On Thu, 06 May 2004 17:19:30 GMT, "Frank Dresser"
wrote:


"matt weber" wrote in message
.. .
On Wed, 05 May 2004 13:40:55 GMT, "Frank Dresser"
wrote:


"matt weber" wrote in message
.. .

The send puts the receiver in standby, filaments remain on but that is
about all. I suspect it disconnects the B+, or at least mutes the
audio.

The switch is between the center tap of the high voltage secondary and
ground. Opening the switch disables the B+ circuit. Closing the switch
with the tubes warmed up forces a large surge current through the
rectifier
as it charges the filter capacitors.
Not really. The vacuum tube rectifiers have very high internal
resistance, it is why you can safely use cap input filtering.

Yes there really is a large surge current when hot switching vacuum tube
rectifiers! In fact, they went to the trouble to develop a spec for the
maximum hot-switching transient plate current. The RCA tube manual says
it's 2.5 amps for the 80/5Y3. That's with the tube manual's recommended
input filter cap value of 20uFd. This radio uses an input cap something
like 40 uFd.

I'd rather play it safe, and not hot switch the rectifier.

The tube
itself is a very effective surge limiter that gets better with age as
the thorium on the cathodes is evaporated off.

The hot switching problem gets worse as the tube ages. As the emissions go
down, the voltage drop goes up. Hot switching old, high voltage drop tubes
can cause internal arcing.

Hot switching vacuum tube rectifiers needlessly reduces their useful life.
Not appreciably. Rectifier failures are almost invariable the result
of filaments burning out. Remember that both the anode and cathode
have considerable mass, so they tolerate short term gross overloads
very very well. In fact that is the basis of the so called vacuum tube
sound. Most tubes that are rated for a few watts, can easily put out
tens of watts for a few seconds without damage, and a Tube like a
4PR400 which is rated 400 watts can actually take a several hundred
thousand Kilowatts for a few milliseconds at a time. If you tried
that with a solid state device, it would be toast . Short of getting
them hot enough for the seals to fail, or elements or connections to
the elements to vaporize, there really isn't any such thing as short
term damage from overload unless the tube is already at the end of
service life anyway. (you eventually boil off all of the thorium on
the cathode).

By the way, there's no thorium in the 80/5Y3. These tubes use oxide coated
filiments. I'm not aware of any thoriated tungsten filament tubes used in
normal consumer electronics since the days of the '01A.


If you did it with a solid state rectifier, the rectifiers burn up
unless you protect them. Some of the high voltage/high vacuum
rectifiers like the GZ34 had trouble delivering 250ma with 800 volts
on the plate.

That's what happens when tubes are operated out of their ratings. The RCA
tube manual says the maximum plate voltate allowed for the GZ34/5AR4 is 475V
rms, with a capacitor input filter. That's a peak voltage of 672V. The
recommended rms voltage with a choke input filter is 600V, but the output
voltage from the choke input filter will be lower than the rms voltage
input.
I suggest you review the design of the EICO 720 transmitter, which
used a GZ34, and plate voltage on the 6146 was a nice round 600 volts.
Never heard of the GZ34 failing, The 5AR5 is no an absolute
replacement for the GZ34. The GZ34 was a lot tougher.

I would add that it was permissible to operate tubes well above the
commercial ratings, in fact many had two rating CCS (Continuous
Commercial Service), and a much higher, manufacturer sanctions, ICAS
(Intermittent Commercial and Amateur Service). The ICAS ratings were
often 25-30% higher.

The recommended maximum hot-switching transient plate current for the
GZ34/5AR4 is 3.7 amps.

I'm sure some people exceed these ratings, but at the cost of life of the
tube.


That's one of the reasons really big, vintage power
supplies user Mecury Vapor rectifiers. They have much lower
resistance, and you haven't seen a rectifier until you have seen a big
3 phase 800 amp mercury pool rectifier.

I'm sure such rectifiers also had a much higher hot-switching spec than
either the 80 or the GZ34. Although it doesn't matter much. Mercury vapor
rectifiers were almost always used with a choke input filter to avoid
letting the thing turn into a relaxation oscillator. Mercury vapor
rectifiers aren't often found in consumer equipment, either.


