Updated 13th December 2010.
Also Local Oscillator version .HEX file has been added to the page.
Reverse DDS PCB TOP
Reverse DDS PCB Bottom
Low cost crystal oscillators are simply not stable enough for the higher microwave bands .
Reverse DDS is a system to lock crystal controlled oscillators to a high quality reference...and it can be retro fitted
to your existing oscillators if they have or can have varicap fine tuning.
The Reverse DDS system described here is based on work done by Andy G4JNT with further assistance from Chris G8BKE and using a PCB developed by John G8ACE.
Using RDDS any crystal oscillator such as our local oscillators for the microwave bands, that are typically in the range 90 to 140MHz, can be locked to any
high quality reference in the range 5 to 20MHz. A GPS disciplined reference is ideal but many good reference oscillators are available as surplus or even "lurk"
in our junk boxes. And they are likely to be much better than the custom crystals we can buy at reasonable prices today.
Odd frequencies such as 135.666667MHz (3400MHz LO for 144MHz IF) and even odd references e.g. 8.192MHz are no problem as the DDS tuning word
can be calculated and the DDS device programmed to suit. With the RDDS system described here this can be done either
using a simple RS232 connection to change one or more of the 16 provided local oscillator frequencies, or if you want to highly customise
your system, the source code for the controlling PIC is available. The PIC which controls the RDDS can be programmed separately or in circuit,
as the RDDS module has a connector compatible with the PicKit series of programmers from Microchip. An expansion header provides access to
additional PIC I/O and the supplied Local Oscillator code can accommodate “band switching” so that only one RDDS
can meet all your Local Oscillator locking needs for up to 16 different frequencies. You can even do away with crystals altogether and use a
free running oscillator with a suitably high Q resonator such as a coax line, although you will have
rather higher phase noise than with a crystal oscillator if you opt for this route. Perfectly adequate for a shack signal source though.
In addition to locking Local Oscillators the RDDS is designed to lock microwave beacon OCXOs, which are often on very "odd" frequencies
(e.g. for a 10GHz beacon on 10368.955MHz using an OCXO at 108.009948 multiplied by 96). The PIC can provide FSK and,
given an accurate time reference such as GPS and suitable PIC code, even JT4G identification for serious weak signal microwave beacons.
What is Reverse DDS ?
Normal Direct Digital Synthesisers use a reference oscillator usually multiplied up to, for example, 1000MHz,
to clock the DDS device which then generates the required frequency. For microwave applications, this is then
multiplied up to the required frequency for a Local oscillator or signal source.
One problem with DDS is that the multiplication process increases phase noise and unwanted spurs.
In reverse DDS the wanted frequency is derived from a crystal oscillator in the usual way,
and multiplied up as required. A sample of the crystal frequency is used to clock the DDS device (AD9851 in the RDDS Module) which is programmed
to generate the reference frequency e.g. 10 MHz. So it is Direct Digital Synthesis but in reverse! This synthesised reference frequency is then phase
compared with the high quality reference and the difference signal is used to control the OCXO using varicap(s) that would normally be used for fine tuning.
The resulting signal, when multiplied to microwaves, has much lower phase noise than a signal generated by conventional Direct Digital Synthesis.
A big advantage of Reverse DDS is that it is a simple retro-fit to your existing OCXOs if they have, or can have, vricap fine tuning.
RDDS at 108MHz multiplied to 10.368GHz
Note that the spurs visible on the spectrum with 50MHz span were at ~ +/- 900KHz.
They could not be detected with the station 10GHz system although the wanted signal
was 5 9 + 40dB or more.
With the station 10GHz system tuned 5-10KHz away from the wanted signal so that it was outside the pass band,
no trace of the beacon signal or increase in noise level was noticed.
Comparison between RDDS using AD9851 and DDS using AD9852
RDDS multiplied to 12GHz
DDS multiplied to 12GHz
Note the high level of Phase noise on the signal generated by normal DDS