[AMPS] Re:

[Richard W. Ehrhorn]

Rich & interested others...

Observations, opinions, caveats, and a question...

1) Ran into Tom at Dayton and after reading all the flak here was surprised 
to find that he's still a pretty decent guy who seems to know amps well. 
It's so disillusioning!

 2) Never met Ian, but the gut feeling grows that anyone who challenges him 
in the area of network analysis or any fundamental EE stuff has a real good 
chance of losing. Very respectable stuff, Ian - I appreciate it.

3) Too many guys are talking about output circuit "Q" without specifying 
"loaded Q" or "UNloaded Q." They're very different things and the ambiguity 
can cause confusion. I absolutely agree with Tom, Pete and others that 
inadequate loading of a class AB, B or C amplifier, whether the output 
network is "old-fashioned" parallel resonant, a pi, or a pi-L, results in 
excessive "LOADED Q" of the tank circuit which can AND DOES often create 
peak rf voltages several times the DC plate voltage. And it's not magic or 
very hard to understand. I defer to Ian, however, for the rigorous proof, 
because it's easy to see that he won't have to work as hard as I would to 
dig it out(!)

4) Maybe I've missed someone else's similar description, but I believe the 
logical, conventional, and easiest-to-understand explanation of how a 
common parasitic suppressor works is that [a], it does not ABSORB VHF/UHF 
parasitic power or energy, but PREVENTS (or "suppresses") the parasitic 
oscillation from occurring in the first place; [b] it does so by lowering 
the loaded "Q" of the existing parasitic resonance(s?) in the anode circuit 
to the point where feedback loop gain at the parasitic resonant frequency 
(-ies) is too low to support oscillation. The trick, if you want to call it 
that, is to introduce enough loss (resistance) into the parasitic resonant 
circuit to do the job without absorbing so much of the 
fundamental-frequency power as to either overheat itself or unduly reduce 
amp efficiency/output.

5. Rich: I never had an SB-220, but if you'll post the approximate 
dimensions of its 80M tank L (conductor dia., coil I.D., # of turns, 
overall length) I'd like to make a similar coil, tweak it to ~10 uH, and 
measure its impedance at 110 MHz on a good network analyzer. My guess is 
that, at 110 MHz, distributed capacitance of the coil, mostly, makes it 
look quite different from "2pi f L" = 6910 ohms reactive. 'Course it may 
look quite different even from what's measured on the bench when it's 
located "in situ," in the amp.

What's the point? IMHO speculation about what an HF tank looks like at VHF 
tends to ignore distributed capacitance of the coil winding and its tap 
leads, stray (or "parasitic"!) inductances in variable capacitor and other 
rf structures, miscellaneous other little reactances and losses that may 
become significant at VHF/UHF, and the plethora of resonances - series and 
parallel - that the whole ensemble creates to snare the unwary. Sort of a 
manifestation of the principle of unintended consequences ("PUC").

6. Another caveat related to the "PUC," which I haven't seen mentioned in 
references to paralleling multiple capacitors "to get rid of troublesome 
resonances," more or less: If you acknowledge that every capacitor has some 
parasitic (sorry!) inductance as well as various other distributed 
reactances, it's apparent that multiple caps in parallel have more series 
and parallel resonances collectively than they do individually. WAY more. 
If you don't check them out and yet don't get bitten by them, consider 
yourself lucky.

Try paralleling 3 or 4 common "850 series" doorknob caps, say by bolting 
them between flat plates of copper to minimize stray inductance. Really 
clean, right? But then investigate with a GDO, or better yet, a vector Z 
meter or a network analyzer. BTW, this isn't an academic exercise - just 
another lesson earned the hard way and pretty obvious after-the-fact. 
Pretty obvious a priori, too, if we consider a simple cap equivalent 
circuit as consisting of the nominal C in series with a small parasitic L, 
the pair then paralleled by a small parasitic C. Put a couple of those in 
parallel and it quickly gets complicated to calculate all the series and 
parallel resonances. Bottom line? If you're lucky, paralleling 2 or more 
caps may avoid resonance problems and effectively just give you C1 + C2 + 
... + Cn overall. If you're NOT lucky, it may give you a nasty composite 
resonance right where it hurts.

7. Finally, I think Carl's reference to resonances created in or by shorted 
turns of bandswitched tank coils is highly relevant. Extremely high-Q 
resonances often are created and can result in very high rf voltages, and 
unintended coupling among circuit elements to boot. These tend to occur at 
unexpected frequencies (e.g., shorted-out 160 & 80/75M sections of a pi 
coil may exhibit a bodaciously hi-Q resonance around 14 or 21 MHz, along 
with rf voltage that can create a half-inch or longer rf arc.) This can be 
absolutely mystifying until you look for the resonance with a simple GDO. 
Yep - bet most of us learn this "the hard way," too.

Suspect the only reason the home-brew amps I built years ago (and maybe 
yours) didn't/don't self-destruct from these sorts of things is that the 
unintended resonances are typically quite high-Q and have far more NON-ham 
spectrum than ham frequencies to inhabit by chance. Result: many of us, by 
sheer luck, never operate our amps close enough to these unintended 
resonances to trigger the big arc or its counterpart, the big heat created 
by excessive circulating current. It's sort of like the fact that we never 
know how close we came to being hit by a log falling off a truck if it 
doesn't actually quite break loose.

73,    Dick   W0ID
  


--
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