From: w8jitom@postoffice.worldnet.att.net
To: amps@contesing.com
To: <amps@contesting.com>
Date: Thu, 8 May 97 08:34:46 +0000
Subject: Re: [AMPS] tube structure/failure.
Priority: normal
Skip wrote:
> I don't agree.
Hi Skip,
Fair enough. It's great to see someone disagree without personal
attack.
> While trying to source a problem in a single 3-500z amplifier, I observed
> a flash and failure of the tube fillament structure. The amplifier was a
> homebrew based on the Henry amplifier line and the many Handbook designs.
What were the conditions when the tube flashed? Was the PA drawing
grid current withjout drive? Were grid or catode currets excessive?
> This amp was plagued by "diode hash" in the standby mode which turned out
> to be a parasitic just about 50MHz. This was the second tube to "die" in
> the same amplifier within two months.
To oscillate at 50 MHz, the grids must have had some seriously long
leads or a tuned circuit near the area of 50 MHz.
Even if the tube oscillates, any damage would have to be caused by a
parasitic that indicated visable current on the elements.
The grid current would have to be excessive, or the cathode current
VERY excessive. Let me explain why....
The main cause of failure of a good tube element is rooted in
thermal problems (ignoring mechanical shocks, like dropping
the tube). In order to excessively heat an element, excessive power
must be applied for a finite amount of time. In a tube like a 3-500Z,
the saturated cathode current is about 100mA per watt of filament
power, or about 8 amperes.
As a matter of fact pulse duty charts show (operation with elevated
filament voltage) 2000 volts of positive grid voltage causes the peak
cathode current to just reach 10 amperes.
Since that current causes electrons to smash into the grid, kenetic
heating of the grid is the primary problem with heating. Grid
dissipation is almost 10 kW during that condition. If the pulse
duration is excessive, the grid fails much sooner than the cathode.
In any case, the thermal lag of the filament and grid require some
time to be expended at high currents (dissipations). The grid
actually has to become white hot to melt and fail, since it is a
tungsten alloy.
The meter's response is much faster than the grid heating, and any
current large enough to cause a grid failure would surely show
as greatly excessive current.
We aren't talking about a nanosecond pulse failure, this isn't an
FET. It's a massive vacuum tube.
At 50 MHz (the frequency you claim you measured the parasitic at) ,
the tube can be operated normally at several hundred watts output or
500-600 mA of total cathode current for DAYS without failure.
Common sense tells me if you had a parasitic strong enogh to wipe
out the filament, it would have loaded the power supply heavily. It
takes average power to cook the elements, not peak power. Time is
absolutley a factor.
> The big fix was to replace the garbage grid current metering circuit with
> a "cathode" return lead type. The grid pins of the 3-500z socket were then
> tied directly to the chassis ground through large low XL straps.
Grids should ALWAYS be as directly grounded as possible in a
grounded grid triode PA. There are many reasons, and stability is one
of them.
One final question Skip, a 4-400 has LESS gain than a 3-500Z in
grounded grid. So you had a LOWER gain tube fail.
Were these tubes all brand new tubes? What was the inrush current?
Did the HV positive supply line have an appropriate surge limiting
resistance designed in, or was it a low resistance path?
How high did the grid and cathode current get when the PA was
oscillating?
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
|