String fullTitle = 'undefined';
int postNumber = 2790407;
String image = '1713831228357741.jpg';
String date = '04/22/24(Mon)20:13:48';
String comment = 'anons im having trouble selecting ideal transistors for use with an RF amplifier design < 100mhz. It seems that the tradeoffs are just too great- you either have low impedance gain or high impedance gain, and you either have low current and low noise but also poor handling, or high current and higher shot noise. rBB is a big part of this and is rarely documented, because it's usually lower than other noise figures. beta is sometimes very high but this isn't necessarily a good thing. have been reading a bunch of documents
pic related, this shit pisses me off
whether it's nexperia, onsemi, nxp, zetex, or infineon, choices for good transistors designed for HF are limited and most offer shit noise figures or require end users to test tons of models to characterize what the mfg didnt think important
trying to design a loop amplifier based on wellbrook and has:
good linearity from 5nv to 500mV, positive gain from 100khz-10mhz(used a LPF 2nH capacitor for this)
consumes under 100mA(ideally under 30mA) of current, powers off 3-5V bias-T power insert
really good, low noise figure
have had some good results with phemt but biasing them while keeping self-oscillation down is hard
a few good jfet exist, but the square law behavior is not desirable
learning transistor behavior has been a giant pain in the ass and it feels like nobody designs transistors anymore for HF
its all crap they designed decades ago';
'someone is actually transmitting something worth listening to below 100mhz?';
'>>2790407
Can't you pick transistors designed for even higher frequencies (VHF, UHF)? Or is all that stuff done monolithically?';
'I hate to say it but what you're experiencing is par for the course for RF design, from shitty and downright incorrect models, to a complete lack of perfect-fit RF transistors.
You need to design multi-transistor amplifiers and lots of buffring.This is the only way you might possibly get the performance you want. Look into cascoding.
>>2791835
Pretty much anything high frequency is done monolithically nowadays';
}