[Lowfer] Re: Amplifiers for 136k

[James Moritz]

Dear Lowfers,

I would certainly agree that with a 100W PEP limit, a linear amp would be 
the way to go - overall efficiency is likely to be about 50% or so , so you 
would end up dissipating less than 100W. The efficiency reduction is not a 
big deal if you are talking about a limit on TX output rather than on the 
DC input as in the case of the current lowfer rules, and a linear is 
certainly more flexible for the soundcard-generated modes, which seem to be 
becoming default. Bear in mind modes like Jason and the slow BPSK modes are 
more or less 100% key-down for extended periods, so better cooling is 
required than in typical 100W HF rigs.

Having said that, the sound card is really a nuisance for some modes - 
Wolf, for example, where frequency accuracy is important - and you have to 
provide circuitry to translate from audio to 136k. It is quite possible to 
apply amplitude modulation to a class D amplifier for "linear" modes, as I 
have done with my "Decca" TX. A high-level modulator for 1-2kW output is a 
big and complicated thing, but one for 100W would not be a major 
undertaking, it would only require a single pass transistor. In my case, 
overall efficiency is about 80% when transmitting envelope-shaped BPSK, and 
this could be improved, if you wish to minimise global warming.

For linear amplifiers, the MOSFETs designed for audio PAs are probably 
best, even though they are more expensive than the switching types - for a 
given current/voltage rating, they can dissipate more power, and have 
better bias stability under linear conditions. The much-vaunted negative 
tempco of MOSFETs only holds over a certain part of the operating range, 
depending on the design of the device.

Transmission lines are not a problem at LF - a 100m (ie. 330 feet) reel of 
ordinary RG58 coax had a loss of only 0.6dB at 136k - the bigger stuff will 
have even less loss. I use several pieces of RG58 coax salvaged from old 
LAN cabling without trouble at the 1.2kW level (but beware the BNC 
connectors fitted, which are often poor quality, and fitted incorrectly as 
well). There is a certain elegance to designing a PA which uses a loop 
antenna as it's tank circuit, but it really isn't a good idea for 
relatively high power out. Apart from having to re-design the PA every time 
you alter the antenna, any reasonably sized loop will have multiple 
resonances at HF and above, so will tend to radiate square-wave harmonics 
rather than filter them! Impedance matching using transformers or reactive 
networks is quite straightforward at LF, so you can match and filter at 
whatever impedance levels you like within reason. Any sensible design of 
tank circuit, whether for class A,B,C,D or whatever amplifiers will have a 
low loaded Q (say 5 or so), so will not need adjusting over a bandwidth of 
less than 2%, which is what we have on 136k.

The antenna will have much higher Q unless it is very large - my 50m of 
wire with it's loading coil has a Q around 100, and I find it necessary to 
re-tune if I change frequency more than 200Hz or so. Tuning at the antenna 
is OK if you are operating only on a fixed frequency for beacon type 
operations, but if you need to QSY, you will soon get fed up of running 
about in the dark re-tuning things -  this will be difficult to avoid in a 
narrow band with higher power stations in it, where everyone having their 
"own" frequency is not really possible. I use a geared motor to drive a 
variometer, with a simple forward/reverse switch to do the tuning remotely, 
which works perfectly well. The resistance of the series-resonated wire is 
transformed from about 30ohms to 50 ohms to match the coax feed, using a 
ferrite-cored transformer with multiple taps. The class D PA has 
fixed-tuned tank circuits as above, and is also matched to 50 ohms with a 
transformer. There is a 2 x pi-section low pass filter on the TX output, 
which gives enough attenuation at HF so that the harmonics are only about 
S7 on an HF dipole a few metres away from the LF antenna. It is only 
neccessary to change the antenna transformer taps when changes are made to 
the antenna - the rest of the time, the remote tuning switch is the only 
tuning adjustment required to the transmitter when changing frequency, 
which only takes a few seconds.

It's true that a 100W LF station with a modest antenna that is 100miles or 
more from other LF stations is unlikely to cause problems with key clicks, 
whatever is done with the modulation. But definitely more care needs to be 
taken with modulation/key clicks in more ambitious set-ups - with a really 
big antenna, 1W ERP could be achieved, putting signal levels up about 30dB 
on current Lowfer levels, in 1/15th of the bandwidth. I guess that makes it 
15000 times more probable that such a station would be interfering with 
someone!

Cheers, Jim Moritz
73 de M0BMU