Transistor for 2.45 GHz Power Amplifier

[ON5VO]

Hi All,

For a university course where students design and measure various RF circuits on PCB, we're looking for a transistor that they can use for the design of a power amplifier at 2.45 GHz targeting 20 dBm or higher. The main goal is not to achieve certain specifications, hence the vague power specification, but put theory into practice. Our previous searches have led either to too high power devices, which are quite expensive (>€15) and will be difficult to cool down such that the students will not be able to measure up to the compression point, or medium power devices, with only S-parameters models not suitable for nonlinear load-pull simulations. In previous years we've used a QPD0005 (used different specifications), which had a very nice model that allowed to simulate the voltage an current waveforms at the transistor, not just the package, but it was notoriously difficult to cool down and the model was inaccurate when using a lower supply voltage (confirmed by vendor) when trying to reduce the output power. Another year we've used the BLP15H9S10XY, which was doable, but the targeted frequency this year lies beyond its operating range.

Do you have suggestions on which transistor we could use?

[BigBoss]

If the P₁ dB compression point is around +20 dBm, it will not be a power amplifier. In case the target is higher than, the choices are limited nowadays.
The RF power transistors are really expensive, and cheap "medium power transistors" disappeared. Although you find NOS transistors, the models will not be very accurate or not compatible with modern simulators.
Some old Motorola transistors are still manufactured in India and Korea under MOT license, but I'm not sure that they supply recent RF models or not. I saw some Chinese semicon. Companies were still manufacturing
old Fujitsu, Toshiba, and Hitachi RF medium power transistors, but I don't know how to find and buy them.
It sounds hard to find those nice components nowadays. They have all gone...

[NiHaoMike]

There's the RFX2401C, commonly paired with the NRF24L01 for somewhat longer range applications. It's both a PA and a LNA, but it can be pin strapped to use it as a PA only.

[LaserSteve]

Talk to the people at RFPARTS.com and see what they can sell you in bulk and buy a big supply, stock up for a few years worth of classes

Then consider one of the cheap 2-3$ broadband MMICs for 2-3.5 Ghz that are available.  I assure you they not  are as easy to use as their data sheets assure you.

Steve

[ftg]

Has RFPARTS.com changed their policy on selling abroad recently?
Last time I looked they were domestic USA orders only company.
If one wanted to go with surplus parts, one similar company in Europe is rf-microwave.com from Italy.

I cannot suggest any suitable transistors with good non-linear models, as I'm not aware of any that would fulfil your requirements.
When I had similar sounding classes in collage we just used the cheap and cheerful BFP420.
But those are only cheap, not high power.

[dietert1]

When i look up digikey they list 32 bipolar transistors in the 1 .. 2 W range with 6-8 GHz transition frquency. Three are in stock: 2SC3357, 2SC4703-T1-AZ and NE85634-T1-RE-A. Didn't research their datsheets nor simulation models but i'd guess any one of them should serve.
They also list higher frequency transistors, yet without stock, except the BFP490E6327. That one is 4.5 V and fT = 17.5 GHz.
Anyway, it can be more educational to make an amplfier with parts close to their limits.

Regards, Dieter

[MT4S301]

YP601238T, a chinese unmatched GAN HEMT available on LCSC. I didn't find linearized or large signal model on mfr's webpage.

[BigBoss]

Even though these are not RF power transistors, they can reach almost +20dBm power output level.
IMHO, they are pretty appropriate components for playing on to learn some fundamentals of RF.
Also, Minicircuits might hopefully supply modern and accurate nonlinear models. Look at the Excel file too.
https://www.mouser.fr/fr/c/semiconductors/wireless-rf-semiconductors/transistors-rf/?m=Mini-Circuits

[LeslieGuo]

For a teaching lab I'd weight 'debuggability' above raw performance. A few things that matter more than which transistor:

1. Stability first. Students will build it, and it will want to oscillate -- check K-factor across the whole band and well out of band (a 2.4 GHz PA loves to oscillate at 5-6 GHz where nobody's looking). Pick a part whose datasheet actually gives S-parameters, and have them simulate stability circles before touching copper.

2. Match to what you can fabricate. Microstrip matching on standard 1.6mm FR4 at 2.45 GHz is very doable, but the students need the exact stackup (er, copper weight) -- a 10% er error moves your match noticeably. If the course has access to a VNA, make 'measure your own microstrip' week 1.

3. Thermal. Even 100-200 mW in a SOT-89 gets properly hot without copper pour + vias. Budget the thermal design into the layout exercise, not as an afterthought.

On parts: for this use case I'd lean toward something with a vendor reference layout to copy (BFP420-class SiGe is cheap and forgiving; the integrated PAs like RFX2401C get you on the air faster but teach less about matching). Either way, make them build the reference design first, then modify the match -- that's where the learning is.