It's a poor circuit design which was
commonly used back then.
I disagree. Unlike a solid state rectifier, the vaccum tube rectifier
provided surge protection. That is just the way they work.

Exceeding the maximum hot-switching transient plate current will reduce the
life of the rectifier tube.
Why do you believe that? What is the physics involved?
Unlike solid state devices, they have such large thermal mass that you
really needed GROSS long term overload to damage them. Eimac built a
whole line of tubes that were designed for massive (orders of
magnitude) short term overload.



The triode section of a 6K8 will normally draw about 3 or 4 ma. All of that
current flows through the oscillator coil. The mixer section draws 2 or 3
mils through the plate and maybe 6 mils through the screen. So, in a normal
Hartley hookup, at least 3ma flows through the top section of the oscillator
coil, and and at least 11 ma flows through the bottom section of the coil.
I know that's still not much DC current, but is much higher than your
estimate. But wait! There's more!! The oscillator has a parallel resonant
circuit, and as all good EEs know, a parallel resonant circuit has a very
large circulating current, which is limited by the Q of the circuit. Higher
Q means more current.
Perhaps, but the more resistance there is in the coil, the lower the Q
will be. Q is the ratio of reactance to resistance. So if you have
enough resistance to make the I^2 losses reach milliwatts, the Q will
be lousy (and usually is).

even if you work from 10ma, 10 ohms will give you a whole milliwatt to
dissipate.

The tubes in fact cool by both convenction and radiation, both are in
large part take up by the chassis, which is metal. Metal is a pretty
good conductor of heat, so to suggest that heat generated on the top
of the chassis doesn't also end up underneath is nonsense, but I will
concede that resistor to set up screen and bias voltages as well as
the bleeder across the B+ cap are likely to generate substantial
amounts of heat as well, far more than the I^2R losses in any coil
except perhaps in a filter choke.


The oscillator's feedback ratio will also have a
large effect on the current of the coil. More feedback means more coil
current. All that extra current warms the oscillator coil a bit. Not
finger burnin' hot, just a bit. I don't know how many orders of magnitude
the AC current is higher than the DC current, nor do I know how many degrees
the temperature of the oscillator coil changes as it warms up. I don't
care. The effect is that the frequency drifts. The radio has better
frequency stability if the B+ isn't interrupted.

I have
never seen the coils in a receiver get even slightly warm from I^2 R
heating.

It doesn't take a huge temperature rise of the oscillator coil to cause a
few hundred parts per million drift at 10MHz. I'll get about 100 Hz just
from the air conditioner cycling on and off in the same room.
How much of that is temperature, and how much is a result of voltage
reduction as a result of the AC unit cycling on?
Very
annoyiing on SSB. But these aren't very good radios for that sort of work.

No, it's not the voltage shift, either. My SX-62 drifts with temperature,
and that one has a voltage regulated oscillator.


They are heated far more by radiated and convection energy
from the filaments, rectifier, and Audio output tube heat dissipation.
In most receivers, the filament power dwarfs everything else.


In most receivers, the tubes are above the chassis and the coils are below
the chassis.

If you're saying that there's more to the observed frequency drift as the B+
is switched on and off, you have a point. But most radios, including the
S-20R, have the coils under the chassis and the tubes above the chassis.
However, there's some under chassis power resistors which will also
contribute to oscillator coil heating and frequency drift.


If you
have a reciver that is rated 40 watts, and has an audio output of 1-2
watts, the power isn't in the B+. In an All America 5 design, 90+% of
the power dissipated is in the filaments.


Well, let's run the numbers. The typical AA5 uses 150ma tubes in a series
string rated at 120V. That's 18Watts. If the input power is 40 watts, the
total percentage consumed by the filament string is 45%, not 90+%. But AA5s
use a higher percentage of their power in the heaters because the audio
output tube has a high power heater to optimize it for lower plate voltge
use.
If what you say is true, then were pray tell does the other 22 watts
go. It certainly cannot be output. A 50C5 cannot delivery anything
like that, and frankly, if you tried to dissipate 22 watts in the
other 4 tubes, they'd burn out in a matter of hours. YOu are also
assuming an AA5 uses 40 watts, most are more like 25.


