NO Think again, the high order bit of a DDS output is very definitely
not a squared up version of the fundamental !  Used on its own it would
have a spectrum not totally unlike noise.

Before the DDS output anti-allias filter, for a given output frequency
all terms of   Fclock +/- Fout are present, as sampled sinewaves with a
SIN(X) / X amplitude roll-off.   Also present is what effectively is
quantising noise (Nq) .   The level of this  Nq is given approximately
by  6.N dBc, so for a 12 Bit D/A in the DDS (as in the AD9850),   Nq can
only be 72 dB down at most.   In fact the manufacturers clain about
65dBc to play safe.   Just using the MSB gives 1 bit quantisation so
spurious products will be at 0 - 6dBc.    To generate a square wave, you
have to generate a proper sinewaveform, low pass filter and feed back
into a high speed comparator.  The AD9850 has such a comparator built in
and all you have to supply is a Clock, command word and output filter.

I once looked at using a PIC plus external D/A for a low speed DDS, and
for audio freqs it could be done, but again with spurious limitiations.
Assuming 8 bit arithmetic - any more would slow it down considerably and
you do need a sine look up table with this resolution as well - the
output could only manage around -40dB spurious and determining timing
would be a horrendous programming job.  A 32 bit accumulator with the
attendant maths all within a critical timing loop.

For MFSK, why not just use a DDS chip / module and send it commands to
set the frequency.   Some of the the now obsolete DDS chips such as
AD9831 and AD9832 can cost not much more than a PIC .

Andy  G4JNT

-----Original Message-----
From: Alberto di Bene [mailto:[email protected]]
Sent: 2001-03-20 15:11
To: [email protected]
Subject: LF: Re: WOLF - am I missing something ?


>  Then I abandonned software
> timing by using an external interface (a PIC) on the serial port and
> ended up with something very similar to the VE2IQ I/F.
>

A few days ago I was thinking on how to generate a programmable
frequency for an MFSK scheme (I did post a message on this reflector),

and the DDS was one the possibilities. Thinking again at it, after all we just need the high order bit of the phase accumulator, as the TX is driven with a

square wave. This makes the DAC completely unnecessary.
Would it be possible then to implement a 32-bit phase accumulator with
a PIC, to generate a frequency on the 137 kHz band ?
My knowledge of PICs is quite limited, any PIC guru with an answer ?

If positive, then it would be possible to make a really inexpensive outboard

generator for such an MFSK modulation scheme.

73  Alberto   I2PHD

--
The Information contained in this E-Mail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).

prohibited and may be unlawful.