I was thinking about how the original post was about not having L and C
labelled
on the tuner, and translating knowledge of L and C in the tuner into "what's
going on"
(i'm wrestling with the pi network on the input of an amp, so the theory behind
matching networks is something I'm revisiting right now)
G3YNH has a nice page here, on "what's going on"
http://www.g3ynh.info/zdocs/z_matcing/part_2.html
If you look at his description for T tuners, he presents a nice way of tuning,
if you had a antenna analyzer for an exciter, that could indicate the sign of
reactance.
i.e. my interpretation: (for a T matching network)
-Set L, Cin, Cout to max
-Reduce L until you get R=50
-If the measured reactance the analyzer sees is postive, reduce Cout till you
get it to 0. Leave Cin at max.
-If the measured reactance the analyzer sees is negative, reduce Cin till you
get it to 0. Leave Cout at max.
hmm. I suppose this could be used with a MFJ-259B which doesn't indicate + vs -
reactance..Just tweak Cin or Cout after the first step, based on whatever gets
you to 0. This is a slightly different procedure than advocated in the arrl
article. The theory G3YNH shows seems right?
He reiterates the point, that with a lossy coil, combined with not enough C on
one side or the other, can let you tune all the transmitter power into coil
losses.
And now for a contentious point he makes
-for a T network, losses increase with reduced inductance (parallel inductance)
-for a Pi network, losses increase with increased inductance (series inductance)
His reasoning: The Q of a air coil goes down when you have a small number of
turns
(uses example of Q just = 20 to 30 on 10M with small number of turns).
So if Q increases with N (turns), you should use the most N.
i.e. it's not just about a goal of "minimizing L". (for T network)
Am I understanding right?
-kevin
ad6z
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|