Amps
[Top] [All Lists]

[AMPS] Amplifier Experiments!

To: <amps@contesting.com>
Subject: [AMPS] Amplifier Experiments!
From: w8ji.tom@MCIONE.com (Tom Rauch)
Date: Tue, 26 May 1998 07:46:01 +0000
To: <amps@contesting.com>
> Date:          Mon, 25 May 1998 22:40:31 -0500
> From:          Jon Ogden <jono@webspun.com>

Hi Jon,

> >You changed tubes, and the thing popped. IMO, you probably have a 
> >gassy tube.
> 
> Why would a gassy tube cause a resistor combination that should be able 
> to carry several amperes of current to pop like a firecracker?

Because the result of even the slightest amount of gas is the tube 
flashes over and conducts very heavily.

The tube no longer has to depend on cathode emission, which is 
limited by the space charge around the cathode.

It's even tough to test for gas, because the gas can be gettered not 
only by the intentional gettering agents put in the tube, but also by 
the arc itself. 

Let me give an real world example.....

When a certain tube designed to operate at about 2600 volts or 
so arrives at a testing point in a manufacturing process, the tube is 
high potted to 8 kV. That is the normal mistuned peak anode voltage 
expected.

The tubes are sorted by those that fail hi-pot, and those that 
don't (about 40%). The tubes that fail go to a RF generator that 
operates them at gradually increasing voltage, getting  the anodes 
red. After that they go back to highpot.

Any that still fail highpot are labelled bad.

About 60% total pass highpot after the aging, and go back to testing 
where they are operated in a PA with HV fault detection. The detector 
watches the grids for voltage spikes that amount to a 2 ampere or 
more current spike at the grid.

Any tube that faults more than three times in 15 minutes is kicked 
out by a three strike counter circuit. Nearly half fail this test. 

The total yield is about 40%,  with only a few causing future 
problems. Using the hipot alone as a go-- no-go test almost half of 
the good highpot tubes will fail in service, and almost half of 
useful tubes would be rejected. 

> there is no blue glow of any kind whatsoever in the tube.  So what caused 
> things to go pop?  It certainly was related to putting the cover in 
> place.  It's definitely something odd.

If the cover hit the tube anode, or was too close to it, it could 
cause the POP.

If the choke or tank was NEAR series resonance, and the cover's stray 
capacitance to the component caused the component to reach resonace, 
it could cause the POP.

It could also have been just a coincidence the tube arced 
after putting the top on.

Whatever the problem, you had an arc because filament emission would 
not and could not cause the problem 

> And I wouldn't expect a gassy tube to have more gain than a non-gassy 
> tube.

Why not? Gain has nothing to do with gas. Gain is determined by the 
ratio of driving impedance to output impedance of a GG amplifier. 
Those impedance, as well as the mu, have little or nothing to do 
with the potential for a high vacuum arc.  

> The DC blocking caps are connected to the "nub" on the end of the cap 
> with a homemade metal clip that the original builder used.  The clip is 
> made out of 1/8" copper stock.  From the clip, another short piece (about 
> 1/2" to 1") runs to the blocking caps.  I have three paralled caps held 
> in place by two brass plates about 2" square.  From the blocking caps to 
> the anode in a straight line it's about 7".  It's a little longer than 
> that when adding in the supressors, etc.

That's a pretty long lead, no wonder the anode resonance is so low.

The lower the anode resonance, the harder it is to de-Q the anode 
without adding loss at the operating frequency.
   
> The vacuum cap is connected to the B&W 850A coil by about a 1/2" piece of 
> braid.

Braid has several times the RF resistance of a smooth conductor. 
That's a high current area because of the VAR power in the tank. 
Braid should never be used at that point.

> The distance from the end of the vacuum variable to the support/grounding 
> bracket is about 5.5 inches.  The distance from the bottom of the chassis 
> to the center of the vacuum cap is about 5.5".  The bracket looks like 
> it's 1/16 or 1/8" metal stock.  It's pretty stiff.  I just noticed it's 
> connected to the chassis by two brass 90 degree angle irons.  So it's 
> grounding might not be the very best.  I just noticed this.

Sound like a difficult tank to de-Q at VHF. Tuning caps should 
be solidly mounted on sheet metal connected directly and solidly to 
the chassis with as close to zero length as possible. The shorter 
that path, the less suppressor ESR required.
   
> OK.  I see what you are saying now.  Interesting.  A good practical 
> technique I haven't seen in any books.  I think it makes sense.  If I 
> think of the RF as a wave propagating along the cap to the output, the 
> cap would have a better effect than to just connect the RF output to the 
> inductor.  Perhaps that's why the cap has been kinda odd.

That's right. Build the tank considering VHF signals , and you'll 
have a very good HF tank.
 
> It's interesting.  I just went and looked at the old Johnson Viking 
> Invader that I've been using for parts (don't worry it was in BAD shape 
> when I got it).  Anyhow, the output on that amp is not connected as you 
> suggest.  It's similar to what I have now.  So do even all of the 
> commercial amps follow your idea?

It's not "my idea". It is just how things happen to work.

I've even seen TVI filters built the "wrong way", and they show peaks 
in response measurable up at UHF. 

Older amplifiers, manufactured prior to FCC requirements for type 
acceptance, have poor attention to details like this. Even a few new 
ones and TVI filters do!

73, Tom W8JI
w8ji.tom@MCIONE.com

--
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

<Prev in Thread] Current Thread [Next in Thread>