Hello Tobias and everyone else.


Thanks for the detailed info, much appreciated.

I  managed  to change the divider again, but this is probably the last
time  as  a pad is lifting. I also put back a 22k resistor for the 10K
R16.  The  new  divider  one  didn't  have any output at all, save the
tiniest  bit  of  ripple:(     So  I  put  back the 10K and output was
intermittent  and  much  like  before.  But I noticed if i touched the
divider  on its plastic shell the the thing gave a much better output,
good  square waves, but they would either stop themselves, or if power
to  the board was removed then re-applied, it wouldn't start. The input
was   on  frequency  and  strong. But now I could sometimes just hover
a  finger  right  above  the divider chip and it appeared a capacitive
effect would start it.

When it did output clean square waves they too were on frequency.
There was a proper 90 degree phase shift. I then put in a 6.8k for R16
and it's better still. Dare I go lower, or is this telling me
something I am not hearing?I haven't changed the crystal yet. I don't
think the issue is now the oscillator itself, but driving the divider.
I am sure Andy is right that a Schmitt trigger buffer would be best,
but I didn't really want to start redesigning something I know others
have had running 24 hours a day with superb results.

I'll  upload the oscillator wave form at pin 3 of the divider as it is
now.

http://www.gatesgarth.com/newR17-to--pin3.jpg


Friday, January 26, 2018, 9:33:04 PM, you wrote:

> Hi Chris !

> The first screenshot shows an un-symmetrical square-wave that will
> generate two frequencies to the 1:4 divider and Tayloe-mixer. An FFT of
> this signal (R17-to-pin3) should show two peaks and not only one as
> expected. A standard frequency counter will show only the arithmetic
> mean value of both frequencies, as it is just adding zero-crossings
> during gate-time.

> The Softrock series 6.x had a different type of I/Q divider that used
> three flips-flops. They did not have a problem with such an
> un-symmetrical square-wave, as there was an inherent divider by two,
> which forced the oscillator signal to be symmetrical to the mixer.

> You could try to move the bias-level of Q2 transistor to a voltage where
> the output (square-wave) signal of Q2 will be almost symmetrical.
> If the oscillator at Q1 is not running smoothly, this is where to look
> for variations of C's or R's to make it run stable.

> 73 de dg3lv Tobias

--
Best regards,
 Chris                            mailto:[email protected]