[Amps] Instant-On

[David Kirkby]

DEHAYS Dominique wrote:

> And a cross-over pair of 'caps' for 
>neutrilisation. In the end I actually removed these cross-over wires, as 
>I found I could only get more RF feedthough with them when measured on a 
>diode detector with cold tubes, not less. It was stable without them - 
>
>Hi David ,
>you were young , but quite lucky I think. I did some experiments  on 
>neutralizing yesterday on a Motorola quarter KW , which is a PP conduction 
>cooled 4CX250 equivalent tube .
>This amp was modified by myself on 144Mhz , and is in a stage close to be 
>"fired up".
>I measured the reverse gain with the original neutralizing circuit ( abt 
>the same than W1SL design you are talking abt , but with serie inductors 
>in the feedback loop) and found it to be only -26db . The forward gain is 
>expected to be around the same figure , so the safety margin for stability 
>is 0db! The neutralizing circuit was probably OK on the unmodified amp 
>design frequency but not good on 144Mhz.
>I took off all the feedback links , and measured reverse gain for fun : it 
>was then -13db , so with no neutralizing this amp is an expensive 
>oscillator.
>
In the original design (you were right, it was the W1SL), the cross-over 
wires had pieces of metal on the ends. I can't the recall size,  but 
they were about 6mm square. (I'm going back nearly 25 years).

When I first built the amp, I neutrilised it using some sort of power 
meter on the output, moving those caps around to get minimum feedthough. 
I never had an accurate power meter, spectrum analyser or anything else 
like it, so I could not say what the coupling was.

As soon as the screen voltage was applied (I always did that last), the 
HT fuses blew. I did not know what was happening, but speaking to John 
Nelson, G4FRX, he said the amp was probably oscillating. He said the 
size of the caps in the original design were too large for 'modern'  
sockets. I can't recall exactly what sockets I used, but there were the 
correct Eimac ones for VHF use (SK-600 perhaps?).  John suggested I 
removed the tabs on the end of the wires, but kept the wires. So I did 
that, and immediately the RF feedthough was much lower. But it soon 
became apparent that whatever I did, I could always make it lower still 
by removing them completely.

There was one thing that was odd about my amp. The drive power required 
was very low. I think I used around 700mW for 600W output in AB1, 
whereas 2-3W was more typical for that design. So my forward gain must 
have been around 29dB.

Whether there was an element of positive feedback I don't know, but it 
never seemed to break into oscillation, no matter what antenna was on it 
- good or very bad. There was never any evidence of grid current meter 
twitching. It's also probable that for an instant the amp was running 
into an open circuit, as the relays were not properly sequenced. (again, 
forgive the fact I was young at the time).

It must in fact have seen a very wide range of load impedances, as the 
dummy load I used was a reel of RG58 coax which I partially unreeled. 
But even then, too long in testing and the amp would melt the cable a 
few metres from the amp (not on the reeled up bit). Given the point 
where it melted would not have been the same each time, the almost 100% 
reflection could have been just about anything, depending on the ext 
length at the point the cable shorted. (I always just cut of the cable 
up to the point it had melted, re-attached the N-plug and used it again.

>Doing the same shape links like in  the original amp , but without 
>inductors gave a -35db reverse gain , then perfectly OK , but consistency 
>of this figure with different tubes is questionnable , so I added series 
>variable caps in the links and could tune the reverse gain to -40db . I 
>verified on a network analyser this solution was not too " narrow band " , 
>and found it was OK. Playing with loop shape can give -50db , but it is a 
>marginal improvement in my opinion.
>

If you think about it, if you can get the phase right, then minimum 
feedthough is not best, as that given an open loop, which is I know how 
99.9% of amateur amplifiers work. Ideally, if you get some negative 
feedback in, linearity would be expected to improve and drive power 
increase. Of course, getting the phase right would not be easy, although 
I have considered doing this on another amplifier, where a coupler is 
used on the output to couple off some power back to the input. But you 
would really need a VNA to do this properly, and I don't have one, 
although a friend of mine does.

>Probably the reverse gain you get in the w1sl design with noneutralizing 
>caps was a bit higher than the one in Motorola's amp because your tube 
>sockets are better than the simple one used in the quarter KW , but safety 
>margin must be quite low.
>
Perhaps. I will never know. The amp was sold long ago and I did not have 
access to any decent test equipment at the time.

>Sorry for my " far from perfect " english
>
Don't worry about it. Your English is very good. Mine is far from 
perfect, and in emails I don't tend to be too fussy about the grammar.

-- 
Dr. David Kirkby, 
G8WRB

Please check out http://www.g8wrb.org/ 
of if you live in Essex http://www.southminster-branch-line.org.uk/