverable. The wiring length at the launch is part of the electrical result, not a mechanical detail. Comparisons before a distributive stage is replaced are listed on the amplifier cross reference hub, the surrounding bias and filter parts are under RF component sourcing, and availability of both forms is quoted from the amplifiers and comparators index.
Request Quote — In Stock, 100% New & Original, Full Warranty