Content-Transfer-Encoding: 8bit
Message-ID: <5.1.0.14.0.20030815154505.00b0a898@gemini.herts.ac.uk>
Date: Fri, 15 Aug 2003 16:48:13 +0100
To: rsgb_lf_group@blacksheep.org
From: "James Moritz" <j.r.moritz@herts.ac.uk>
In-reply-to: <000d01c362a3$216eb2c0$5967fea9@p4o2b7>
MIME-Version: 1.0
Subject: Re: LF: Variegated witterings.
Content-Type: text/plain; charset=us-ascii; format=flowed
Sender: <owner-rsgb_lf_group@blacksheep.org>
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org

At 21:29 14/08/2003 +0100, you wrote:
>Now the bitter bit. Things were now even worse, with it proving impossible
>to get anything like the previous output figures or a decent waveform on
>136 or even 73 at any over a couple of watts so I hastily put things back
>the way they had been and opened the Pommeau.
>
>So  todays puzzler is;
>
>How come?  Anyone got any ideas? I've gone into sulk mode.
>
>Ian GI8AYZ/MI0AYZ
Dear Ian, LF group,

The main effect of the mods I did to the Maplin audio amp (very similar to 
the B&K modules) is to increase the slew rate by increasing the bias 
current in the input stage - this increases the output voltage swing at 
high frequencies that the amp can achieve and hence the output power for a 
given load. It is not easy to see how that would reduce the output.

One thing to check is that the DC bias voltages in the circuit should be 
roughly the same after the mods as they were before. If there is a wrong 
component, it is most likely that the voltage at one or more nodes in the 
circuit will be stuck near one supply rail or the other. In what way is the 
output voltage waveform not "decent"? I found that my amp had to be 
overdriven to get reasonable output power and efficiency, so the output 
voltage waveform was not very sinusoidal, although with a high Q resonant 
load as most LF antennas are, the current waveform will always be a nice 
sine wave even with really horrible driving waveforms.

One possibility that springs to mind is that speeding the amplifier up has 
resulted in instability problems - this type of circuit, with wide 
bandwidth (a few MHz), and an overall negative feedback loop, is sensitive 
to the load impedance over a wide frequency range, not just at the 
operating frequency. In particular, a load which is capacitive at HF will 
often cause the amplifier to be unstable at some frequency due to excess 
phase lag introduced into the feedback loop. A series tuned antenna by 
itself will be inductive at frequencies above resonance, but if there is a 
fairly long transmission line between loading coil and TX, the shunt 
capacitance of the line will cause the overall load impedance to become 
capacitive above a few 100kHz. If matching is done using a tapped loading 
coil with one end grounded, this will also introduce additional resonances 
at higher frequencies. If the amplifier output voltage waveform is OK with 
a resistive dummy load (or no load at all in the case of the audio amps) 
but not when the antenna is connected, this is probably why.

With my audio amp, I found capacitive loads either resulted in severe 
"ringing" at the output, or caused a blurred waveform due to the thing 
taking off at some unrelated frequency. My cure was to add a low-pass 
filter between the matching transformer at the amp output and the coax 
feeder - the usual pi-section designs won't help due to the shunt 
capacitors at their input, so I used a tee configuration instead with a 
series inductor at the filter input. The low-pass filter effectively 
isolates the amplifier from the ipedance variations at the load. After 
this, I had no trouble. I have occasionally also had problems with class D 
amps and peculiar loads - these were also solved using a low-pass filter at 
the TX output.

Cheers, Jim Moritz
73 de M0BMU