Direct Frequency Synthesiser Dave Powis, ND8P / G4HUP Sat Aug 11, 2007 MVUS Microwave Day
Introduction • Direct Frequency Synthesis is a method of locking a high frequency to a precise reference, such as a GPS derived 10MHz source • This presentation covers the basic principles of DFS, and describes the design of of a flexible dual loop DFS
Frequency Locking • There are a number of schemes available to amateurs for locking of oscillators to precision sources: • Injection locking • PLL’s • CT1DMK Reflock • DFS ......etc...... • All have relative advantages – and disadvantages
Direct Frequency Synthesis • In a DFS, the 10MHz input is multiplied and divided to create signals that can be mixed together to give the final output frequency • In theory, the quality of the output signal should be as good as a crystal oscillator • But good filtering is required at each stage of the process to minimise the products of the multiplication and mixing processes.
A simple DFS • For that 2304 transverter with a 2m IF – well 90MHz is a good starting frequency (x24) +10dBm
90MHz
0dBm
10MHz Input Amp
Diode Mult
LC Filt
Ampl
Simple, huh?
Xtal Filt
Previous designs • Brian Justin, WA1ZMS – for his extreme microwave work • Sam Jewell, G4DDK – 96MHz (23cm with 2m IF) – Scatterpoint • Dave Robinson, WW2R/G4FRE – 90.667MHz for 2320Mhz with 2m IF • All these designs need further processing to get the output frequency
96MHz DFS – G4DDK 2MHz Div by 5
6MHz Tripler
LC Filt
96MHz Diode Mult 10MHz
90MHz
LC Filt
Mixer
LC Filt
84 / 96MHz
Ampl
Xtal Filt
Scope of the single loop DFS • This method of combining enables several useful LO frequencies to be synthesised: • 96MHz - 23cm • 90.667MHz – 13cm (2320MHz) • 110.667MHz – 1.2mm (24048MHz) • Also 92MHz and 95 MHz • But that’s about it for 144MHz IF’s!
The Dual Loop DFS • By adding a second divider and mixer loop many more useful frequencies can be produced – but the penalty is more ‘grass’ • WC8VOA uses a DEMI transverter for its 10368 EME station. • With a doppler shift of ±20kHz max, you need to know where you are! • Even the MicroLO does not give adequate accuracy or stability for this application
The Dual Loop DFS • 183.3333MHz is used as the xtal freq by DEMI – so a DFS at 94.6667MHz could be doubled to provide this input • To generate the 94.6667MHz requires 90MHz (easy!), 4MHz (also easy) and an additional 667kHz • 10MHz dvided by 15 gives 667kHz – so that’s easy too!
Dual Loop Schema 667kHz Div by 3
2MHz Div by 5
LP Filt Mixer
4MHz Doubler
3.333 / 4.667MHz LC Filt
LC Filt 4.667MHz
Diode Mult 10MHz
90MHz
LC Filt
Mixer
94.667MHz O/p Filt
85.333 / 94.667MHz
Flexibility of the Dual Loop DFS • In addition to the versatility of having 2 loops, there are 3 places where the second loop can pick off its starting signal Div by n
a 10MHz
c
b Div by m
Mult y
Mult x
LC Filt
LP Filt
Flexibility of the Dual Loop DFS • Further flexibility comes from: • the choice of multiplication factors in each chain • the selection of upper or lower side frequency at each mixing process • Using different input frequency – 5 or 15MHz for example are common GPSDO outputs
Another example • 10368MHz transverter with 144MHz IF, requiring 106.5MHz LO • Two possible schemas from 10MHz: • 90MHz + 3 x 5MHz + 3 x 0.5MHz or • 110MHz – (3 x 1MHz + 0.5MHz) • Better still in this case is a 15MHz input - 105MHz + (15/10)
Other possible combinations • 94.75MHz – 902MHz • 101.75MHz – 3400MHz
• 95MHz - 2424MHz (JA 13cm), • 92MHz – 3456MHZ, and • 104MHz – 5760MHz can all be achieved with single loop DFS systems
Performance Achieved • So far a prototype breadboard has been built – not a pretty sight! • Output spurious are approx -45dBc • Currently translation to a PCB design is in hand – 148 x 74 x 30mm case. • It is expected that this will result in a 10 to 15dB improvement in spurious products visible in the output spectrum
Performance Achieved Prototype filter testing
Prototype output spectrum, 70 to 115MHz
Conclusion • I have shown that a complex frequency can be derived from a precision source • Further work is needed to complete a usable solution for the VOA station • I hope to report on this work, and have the DFS unit available at MUD this year.
transverter has a frequency offset and has temperature variations. ⢠GPS locking to Z3801A will eliminate these problems. ⢠A DFS at 94.667MHz can be doubled ...
Dec 5, 2007 - The main circuit blocks are shown in Fig 1, and are described in the following sections, which refer to the circuit diagrams at the end of this ...
comb of frequencies at 10MHz spacing and the following filter selects the wanted harmonic at. 90MHz. A two-stage amplifier increases the level to +7dBm (easily ...
PLL. UMA1022M. Low voltage. Low noise. Date: December 20th, 1996 ..... and ground pins are provided to the analog (charge pump, bipolar part) and digital.
3.1 Basic Loop Filter Design Procedure. ..... After the supply voltage is switched on, the different registers are loaded with following preset values.
A close in noise value of -86.5 dBc/Hz has been measured in the loop bandwidth ..... is settled (i.e. acceptable frequency or phase error with respect to target). .... In this section, a design example based on the third order PLL for GSM is shown.
and local oscillators used in the detection of moving tar- gets in such ... -30°C to +70°C. However, for higher operating temperatures, please contact the factory.
frequency, a correction factor (in +dB) is shown on the label ... correction factor is added to the power meter reading to yield the value of true ... 1% duty cycle.
performance. ENR outputs of up to 35 dB allow detection of receiver decenoration and developing faults. In addition to the standard product line of system Noise ...
be done only with the help of a magnifying glass of at least 4 dioptres. .... HF-PCB ist üblicherweise mit 2-Komponenten-Silberleitkleber eingeklebt und wird bei ...
Chip '5 \ ~-._ to \ ~. - PHOVEN RELIABILITY h ,2 .0 at âs. a. . 2 \Lu. 'ii'; DESCRIPTION 3 9 new ~., .'I. I The NEBOUt AND NEeooz are medium power GaAs FETs.
This makes copying of weaker SSB signals quite difficult. The blame for this effect can be laid directly at the door of the crystal itself. I have been pursuing a cure ...
The downconverters have conversion losses ranging from. 6 dB in K-Band to 8 dB in ... A calibration curve is supplied that allows accurate absolute power ...