For example, the 50L6 uses a 7.5 watt heater, while the 6F6 uses a
4.4
watt heater. Hallicrafters substituted a 6K6 for the 6F6 in their later mid
level radios. The 6K6 had an even more economical heater, at 2.5 watts.

Anyway, these tubes are above the chassis, the coils are below.


Radiated heat goes up at
T^4, so a reduction in power input of 10% results in a change in
temperature that is tiny (on the order of 1.7%)....

Radiated from the above chassis tubes to the below chassis oscillator coil?
Ignoring the actual heating effects in the oscillator coil itself, I have to
figure the under chassis B+ dropping resistors have a far larger effect on
the temperature of the oscillator coil than the above chassis tubes do.
Anyway, the frequency drift after switching the B+ starts right away, and
that implies the source of the drift is right in the oscillator circuit.


That is often measured in tens of wattts. What is dissipated in the
coils is microwatts to milliwatts. Ambient temperature inside the
cabinet had far more to do with coil temperatures then the current in
the coil.

But the under chassis temperature will rise relatively slowly after the B+
is switched. The frequency drift after switching starts immediately.


The tube and coils cool a bit in the
send position, and rewarm up in the receive position. The frequency
shifts
as the temperature shifts.
Not it if was well designed. Designer did two things. They used
regulated voltage on the oscillator,

Oh. How many S-20R radios have been designed with regulated voltage for the
oscillator?

and NPO caps, negative
temperature coefficient, so the temperature of the coils would drive
the inductance one way, the caps went the other way, cancelling the
changes out.

Temperature compensation might work over a small range of frequencies. As
the tuning capacitor is closed, the reletive effect of the temperature
compensation will be reduced. These radios use a bimetal sort of temp
compensating capacitor a couple of inches away from the oscillator coil.
It's nowhere near the above chassis tuning cap, which is subject to the heat
from the tubes. The temperature compensation doesn't work very well. The
compensation capacitor is also microphonic.

As long as we're on radio design, a good frequency stability technique would
be use of low expansion coefficent coil forms such as some of the ceramics.
The forms on this radio are bakelite. Bakelite isn't as good as low
expansion ceramic, but it better than cardboard.

This radio wasn't designed for a high level of frequency stability. It was
designed to be a good value for the money. From that point it was a

"matt weber" wrote in message
...
On Thu, 06 May 2004 17:19:30 GMT, "Frank Dresser"
wrote:

Hot switching vacuum tube rectifiers needlessly reduces their useful
life.
Not appreciably. Rectifier failures are almost invariable the result
of filaments burning out.

That is simply not true. Rectifiers generally fail from poor emissions.
The oxide coating on the cathodes gets used up. Rectifier filaments
sometimes burn out, but that risks a catastrophic failure in which the
broken filament touches the plate and shorts. The rectifier short will ruin
the power transformer quickly. There's still alot of old radios around,
although most of the original rectifier tubes have been replaced.

The rectifiers in AA5s are an exception. They have a crimped area in one of
the internal wires which acts as a fuse. Problems in the radio, such as a
heater-cathode short in one of the tubes will blow the rectifier's internal
fuse, rather than risk burning the radio.


Remember that both the anode and cathode
have considerable mass, so they tolerate short term gross overloads
very very well.


By "mass", I'll guess you mean thermal mass. OK, short term overloads don't
necessarily overheat the tube's internal parts. So what? It's the
cathode's oxide coating that gets used up.

Drawing excessive current through the rectifier is like chirping the tires
on a car. Does chirping the tires cause extra wear? Yes. Will it ruin the
tires right away? No. Will it shorten the life of the tires? Of course.


In fact that is the basis of the so called vacuum tube
sound.

Whaaa.... No, please, please -- don't explain.


Most tubes that are rated for a few watts, can easily put out
tens of watts for a few seconds without damage, and a Tube like a
4PR400 which is rated 400 watts can actually take a several hundred
thousand

Yes, that's nice. What piece of consumer gear uses a 4PR400? The damage
from exceeding the hot switching transient current spec on rectifiers isn't
immediate, it's cumulative. Is this a difficult concept?


