Zozo wrote: ↑Wed Mar 03, 2021 10:47 am
Another point which makes me favor VXO's is there's less components needed, No synthesizer IC or supporting controller. You can even put the code writing to one side. One of the only down sides to the VXO was ordering Crystals, But now you can order one very cheaply and they get cheaper as you increase you order quantity.(Established pirates long term investment maybe).
There are three gigantic problems with a VXO - the need for a specific crystal (quite expensive and "unusual" frequencies engender unwanted interest), and very poor modulation linearity. With a VXO you're "pulling" the crystal off its natural frequency. The deviation from the centre frequency is simply NOT linear. Back when a Big Broadcasting Concern used VXO VHF rigs, they had to use "pre-distortion" networks to correct for the non-linearity, which didn't work particularly well. The third problem (related to the second) is that the available deviation is limited.
VXOs are fine for voice communication, but no good for high quality FM audio, in my experience.
The deviation problems are why PLLs are generally used for VHF high quality audio transmitters. You can use cheap, standard crystals - I pay around £0.26 for the ones I use - and the ICs and diodes for a basic PLL cost around £1.40 if you shop carefully. If you make the PLL in SMD, it can be tiny, and if you put the crystal in an oven, you can get spectacular stability. If you're being really flash, you can receive 198kHz off-air, square it up and divide it to somewhere useful as the reference for your PLL. The frequency of your rig is then controlled by the frequency source at the National Physical Laboratory!
The inspectors from OFCOM complained that a MW RSL I did was a few Hz off-channel - I demonstrated that the problem lay with their inaccurate frequency counter, not my NPL-referred rig!
Actual rig stability is often a function of the PLL loop filter. Roger Howe had a problem with one of "his" designs that "hunted" around the required frequency. The problem was actually down to Roger incorrectly noting down the time constant components in my PLL circuit at the time......
Way back, there were three or four predominant exciter designs:
The first came from the 1970s, which was a doubler-doubler circuit, and had a bizarre "totem pole" coil assembly in the middle of a large square PCB. It used five RF transistors, one more (low noise one) for the modulator, and a couple for zener-based regulators. It was a free-running VFO design, and was capable of good stability if carefully constructed. It was used as the exciter for a lot of Radio Invicta QQV06-40 rigs. The circuit originated from "The Tadworth Mouth".
The second one was another doubler-doubler (later simplified to a tripler variant) that was used for the free-running RFL rigs, and widely coped. The circuit originated from "Michael Martin", and used a Colpitts oscillator with silvered-mica capacitors for stability. The doubler stages used under-biased BSX20 transistors with high-Q collector circuits. The doublers had to be screened from each other to prevent spurs. The original circuit used oscillator, buffer, doubler, doubler, filter / amplifier, amplifier to about 100mW, all using BSX20s (6 stages), followed by a 2N4427 for about a Watt. If built carefully, with the right components, it was reasonably clean and quite stable.
BDHK simplified the Michael Martin circuit so that the oscillator ran at one third of the output frequency, then used an un-tuned buffer stage, a tripler, and an amplifier to 100mW - reducing the circuit to 4 BSX20s and a 2N4427.
The third circuit came from the Netherlands, and used a push-pull oscillator (similar to the NRG oscillator, but with only two coils), and a push-pull doubler / amplifier stage to about ½W. My version of it then had a pair of SD1127s (or MRF237s) for about 7 Watts. That was enough to drive a 2N6084 to over 40 Watts if critically tuned! This was the first circuit that I added a (TTL) PLL to for best stability.
The last circuit was a kludge based on a circuit from Radcom magazine, and used two transistors in a sort of "cascode" affair, a doubler with two tuned circuits, an amplifier stage with a toroidal collector transformer to drive the base of a 2N4427. This formed the basis of the BW exciters. It wasn't a bad circuit, but the BW insistence on using a single varicap, and tuning the whole band led to insanely high mod sensitivity, and a propensity for nasty hum, due to RF feedback. Roger did a lot of work on getting the output clean, and for that reason abandoned the doubler, moving to an at-frequency oscillator. This worsened the RF feedback problem, of course. The BW boxes used every trick in the book to try to minimise the RF feedback problems - often rather unsuccessfully.
There were a couple of other designs - the "Thameside" exciter used an RF op-amp as its buffer stage, and I did one that used the SL560 RF amplifier chip as oscillator and buffer in one, and a second SL560 as doubler and filter amp. My 560 exciter also used the SP8629 as the prescaler from Band II (or Band III for link gear), and then used cheap CMOS for the rest of the PLL at low frequencies (around 1 MHz). My exciter looked a bit weird, with three 8-pin DIL ICs and a 4427 at the end. The CMOS PLL bits were beneath the board, with the chip pins bent outwards to make them into "big surface-mounted" parts. I bought a huge boxful of SL560s from a component broker for about £0.02 each and had to use them up!
Variants of the RFL and BW circuits are being made to this day. The latest iteration of the RFL circuit that I've seen uses a half-frequency oscillator and an un-tuned diode doubler (so that it's broadband) into a two-stage amplifier for around a Watt.
The final exciter design I've seen recently uses a 2N4427 as an at-frequency "power oscillator", driving a RD15 FET for over 12 Watts in just two stages!
smiley for once!