From: Winfield Hill
Subject: Re: Q. medium-power RF transistor
Date: 9 Nov 2002 13:50:44 -0800
Organization: Rowland Institute
X-Newsreader: Direct Read News 4.10
email@example.com (carltons) wrote...
> Helmut Sennewald HelmutSennewald@t-online.de wrote:
>> Winfield Hill schrieb ...
>>> I'm thinking about making 20 to 30V pulses with a
>>> 1ns max risetime into a 50-ohm line...
>>> What's a good low-capacitance transistor to use with
>>> say Vceo 30V, 500mA capable (h_fe >10 at 500mA) and
>>> with f_T of at least say 1GHz at Ic 200mA, Vce 15V?
>> Hello Win,
>> I would try four cascodes in parallel with BFG135 transistors.
>> The product of output capacitance of your transistor(s) and
>> the load (50Ohm) has to be below the wanted rise time.
That's a nice transistor, reasonable beta at 100mA, with just
over 3pF of Ccb+Coe capacitance. I could parallel four with
a 50-ohm pullup to 25V (each in common-base mode, Vcbo = 25V),
and still have RC under 1ns. And it looks easy to get.
But I wonder, where have all the medium-current 25+ volt RF
transistors gone? All the likely parts I have examined are
discontinued. For example, an ON Semi MRF3010 or MRF134 RF
MOSFET (65V, 1A, Coss + Crss < 12pF) would have been nice a
few years ago. What's the ticket now?
>>> For example in the old days we'd consider a 2n3866,
>>> etc., but this falls short of the mark for the above
>>> specs, and is clearly far behind today's curve. Hmm,
>>> what's a common transistor to get 5 watts at 900MHz?
>> This is a very different application. It is narrow band only.
>> The output capacitance of the transistor(s) is compensated
>> with the inductive collector load(resonance circuit).
That's true, a 900MHz transmitter would be narrow-band tuned.
But if it uses say 20V power supplies and we examine a typical
part used for 5 or 10-watt resonant power, it may have a low
enough capacitance for fast risetime for my flat-top pulse.
> Hi Win,
> I think Helmut is definitely on to something good. I picture
> it myself as a kind of blocking oscillator design without
> the oscillation if you get my direction. ...
Thanks, but I need a logic-controlled 20V step with 1ns edges.
> Another warning bell is going off in my head also. Look out
> for Miller cap. ...
C_cb has to be dealt with, you betcha!