On Sat, 2008-05-10 at 18:11 -0700, Jim Brown K9YC wrote:
> Jerry,
>
> If you haven't already done so, have a look at the measurement
> issues that I've raised in the Power Point file below.
>
> http://audiosystemsgroup.com/NCDXACoaxChokesPPT.pdf
>
> Then take a look at
>
> http://audiosystemsgroup.com/Toroid61Data-2.pdf
>
> Virtually any reflection-based measurement (i.e., S11) will
> give you VERY wrong answers for those chokes wound on #61 (and
> also for the higher frequency coax chokes on #31 cores). It IS
> practical to get good data using S21 (i.e., as the series
> element of a voltage divider.
Ah, we tend to neglect transmission line effects. And you make a very
valid point that there are many lengths of feedline can be series
resonated by the high Q choke/balun. When I was at Collins we built a
250 KW AM short wave transmitter that used coax for the inductance of
the output PI network. We called it "Pi Line." It band switched (had to
cover 3.95 to 26.5 MHz continuously) by running vacuum capacitor banks
in and out (high and low C) at various points along the line. There were
no moving contacts which is handy at a couple kiloamps circulating
current. The concept though started with pieces of RG-59 and clips to
MOVE the shunt capacitors along the line which ignored the fact that the
capacitors pulled to minimum C still had significant C and L and could
easily come to series resonance at a harmonic and those big vacuum
variables were really low loss so the circulating current in that series
resonance was a serious problem. Many a capacitor planned for minimum
had to be moved to keep that resonance away from a harmonic.
Some years later I explored using coax lines for inductances in ham rig
HF PAs. Computing only load impedance at the tubes showed nice
functions. Checking transmission (S21) showed some real problems when
the line sections and the terminating capacitors resonated in
transmission line modes especially at 1/4 wave line where the input Z
soared which would lead to PT tube flashovers or parasitic oscillations.
Following that, I looked at lengths of coax between PA and the typical
low pass filter that includes a series resonant circuit across the coax
connector (m derived filter). When the coax was an odd multiple of a
quarter wave, it matched the low Z at the notch frequency to a high
impedance at the PA coax connector and so increased the harmonic output
through the TVI filter rather than decreased it.
Point is and you made it, that coax combined with lumped resonant parts,
leads to resonances you didn't want in places you didn't expect and
makes some of the additions produce exactly the opposite results you
expected. You can't neglect transmission lines.
As for measuring, there are more techniques than problems, many times. I
like the Q-meter and RX-meter because I used them and found them applied
in my early engineering career. And they became reasonably priced,
though my first RX-meter came from Army MARS.
>
> The Q-meter is fine instrument, but not many folks still have
> one or know how to use one. As the Power Point shows, my
> voltage divider technique is able to get pretty good data for
> some pretty high Z parts with instrumentation of modest cost.
Takes some computation though. I notice wiggles, might they be from clip
lead lengths or imperfections in the 50 ohm terminations not being
purely resistive?
>
> 73,
>
> Jim K9YC
>
One of my projects at Collins after the 250KW was delivered to VOA
(Dixon, Delano, and Bethany stations) was a book on coil design for
power applications. Ordinary coils up to a few KW pancake coils were
measured on the Q-meter but larger coils just couldn't be connected.
Besides the loss of the Q-meter capacitor wasn't all that good compared
to the biggest coil, 18" diameter wound of 1-1/2" copper pipe in a 3'
shield. So we tuned the bigger coils with enclosed big vacuum variables
that had a measured Q on the order of 10,000. We used a pair of small
loops made of 51 ohm precision resistors. We measured Q by bandwidth
with an HP606A generator, a Boonton RF millivolt meter accompanied by a
counter. The millivolt meter had a mirrored scale and I hung a magnifier
in front of the -3 dB point on the meter. We reduced the loop areas
until the Q stopped increasing, then reduced the loop coupling ten dB
more. With that big coil measured Q was around 3000 so holding the
generator (with reduction drive) while turning to read the counter
didn't work. Had to trigger it with a foot switch. The whole test set
was bolted down to a copper topped work bench and supplied by a Sola
voltage regulator. There were a couple other transmitter projects making
RF in the same lab, so the grounding and shielding details were
important. The local HP rep tried to loan me an HP 5100/5110 synthesized
signal generator but it didn't work well because of the level shifts
between adjacent frequencies. But now I own one and find it handy with
measuring quartz crystals but I don't use an indicator with the
sensitivity that I did at Collins, though I could. A wide band scope is
quicker but doesn't have the fine resolution of the 5" mirrored scale meter.
73, Jerry, K0CQ
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|