Kilowatts for a few milliseconds at a time. If you tried
that with a solid state device, it would be toast . Short of getting
them hot enough for the seals to fail, or elements or connections to
the elements to vaporize, there really isn't any such thing as short
term damage from overload unless the tube is already at the end of
service life anyway. (you eventually boil off all of the thorium on
the cathode).


Ahh, I see. Here's a tube tech update. The thoriated cathode became
obselete in consumer radios around 1930. The oxide coated cathode is much
more efficent. I suppose I should have mentioned the oxide coated cathode
earlier.



By the way, there's no thorium in the 80/5Y3. These tubes use oxide
coated
filiments. I'm not aware of any thoriated tungsten filament tubes used
in
normal consumer electronics since the days of the '01A.


Oh. Looks like I did mention the oxide coated filament. Although I did
misspell filament.




If you did it with a solid state rectifier, the rectifiers burn up
unless you protect them. Some of the high voltage/high vacuum
rectifiers like the GZ34 had trouble delivering 250ma with 800 volts
on the plate.

That's what happens when tubes are operated out of their ratings. The
RCA
tube manual says the maximum plate voltate allowed for the GZ34/5AR4 is
475V
rms, with a capacitor input filter. That's a peak voltage of 672V. The
recommended rms voltage with a choke input filter is 600V, but the output
voltage from the choke input filter will be lower than the rms voltage
input.
I suggest you review the design of the EICO 720 transmitter, which
used a GZ34, and plate voltage on the 6146 was a nice round 600 volts.
Never heard of the GZ34 failing, The 5AR5 is no an absolute
replacement for the GZ34. The GZ34 was a lot tougher.


Maybe it's just me, but I do think there's a difference between 600V and
800V.

Here's a spec sheet on the GZ34:

http://www.mif.pg.gda.pl/homepages/f...010/g/GZ34.pdf

Philips only allows 450V rms input at 250 ma on the GZ 34, while RCA allows
475V input at 250 ma on the 5AR4.

Philips does allow higher voltages, but at lower currents.

I don't see much discrepency between the specs of the GZ34 and the 5AR4. In
fact, the choke input rating charts are identical, if you take into account
that the RCA book uses ratings for one plate:

http://hereford.ampr.org/cgi-bin/tube?tube=5ar4


I would add that it was permissible to operate tubes well above the
commercial ratings, in fact many had two rating CCS (Continuous
Commercial Service), and a much higher, manufacturer sanctions, ICAS
(Intermittent Commercial and Amateur Service). The ICAS ratings were
often 25-30% higher.

Sure, run 'em harder, but at the expense of shorter life. It's in the
manuals!


Exceeding the maximum hot-switching transient plate current will reduce
the
life of the rectifier tube.

Why do you believe that?

Because the the tube manufacturers developed a spec for maximum hot switch
transient plate current. Because I know tube emissions go down as the tube
ages, and it makes sense that stressing the tube will make emissions go down
faster. Because I've seen a marginal rectifier arc internally as it was hot
switched.


What is the physics involved?

No doubt the physics of accelerated tube aging from hot switching into a
capacitor are the same as the slower aging from reduced emissions from the
oxide filament. By the way, the oxides normally used are of barium and
strontium. They haven't been using thorium in production consumer tubes
since the days of the Model A.


Unlike solid state devices, they have such large thermal mass that you
really needed GROSS long term overload to damage them. Eimac built a
whole line of tubes that were designed for massive (orders of
magnitude) short term overload.

Wow. Did Hallicrafters use those tubes as rectifiers in S-20Rs?


Perhaps, but the more resistance there is in the coil, the lower the Q
will be. Q is the ratio of reactance to resistance. So if you have
enough resistance to make the I^2 losses reach milliwatts, the Q will
be lousy (and usually is).

even if you work from 10ma, 10 ohms will give you a whole milliwatt to
dissipate.


It isn't a DC current/resistance problem. Due to feedback, the oscillator's
tuned circuit draws power from the oscillator tube.

