Special DDS for the 500kc project.
And other low frequency experimenters
I made some modifications to my DDS to accommodate the precision that
was needed for this experiment.
- Frequency Range: 100KHz to 1200KHz
(DDS works down to DC, but specs not guaranteed below 100KHz)
- Step Size: 1.0Hz
- Harmonics -60dbc of better
- Mirror & Spurs -70dbc or better
- Phase noise -100dbc or better
- Output level 250mV peak to Peak
My design uses a standard 1/2 can crystal 50Mhz oscillator module.
It works up to 20MHz, with 10Hz steps. But for the 500Kc project,
they needed a 1Hz step, but outputs only in the 500KHz range. In
order to get a smaller step size I needed to reduce the frequency from
the normal 50MHz. I ended up using 4.914.304 Hz because this
provided for steps of exactly 1Hz. But in order to get precise
steps the oscillator need to be exactly on. Commercial
oscillators are accurate but cannot be calibrated. So I designed
Crystals at 4,194.304 are readily available (It's 2^24 Hz) as parallel
mode. They come in normal HC49/U and also the short can
version. I tested both and the short can version had much more
temperature drift. So I ended up with the tall can version
specified with a 12pF load, parallel resonant. Mouser part #
520-HCA419-12X They have pretty good temperature stability in the
above circuit. I calibrated them at 75F. At 105F (a nice
hot day outside in Vegas) they drifted up 2.6ppm. Putting them in
the freezer, set at 0F, they drifted down 14.5ppm from the 75F
reading. If one was to assume the drift is linear, and I'm not
sure it is, then one could conclude that if you keep your radio shack
+/- 10 degrees F, then it will drift no more than 2ppm, or 1Hz at 500Kc.
I will supply the electronic parts with the kit. You will need to
build it on a Vector board or something similar, which I don't supply
with the kit. I built one with about 4 inches of cable, so that
someone could put it in a separate temperature controlled mini oven if
they wanted. But I imagine you would put it back to back with the
PCB and keep the runs short. Only 3 connections are needed
+5v power, ground, and the output signal. If it's more than an
inch or so use coax for the signal line. I also let the braid of
the coax act as the power ground, so I used just one wire in addition to
Here's the recommended layout on vector board.
The transistors I am supplying are 2N4124, only because they are a
pretty decent transistor and I got a boatload very cheap. I
imagine any NPN transistor would work.
You connect the master oscillator to the open space normally used for a
can oscillator on the DDS PCB. There are through hole's and SMT
pads in parallel. I put the Signal and the +5v through the holes,
but soldered the ground to the SMT pad.
It is very important to calibrate the master oscillator. You will
achieve much higher accuracy by calibrating the master oscillator at
it's frequency rather than the final output of the DDS. A 10Hz
error in the master oscillator will result in an error of about 1 Hz at
I don't have any laboratory calibrated equipment, I decided just to
use my regular radio (Icom 746Pro) , with
some help from WWV and a PSK31 program. Zero beating a signal using
a radio with normal filters for SSB is difficult. So I tuned to WWV
at 5Mhz by setting the
radio to 4999.000 USB. This produces a carrier at exactly 1000Hz. I
ran my PSK31 program and watched the waterfall. There was a sharp line
just above 1KHz.(a little off, but thats OK) I put a marker at the
exact position of WWV. I then returned to listen to the master
oscillator, and set my radio to 4193.304 USB,
exactly 1KHz below the desired frequency of 4194.304. I then
adjusted the trimmer so that the 1KHz line on the waterfall display
exactly matched up with the poition of the one from WWV.
Note that the 4194 oscillator is rich
in harmonics, I could easily hear the 10th harmonic at 41MHz. And
calibrating against the harmonics may give you even more accuracy.
However the 500KHz output is very clean, I could not even hear the
2nd or 3rd harmonic. But if you lay a pick up wire near the "D" test
point on the DDS PCB you will pick up more harmonics
(this is an unfiltered digital output).
I will be shipping these kits with a detent rotary encoder, which is
most appreciate for the precision needed for this project.
Tuning is very slow, it takes 2-3 detents to move one Hz.
There are two extra holes which are
normally for the speedup button. If you get tired turing the
dial, you can put a pushbutton in there and that will give you some
faster tuning modes. But it's actually easier to set the
frequencies via the RS232 port. If you are using the RS232 port,
you will notice that 3 decimal points are shown for everything.
This is different from the normal unit which has 2 decimal
points. If you use the RS232 interface to fiddle with some of the
parameters ignore the decimal points for every thing except
frequencies. For example the RS232 manual says that the parameter for a
baud rate of 4800 baud is 1.74 In your unit it's 0.174, but it's
actually 174 for both, so just ignore the decimal points.
I will be including the LCD with the kit, but not the cable.
It's simplist to just solder 8 wires from the DDS PCB to the LCD.
If you want a connector you can use 10 pin breakaaway headers, and 10
pin flat cable with IDC connectors on each end. Similarly for the
RS232 cable. I have some that work, with D connectors on each end,
you cut off one of the D connectors and solder 3 wire to the 3 holes in
the PCB. If you want to use a 3 pin header, and a connector that
you might see in some of the photos you can do that too.
I should also say that I have not tested the alternate (44870) LCD
interface. If you read the specs for my standard DDS, it supports
both the supplied Serial LCD supplied and any 44870 LCD.
For this special, I had to chenge the way the decimil points were displayed,
and it's different for the other LCD's
I didn't try to get that working. Stick
with the serial LCD I include with the kit and you'll be fine.
The maximum output frequency for this verion of the DDS is about
1200Kc. Although it will let you tune above that, the output
level will drop off quickly because of the 1MHz LPF.
You get the memories and IF offsets just like the my standard DDS, but I
don't think those will be of much use to you.
$49.99 $37.50, plus $2.99 S.H to the US, includes DDS PCB, programmed micro, 1MHz LPF filer components in lieu of normal filter, and 4.194 osc parts in lieu of 50MHz Osc can. A detent rotary encoder and LCD module are included. The DDS chip is required but not included (AD5932) but it is genrally available as a free sample from the manufacturer. Also not includeed are solder, wire, cables, or vector board for the special 4.194 oscillator module.