Organization: Duke University EE Dept.; Durham, NC
Lines: 9
Distribution: world
Message-ID: <3ts2e3$ckj@news.duke.edu>
NNTP-Posting-Host: ee.ee.duke.edu
Hi,
I noticed that there was a previous posting asking for info. about a starter kit for somebody who developped an interest in the field recently. I second
In article 8a1@cismsun.univ-lyon1.fr, elendir@enst.fr (Elendir) writes:
> Hi
> I've got another question.
>
> I have to design the first RF stages of my RXver. For that purpose,
> I'd like to use the XXXX (a transistor whose name I can't remember).
> I'd like to use it with a source Z of 50 ohms, and a load Z of 50
> ohms also, since this scheme does insure stability and low-noise.
> But, although I encounter no problem for the first device, how
> can I design a network that would transform the Zin of the second
> stageinto a collector load of 50 ohms, and at the same time
> transform the first stage output Z into 50 ohms .
> That is to say a network which, when loaded by the Zout of the
> first transistor would transform it into the 50 ohms needed by the
> input stage of the second, while at the same time transforming
> the Zin of the second stage into the 50 ohms needed by the first stage ?
>
> Thanks,
> Vince
>
> --
> F5RCS - Worldwide Friendship through Amateur Radio
> ENST, Ecole Nationale Superieure des Telecommunications, Paris
You have lost me a bit there, but I'll do the best I can.
Desiging a network to match any complex impedance to any other is simple. All you need is an L-network. I have a program (PC) I wrote to do it, which I wo
wanted. The other way is to make the impedance real (add series L or C) then
transform the one real impedance to another. There are lots of ways to do it. Eg
Pi, L, T networks. There should be a program published in EDNs 'Design Ideas' soon
to do it. I wrote the code, but cant give it to you (as they paid me). All networks
can do the job, its just some are more practical.
The problem I've had in the past is that S-parameters assume the device has 50 Ohm
loads on all ports. Since S12!= 0, matching the o/p to its conjugate for maximum
power transfer means the input Z will change. Even if you calculate its new value,
and match the input, the output matching gets screwed up. So you re-match the output,
but then the input match becomes poor. In other words, it becomes an iterative
process. I've never found a definitative way to do it quickly. (I posted a request
recently for a decent way, but got no replies).
If you have two stages, there is no need to go via 50 Ohms, although I admit it
makes testing easier.
You imply the devies are stable and low-noise at 50 Ohms. Most devices have an
optimum input Z for lowest noise. You cant assume its 50 Ohms. Its usually
put on the device data sheet as TOR_opt (Greek Tor). You should try to make sure you
input device sees this reflection coefficient, rather than 50 Ohms. Its easy to
convert from tor to Z.
If the device
is not unconditionally stable, then you have to be careful what load it see's. Again
assuming 50 Ohms is okay is not really on. You need to calculate a couple of
stability factors. One will tell you if its unconditionally stable, in whcih case your
okay. If not, you have to be careful. See 'Microwave Mesfets' By Pengelly, for how
to do that.
Hope this is of some help. If you want the program to convert one arbitrary complex
impedance to another, please let me know. I'll put it on our ftp site. I should add
there are a number of Application notes by Mototola and HP on this sort of thing.
I'm no expert, but have managed, with a bit of fiddling/luck/insight etc to design a
few amps to be stable and have good return losses.
