Amps
[Top] [All Lists]

Re: [Amps] GS35B Amp 15 m Situation

To: <amps@contesting.com>, "Mike" <k4gmh@arrl.net>
Subject: Re: [Amps] GS35B Amp 15 m Situation
From: "Tom W8JI" <w8ji@w8ji.com>
Date: Mon, 3 Jul 2006 07:19:08 -0400
List-post: <mailto:amps@contesting.com>
> Having an issue with a homemade GS35B operationon 15 and 
> 17 m.  When
> keyed, the power seems to hesitate before going up to 1.5 
> KW
> output.  CW is a mess on these bands due to this 
> situation.  On RTTY
> the power hesitation is so short that it doesn't present a
> problem.  Amp. seems to be okay to about 1200 W, but the 
> hesitation
> returns above this level.  Other bands, 10, 20, 40, 80, 
> 160 work okay.

That's a big tip-off of what is happening.

That effect is almost always caused by the electrical length 
of cables and networks and what they appear like to a sensor 
or detector someplace upstream towards the transmitter. 
Worse case is when a network/ line combination comes out an 
odd 1/4 wl long to a point that is high impedance when idle 
or at the instant of activation.

Consider a 1/4 wl line with an open at the far end. Say it 
is connected to a tube cathode and the tube is biased off. 
At the moment of excitation you'd have very low voltage 
available at the exciter end of that 1/4 wl line because 
that point would look nearly like a short. If you had an 
electronic bias or fault protection circuit looking at that 
point, it would see very low voltage and be slow responding.

It doesn't have to be a transmission line, it can be a 
network with suitable phase delay. A minimum Q network can 
look like a phase delay of around 90 degrees, so it 
approximates the characteristics of a 90-degree transmission 
line with a surge Z0 equal to the surge impedance of the L/C 
combination at the operating frequency. Open the far end by 
having a tube in cutoff or an open relay, and an electrical 
1/4 wl away you have a dead short (except for some small 
resistance caused by component losses).

Almost anytime you have a band-sensitive start delay, it is 
rooted in this problem.

If you are using active bias you need two or more sensors 
that are analog "or'ed" at different points in the system, 
and you should never set quiescent bias into low level 
non-linear regions. That,. along with a response time faster 
than envelope rise, will cure any problems.

73 Tom 


_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps

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