> This can be detrimental to the health of the 3-500Z if the exciter is
> capable of 150-200 watts, and no cathode padding or no external ALC
> cut-back is used.
> Again we agree. I have experienced the transients. That is why I believe
> the cathode padding resistors in Rich Measures kit are a valuable addition
> to this, and any other amp that can be over driven by modern day transceivers.
> It certainly does not hurt anything, and almost assures a clean signal in the
> "heat of battle."
Adding cathode degeneration is a requirement, if the
exciter is capable of the power levels you mention.
ALC will not cure the problem, because ALC is traditionally too slow
to limit leading edges of waveforms.
Also be sure the radio actually keys the PA before RF is applied, and
removes RF before the keying signal is removed. I've seen radios
output RF long after the PA is unkeyed!
> >On 160 meters the voltage
steadily decreases from tap to
> >tap as you move along the switch wafer. There is no "tesla effect" on
> >160, because the inductor does not have self resonant sections at the
> >operating frequency.
>
> That is correct. The failures to my switch were on the 15 meter and
> 40 meter tap positions, and on the 10 meter position. The 10m and 40m
> contacts were nearest the ceramic support insulators, and it melted
> the ceramic!
The problem is voltage to ground in you failure, and occurred at
the weakest point in the switch. The wafer indexing was rotated
to position the ten meter tap away from the mounting screws, but
some contact needs to be near the grounded mounting screws.
> The only failures I have had on 160 meters were the door
> knob padding capacitors. That was before I obtained a book from HEC
> and found out how low the current limits really were on such devices.
> ETO, in their infinite wisdom, failed to supply higher current padding
> caps with their 77DX-to-SX conversion kit. The drift in the original
> cap due to heating was horrific! No wonder..its current rating is only
> 1.6 amps @ 1 mhz!
Be sure you apply that current rating correctly. On lower
frequencies current is determined by the I*X voltage drop
across the component and the voltage drop across the component. On
160, you are well within the rating of the component. The 1.6 amperes
does NOT indicate the capacitor is close to HEAT failure from
excessive current, al low frequencies the current rating indicates
the point of arcing from VOLTAGE failure.
Drift problems are a different issue. Drift problems have come and
gone with ceramic doorknobs. HEC's capacitors were originally quite
stable, while ITT Jennings caps were very poor. When HEC bought out
ITT, HEC's stability went to hell.
> >You can measure this by driving the tank with RF from a generator,
> >and probing it with a hi-Z RF voltmeter. You can either drive the
> >output port with a 50 ohm generator, terminating the anode in the
> >operating R, or you can drive the anode through a series R equal to
> >the anode source impedance from a low impedance source and
> >measure voltage with the tank terminated voltage.
>
> This simulation may be of value during initial design, but when you
> put the power to the circuit, strange and different things begin to
> happen. In a Pi-L circuit the weakest and most vulnerable point is
> the band switch, In a Pi network, it is the tune cap
I've yet to see anything "strange" happen, only things that initially
are not intitially understood. The key is to understand what is
really going on.
So if I'm missing something here, I certainly want to understand what
it is!!!
A network behaves the same way at one watt as at a
million watts, at least until something fails. Networks also behave
the same way driven backwards as driven the normal direction. That's
why networks can be analyzed as two port back box devices.
The voltage at the anode end of the network is determined by the
anode voltage, conduction angle of the tube, and energy storage in
the tank. As the analysis point moves along the network towards
the low impedance load, voltage steadily decreases. Someplace before
the end of the inductor is reached, voltage to ground reaches a
minimum (that's because a pi is really two back to back L networks,
with a low Z point out a bit from one end).
A pi-L network is a simple pi- network driving an L network. The
pi-network is designed for some median impedance value at the
loading capacitor, like 200 ohms. The L section transforms the 200
ohms to 50 ohms through a minimum Q L section.
Other than the slightly higher voltage across the loading
capacitor, the pi-L is no more likely to produce high voltages
anywhere in the tank system than the pi.
> The L section is NOT the problem, I agree. It is a player in the result!
> No strait Pi network design I have ever seen bothered to short any turns;
> commercial or otherwise. It made no difference back then because only
> 5 bands (80-10m) were involved. (B&W 850 series coils, et al)
Virtually all PA's short unused turns. Turns often need to be
shorted to prevent unwanted voltages from building up when operating
on higher bands.
On a 80 through ten meter tank, the designer might get by without
shorting unused turns, but he likely won't get away with it on a
160 through ten PA.
> In the pi-section of a Pi-L network, the voltages on the Pi coil from
> end-to-end are actually higher than in an equivalent strait Pi network.
Not necessarily. It depends on phase shift in the network. In most
design cases, phase shift is less making end to end voltage less.
Traditionally pi-L's operate at slightly less Q in the pi-section
that conventional pi's.
In any event, the problem was indicated to be voltage breakdown to
ground..... not end to end across the switch.
If the switch had rotor to ground or rotor to ten meter tap arcing, I
would worry about the fact it is a pi-L. If arcing was from a tap
terminal to ground, the problem was excessive voltage to ground.
To me, this failure mode indicates an unloaded tank (for
whatever reason). The cause could be exciter transients, excessive no
load supply voltage, load arcing, keying timing, improper tuning, and
so on.
I'm in no position to isolate a cause, since multiple causes
can be at play. But I can say one thing for sure, the reason was
almost certainly not due to the network being a pi-L...unless the
pi-L caused a loading capacitor or rotor to ground arc.
By the way, the pi-L was selected to make the loading
control have increased range on 80 and 160. Removing the L section
decreases tuning range.
No matter what the cause of this particular problem, I'd hate to
see people dismiss pi-L's as "arc prone" systems.
73 Tom
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|