2008-06-29
More Beacon Work
Beacon Transmitter
I put the 80 metre CW beacon into an Altoids tin. This worked quite well shielding the unit far better than expected despite no effort having been made to decouple the DC input feed. With a dummy load the signal is hardly detectable across the room. The boards are held in place with double-sided foam tape, short wires connect to the tin for grounding where required, power comes in via an RCA socket.
Antenna Work
Previous quick experiments with a ferrite base-loading inductor were reconsidered when I felt how hot the inductor was getting. Almost all the transmitter power was being dissipated in the core! Calculations and common sense about high-Q inductors suggested I needed copper wire gauges and coil former volumes I just didn't have in stock, so an order was placed for materials. In the meantime I wound a 159 uH inductor using 0.5 mm wire close-wound on a piece of 25 mm OD conduit. Neatly hand-winding 150 turns without a winding jig is a process that requires much patience, my hands where cramping at the end of the job. The self-resonance frequency of the inductor is about 10 MHz (confirmed with my super-regen dipper), the aspect ratio was chosen to minimise self-capacity of the coil while still having a reasonable Q, while "squarer" coils often have better Q for the same wire gauge they have more self-capacity and hence a lower SRF and peak-Q frequency. This inductor has a Q-peak estimated near 2 MHz, so it is being operated non-optimally. Not being space-wound and in full contact with the former is the largest Q-limiting factor.
While not ideal, this inductor dramatically improved the antenna performance, allowing Peter VK2TPM to hear my feeble signal at his QTH and even read parts of it. Here is a short recording he made after the inductor change. Prior to this he could just tell there was a signal there, but not read it.
The new inductor is somewhat larger in value than needed, my C-jig is used to reduce the effective antenna capacitance and tune to resonance with the fixed inductor. Resonance is quite sharp (a good sign - means the Q is fairly reasonable), and the RF voltage delivered into the high antenna impedance (kilo-ohms in magnitude) is quite high, enough to ionise Neon NE-2 bulbs held near the coil.
This voltage is not healthy for the C-jig polyvaricon BTW, it resulted in a flash-over of the small trimmers on the back of the polyvaricon causing a small region of carbonisation in the insulator edge. This gave an intermittent fault that only appeared as resonance was approached and the voltages rose high enough to flash to the burnt spot - the trimmer insulation was carefully scraped cleaned and the trimmer position displaced slightly from minimum reducing the 'sharp edge' field concentration curing the problem, but the final matching network will use an air-spaced trimmer of higher voltage rating. At higher powers I have no doubt you could destroy a polyvaricon quite easily.
Essentially the antenna appears as a small capacitance (Cant - 10-30 pF) in series with a resistance which is composed on a small value which models actual signal radiation (Rrad) and a larger value that models the losses in the ground system (Rgnd). The coil has its losses modelled as a resistance as well (Rcoil), but basically the entire system resistance appears lumped together once Cant is tuned out by Lload (and Ctune, which acts to reduce Lload and/or Cant). This total antenna resistance must then be matched back to our transmitter output load resistance, 50 Ohms in this case. The inductive part of the L-matching network could be absorbed into Lload but I decided not to do this enabling each section to be tested separately.
Matching is important because the filters and output devices in the transmitter were designed to operate with a 50 Ohm resistive load, reflecting smaller resistances (or more pathological reactive impedances) back into the transmitter can make it pull too much current destroying its output devices or cause it to generate spurious signals. Larger resistances are fairly safe with this particular design, it just tends to poorly load the unit and result in very little output power. Especially bad is inductive reactance which can result in resonances in the output stage and frequency multiplication preferentially producing higher harmonics resulting in overheating - this is unfortunately easy to achieve with the L-match feeding scheme and happens close to the optimal resistive match because of the sharpness of the tuning.
Two different feeding schemes have been trialled; initially my trusty end-fed half-wave antenna matching unit was used instead of an L-network, having just sufficient range to match the moderate resistance (hundreds of ohms) of the resonated antenna. More recent work involved careful measurement of the antenna resistance at resonance (near 300 Ohms) and design of an L-network to match this to 50 Ohms (5 uH inductor and 330 pF capacitor in a low-pass configuration). A pair of 10 uH Ohmite chokes in parallel were used and a silver teflon capacitor from the junkbox, network Q is around 2.2 so the values and component performances aren't especially critical. At resonance the Γ seen looking into the matching network is < 0.15 which is a VSWR better than 1.35:1.
The 300 Ohm resonant load resistance figure suggests very high ground losses (Rgnd), I am using the cast iron sewer back-vent of my building as a ground system, it may not be well bonded at each joint. The radiation resistance is estimated to be in the 1-5 Ohm region for this length of wire, so the total system efficiency is 1.7% at best and perhaps as little as 1/3 of a percent! Radiated power is therefore in the 450 uW - 2.5 mW region ignoring mismatch and any other losses I can't quantify.
Enough to say it isn't getting out very well at present, better antennas are in the works though.
14 comments.
Attachments
| title | type | size |
|---|---|---|
| Antenna Network Circuit Source | application/postscript | 12.276 kbytes |
Parent article: 80 Metre CW Beacon.