chinese 2SC3356 s-parameter measurement

[MT4S301]

The 2SC3356 transistor, date back to the Showa era, was well received by china engineers throughout the 90s. It offered incomparable performance for its time that easily outperformed chinese non-military RF BJTs. This particular type went so popular that many chinese manufacturers give their own equivalents the same name, and today "2SC3356" become the only type of RF BJT from many chinese manufacturers. However for most enthusiasts whether these chinese transistors could rival NEC's 3356s is an unanswered question. That's why I decided to measure s-para of several chinese 3356 myself.
I make this simple test circuit and send it to JLCPCB for cheap FR4 boards with 0.8mm thickness. Solid ground plane and multiple vias at emitter pin.


DUT biasing is established and stabilyzed by RC=100ohm and varying Vcc between 4 to 5 Volts. All transistors are biased at Ic=10mA to reflect modern untuned(wideband) system requirement on linearity
The "FB" at collector are muRata BLM15HD182SN1 which provide >1kohm impedance from 100MHz to 1GHz. It is thus not de-embedded from the measurement results.

Then I melt some solder and turn on a pro-grade VNA. Open and short traces are added to the PCB to enable de-embedding. I put a magnifier in front of my camera for the close-up photo but also introduce aberrations



Here is what I got in the end: S-para of 4 chinese 2SC3356 transistors plus one BFR182 serve as a "known reference".


In conclusion all these chinese 2SC3356s show S-parameters simlar to the Philips BFR92 or Toshiba 2SC5084; for some reason their S-parameter aren't close to what NEC showed in datasheet. Anyway, I hope someone could find my quick/simple measurements helpful

[MT4S301]

Bonus track:
Years ago I came across some "original" NEC 2SA1977 RF PNPs at a brick-and-mortar shop in Shenzhen. After paying a kidney I secured dozens, perhaps the fastest discrete PNP devices. Applying negative supply voltage on Vcc terminal and adjust for Icq=-10mA enables me to test these PNPs as well.

It turned out these 2SA1977 are genuine. Their bandwidth is surprisingly nice despite being PNP

[ftg]

Thank you for doing these measurements!
I have been wondering about those cheap "2SC3356" transistors on LCSC and how close they are to the parameters published by NEC/Renesas.
Here in Europe the RF BJT of choice during the 2SC3356-era was BFR93 or BFR92.
So these being close to them is an interesting coincidence.

[MT4S301]

Here in Europe the RF BJT of choice during the 2SC3356-era was BFR93 or BFR92.
So these being close to them is an interesting coincidence.

That's great since the BFR92 spice model available online could predict behavior of a large amount of cheap alternatives.
In fact I have built 100MHz cascode LNAs with BFT92 model +chinese 3356 devices. The circuit's measured S-para are highly consistent with simulation. Haven't tried higher freq.