I'll try to illustrate it with an Armstrong feedback oscillator, although
the priciple is the same for all oscillators. Imagine a normal Armstrong
oscillator with a tuned circuit at the grid fed through a blocking cap. The
grid leak is connected to ground. The feedback coil from the plate is wound
on the same form, but there's no DC connection to the tuned circuit. But,
when the oscillator starts up, there's no DC on the tuned circuit but
considerable AC. 5 - 10 Volts peak AC is common. In this oscillator, all
the power in the tuned circuit is induced from the plate circuit.

Let's say there's 5V peak AC at the top of the tuned circuit. 5V peak AC
across 10 ohms is 1.25 watts, right? No! This isn't a regular resistance
problem, it's a tuned circuit. The impedance of the oscillator's parallel
tuned circuit is much higher than 10 ohms. But we do get a significant AC
voltage and a significant AC current in the tuned circuit, without a bit of
DC. The upper limit for the power dissapation in the tuned circuit would be
somewhat less than the DC input to the oscillator tube. If it's 3.5 mils at
100V, the upper limit would be .35W. I'm sure the actual dissapated power
is much less than that, but it's still enough AC power to warm the coil
slightly and make the oscillator's frequency drift upward until the coil
temperature stabilizes.

The Radio Amateur's handbook recommends minimal feedback in Variable
Frequency Oscillators to minimize RF heating of the oscillator coils. They
also recommend against using bakelite forms and ferrite slugs, because they
change so much with temperature.

The S-20R uses both bakelite coil forms and ferrite slugs, but it wasn't
designed with super stability in mind. It was made to be a popular radio,
and it was.





The tubes in fact cool by both convenction and radiation, both are in
large part take up by the chassis, which is metal. Metal is a pretty
good conductor of heat, so to suggest that heat generated on the top
of the chassis doesn't also end up underneath is nonsense, but I will
concede that resistor to set up screen and bias voltages as well as
the bleeder across the B+ cap are likely to generate substantial
amounts of heat as well, far more than the I^2R losses in any coil
except perhaps in a filter choke.

Of course, a small part of the tube heat gets under the chassis. But this
heat would take time to reach the under chassis oscillator coil. The
frequency drift caused by switching the B+ happens right away. I'm sure the
under chassis power resistors will eventually cause some oscillator coil
temperature rise and frequency drift, as well.

The frequency drift from switching the B+ is a fact. But if you want to
believe under chassis resistors or tubes cause the drift, that' OK with me.




Well, let's run the numbers. The typical AA5 uses 150ma tubes in a
series
string rated at 120V. That's 18Watts. If the input power is 40 watts,
the
total percentage consumed by the filament string is 45%, not 90+%. But
AA5s
use a higher percentage of their power in the heaters because the audio
output tube has a high power heater to optimize it for lower plate voltge
use.
If what you say is true,

The tube manuals say it's true. If you want to nit-pick, the series string
of an AA5 adds up to 121Volts, not 120. All the tubes have 150ma heaters.
18 watts of heater dissapation is close enough.

then were pray tell does the other 22 watts
go.

The additional dissapation is more like 12 watts, the 22 watts comes from
your 40 watt AA5 number. Anyway, the 12 watts heat the tubes. Do I have to
mention that tubes are also heated by plate dissapation and grid
dissapation? Let's look at the 50L6. At 120 V, it's plate draws about 50
ma and it's screen draws about 4ma. That's a dissapation of about 6.5
watts, just for the audio output tube. Since the tube's plate dissapation
is rated at about 10 watts, it's well within it's tube manual ratings, and
should last a long time. The same excercise can be repeated for each tube,
but I hope you get the point.


It certainly cannot be output. A 50C5 cannot delivery anything
like that, and frankly, if you tried to dissipate 22 watts in the
other 4 tubes, they'd burn out in a matter of hours.

All the tubes in AA5s are operated within their normal dissapation ratings.
The tube manuals say so!



YOu are also
assuming an AA5 uses 40 watts, most are more like 25.

It's more like 30 watts, but 40 watts was YOUR number.

Frank Dresser