Amps
[Top] [All Lists]

[AMPS] Blown TL922A... What to do?

To: <amps@contesting.com>
Subject: [AMPS] Blown TL922A... What to do?
From: W8JI@contesting.com (Tom Rauch)
Date: Thu, 7 Sep 2000 09:11:39 -0400
> >The tank is typically a virtual short for VHF and UHF energy,
> >because it has a capacitor shunting the input.
> 
> That's not quite right, Don. The tune C has inherent inductance, so you

The series inductance LOWERS the impedance. It forms a virtual 
short in the cases I have measured.

> get resonances. The resonance frequencies vary with shape, size and
> capacitance setting, but in all that I have put on the network analyser,
> the lowest frequency resonance is 'series', or low impedance and the next
> resonance is 'parallel' or high impedance. I'm using the term resonance to
> mean points where the impedance goes purely resistive.
> 
> The one I looked at yesterday was 100pF max, .06" spacing and about 2.3"
> cube overall. The first, series, resonance came in at 90-150MHz, depending
> on C setting, and the higher one was in the 200-300MHz region. The higher
> frequency one was low Q.

I measured a transmitting capacitor commonly used in PAs. It was 
series resonant, and a virtual short, at the frequency where Rich 
claims considerable VHF voltages occur across the terminals of 
the capacitor. We were talking about a specific tank system, and a 
virtual short of a few ohms.

I didn't find a high impedance (above the operating frequency) within 
the range of my equipment, which is 1.2 GHz maximum.   

The tank circuit with the transmitting type capacitor typically used 
that I tested was a virtual short for VHF and lower UHF energy. 
 
> The impedance at the capacitor terminal is low either side of the series
> resonant point. How low depends on how far you move in frequency, but it
> doesn't suddenly go high impedance at any frequency 'close' to the
> resonant point.

Exactly. The series resonance produces a wide smooth dip in 
impedance. Rich's guess that it suddenly goes high is a bad 
guess, even though it fits his "theory". 
 
> Adding inductance in series with the capacitor (to simulate the lead from
> an anode) changes the resonant frequencies, but doesn't change the overall
> characteristic.

The lead from the anode affect the impedance at the anode, not 
across the capacitor or into the tank. That assumes the tank is laid 
out with some common sense, and the lead is routed from the tube 
to the tuning cap, and then from the tuning cap to the switch and 
the rest of the tank circuit.

The best lead arrangement generally is in one stator terminal of the 
capacitor and OUT the other lead. This lets the capacitor act like a  
low pass feedthrough.
 
> Adding a L/R parallel circuit in series with the capacitor introduces loss
> as the frequency increases, but does not introduce multiple resonances. As
> I understood what he said, Rich suggested that the separate current paths
> through the L and the R should produce multiple resonances. I said then I
> thought his analysis was wrong. I still do.

Rich's analysis is wrong. I can't measure multiple resonances here 
either. There are some small ripples in impedance, but nothing like 
Rich claims.

I think the problem is Rich only has a grid-dip meter. All of his 
resonance theories, for grids...anodes...whatever, are based on a 
dip meter. 

Dip meters do not measure system impedance or the resonance 
mode or Q, they measure signal suck-out through magnetic field 
coupling to an oscillator to whatever is in the field of the GDO tank 
coil.

Why not measure the impedance at the path loss where the switch 
contacts "arc" from VHF energy? I did that in one PA, and it would 
take a tube driving many thousands of amperes at the frequency of 
instability to arc the switch.


73, Tom W8JI
w8ji@contesting.com

--
FAQ on WWW:               http://www.contesting.com/FAQ/amps
Submissions:              amps@contesting.com
Administrative requests:  amps-REQUEST@contesting.com
Problems:                 owner-amps@contesting.com


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