[Lowfer] KD5UWL dds exciter

[Eric Smith]

Hi all,

Finally _finally_ making some progress on my new dds exciter:
http://www.nutstreet.net/radio/projects/dds/  I'll update this page as
I go along.

Some of you may remember that I've been using the ZL1BPU exciter for
the last few years, and it has served well, but as designed it's only
good up to about 400 kHz.  That has always been fine, but since
modifying my Part-5 license to include 600m I've been wanting one that
would operate into that range.  THIS HAS BEEN ONE OF THE SINGLE
GREATEST HOLDUPS to me getting on 600m.

SO I'm using the Analog Devices AD9851 DDS IC.  When I first received
this item I had trouble trying to solder the tiny little pins (SSOP
package).  I finally mailed one to Accutek Microcircuits who soldered
it to a 28-pin DIP adapter (their part # AK28D300-TSOP).  This was $7
including the adapter and the soldering.

I'll post a schematic and a link to my (ever changing) source code
soon.  I'm using GCC and the avr-libc library.

Years ago Alberto (writes Argo -- thanks, Alberto, as always!) turned
me on to Atmel's "ATmega" family of 8-bit RISC microcontrollers.
These are a blast -- for this project I'm using an ATmega644.  Has
more memory and some other features than I need, but I wanted the pin
count.  The prototype you see in the picture is functional in that it
will produce whatever frequency waveform I want from DC up to a little
over 5 MHz (without engaging the DDS core's 6x clock multiplier) at
approx .003 Hz resolution [ (1 * 12.8) / 2^32 where 12.8 is my system
clock speed in MHz, see AD9851 datasheet], but I'm planning to add
bells and whistles ... including an LCD and pushbuttons to allow front
panel frequency and mode changes without having a PC hooked up to it.

So I used the MCU to clock the phase, control, and frequency words
into the dds core and strobe the frequency update pin after all five
words are clocked in.  Then the output comes spilling out of the DDS
IC, just that easy.

The clock source is one of the two 12.8 MHz TCXO units Stuart was
offering years ago, "free to good homes".  These have proven to be
ultra stable.  I use the clock output of the MCU to drive the DDS core
with the same clock.  I also divide the 12.8 MHz by 12800 in the MCU
using a prescaler combined with 8-bit timer hardware on the MCU to
produce a 1 ms clock which is the basis for a sleep-based delay
routine I use in the project for setup and hold timing.

Questions for you guys -- the digital hardware and software of this
thing is dead simple to me, it's the analog side of things that I
always struggle with.  So my questions for you are:

 - given this 1V P-P waveform I'm producing, what is the best way to
drive an amp?  That is, do I need to square this up?  Or will
something like Lyle's complementary pair final do that anyway?
 - any (simple) 600m final suggestions out there?  (the amp will
quickly become my next biggest holdup)

Thanks, guys,
Eric  KD5UWL / WD2XFX