I enjoyed looking at the PIX of the compact amplifier of Larry's - nice
piece. This email QSO stuff is funner than HF, cause it includes images
now. Its also easier to jump in an grab the mike, as it is a serial
transmission, not parallel like open HF operation is. Only my packet gets
through when it gets through. The QRN is gone, but boy, there is a lot of
QRM here! Nice thing is that everyone has a chance to get through, all at
the same power level. Here, I CRANKED UP MY OUTPUT BY 3DB, BY GOING TO
CAPITAL LETTERS. I could send a bunch of HTML background noise, and I
suppose that is like splatter on my rig.
Now for the real meat of my entry:
I agree wholeheartedly that the Sprague doorknobs are really made to work
filtering 15 KHZ in TV HV supplies, not in the MHz. They might work, but
the losses and subsequent internal heating are high. Not to mention the
resonances in them, since they are high K dielectric and probably have a
lot of internal inductance. But if you can get by with them, then so be it.
I believe it is a gross misuse of a component in a circuit that should have
a better one. If it fails, you get DC on your output network, and so on. Is
that safe? Is there an RFC or L to ground the network? Worse yet, does the
DC come out to your antenna feeder then?
The Centralab style ceramics are much better, the ones < 1 nF. In DC
blocking application, you probably don't care so much that the capacitance
drifts as they heat, as you would hopefully design the thing to have a
capacitive reactance maybe 10 times lower than the impedance of the circuit
it is in, at the lowest operating frequency. To get enough C, for a low
Xc, you might have to parallel the Centralab 470 pf or 560 pf, or 330 pf,
etc. At 10 meters you have to watch for the resonance of the caps. Also,
when placing caps in paralllel,use the same values and be sure to strap
them so that the electrical length and inductance to each cap is identical,
for current sharing. Never place them linearly along the strap or mounting
plates, in line with the current flow, but place them on a T plate, which
causes current to divide equally, with the caps all perpindicular to the
ciruit. I am not sure how to show this easily on email.
This is good, all caps marked with x, bodies all in parallel
| x|
-----------| |--------------
| x|
This is bad, caps are mounted so that some have lower L connectiions.
------------------
x x x x
----------------------
FORGET IT, i cannot explain. Maybe Rich or Tom or someone else?
Keep the loops between caps short, to minimize inductance and to prevent
resonant circuits within. See Dittrich/Philips application text TUBES FOR
RF HEATING for a lot of discussion on this.
Better yet, splurge and find some big ceramic jobbies, like the Draloric,
or the Unilator or the High Energy. I got some custom made, 20,000 pf, 25
kV, ceramic disk, about 6 inches in diameter, with a ring of tapped holes
around the perimeter. Hi Energy made them for about $500 each. It took 6
months to get delivery, but it is a really nice low Z capacitor. I wish
these were available from Fair Radio! I use them at 100 Kw, at 2.8 MHz,
plate voltage is 15 kV DC.
In 1998 we ran some self resonance checks on various capacitors in a
fixture made of heavy copper strap and an HP 4193A impedance meter. The cap
was basically fixtured across the probe of the vector Z meter:
CDE 272 (Cornell Dublier mica block) rated 0.02 uF @ 2 KV:
Series Resonance at 4.5 MHz then another resonance (parallel) at about 40
MHz. Would be useless here.
Centralab 858 Ceramic doorknob style rated 1000 pf, 5 kV, X5V dielectric
Series Resonance at 28 MHz. much higher Q, nothing else beyond 81 MHz, but
stays inductive there
Put 8 x 1000 pF in parallel on a pair of round plates, around the
circumference.
Series Resonance at 12 Mhz, still clean to beyond 81 MHz, but all inductive.
73
John
K5PRO
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|